Celestica Inc. Aktie

Hello, everyone. Thank you for joining us, and welcome to the Celestica Third Quarter 2026 Financial Results and 2026 Investor and Analyst Day Conference Call. [Operator Instructions] I will now hand the conference over to Matthew Pallotta, Head of Investor Relations. Please go ahead.

Good morning, and thank you for joining us on Celestica's Third Quarter 2026 Financial Results and Investor and Analyst Day conference call. On the agenda for today's call, we'll begin with our third quarter financial results, followed by our 2026 Investor and Analyst Day. At the conclusion of the prepared remarks, we will open up the lines for Q&A. On the call today, we have Mionis, Chair of the Board and Chief Executive Officer; Todd Ankenmann, Chief Financial Officer; and joining for our Investor and Analyst Day portion, Mandeep Chawla, Group President of Global Markets; Stephen Dorwart, President of our Connectivity and Cloud Solutions segment; and Todd Cooper, President of our Advanced Technology Solutions segment.

Please note that during the course of this call, we will make forward-looking statements, including statements related to the future performance of Celestica, our business outlook guidance for the fourth quarter of 2026, our 2026 and 2027 annual outlook and anticipated trends in our industry and their anticipated impact on our business. These are based on management's current expectations, forecasts and assumptions as of October 26, 2026.

These forward-looking statements are subject to risks and uncertainties that could cause actual results to differ materially from our expectations and their potential impact on our results cannot be reliably predicted at this time. For identification and discussion of the material assumptions, risks and uncertainties, please refer to our public filings with the SEC and on SEDAR plus as well as the Investor Relations section on our website. We undertake no obligation to update these forward-looking statements unless expressly required to do so by law.

In addition, during this call, we will refer to various non-GAAP financial measures. We've included in our earnings release found in the Investor Relations section of our website, a discussion of those non-GAAP financial measures and a reconciliation to the most comparable GAAP measures. Unless otherwise specified, all references to dollars on this call are to U.S. dollars, all per share information is based on diluted shares outstanding and all references to comparative figures are a year-over-year comparison. With that, I will now turn the call over to Rob.

Thank you, Matt, and good morning, everyone, and thank you for joining us on today's call. We're excited to be able to share with you more Celestica's role in advancing AI infrastructure. Today, we will discuss a number of important developments and the long-term outlook of our business. We are seeing unprecedented momentum, positioning us for a strong end to 2026 and further acceleration into 2027. Before diving into the Investor and Analyst Day portion of our call, Todd will walk us through our third quarter results, provide our fourth quarter guidance and share our updated 2026 outlook. Todd, over to you. .

Thank you, Rob, and good morning. It's a pleasure to join you for my first call today in this capacity, and I look forward to engaging with you all. Now turning to our third quarter 2026 performance. Revenue in the third quarter was $5.83 billion, up 83% and above the high end of our guidance range, as strong demand and solid operational execution led to higher-than-expected revenue in both our CCS and ATS segments. Our non-GAAP operating margin was 8.5%, up 90 basis points driven by higher margins in both our segments. This once again represented the highest quarterly non-GAAP operating margin in the company's history.

Our adjusted earnings per share was $3.30, exceeding the high end of our guidance range and an increase of $1.72 or 109%. Moving on to some additional metrics. Adjusted gross margin was 11.2%, down 50 basis points, largely due to mix in our CCS segment. Our adjusted effective tax rate for the quarter was 18%. Lastly, our strong earnings growth and disciplined working capital management resulted in adjusted ROIC of 64.7%, up approximately 27 percentage points compared to the prior year period.

Before moving on to our segment performance, we'd like to inform everybody that we have renamed our CCS end markets to better reflect the program composition of our portfolio. Effective Q3 and going forward, our communications end market is renamed working, and our enterprise end market is renamed compute and storage. This update is a change in naming convention only, and the composition of the end markets remains unchanged and does not impact previously reported financial results.

Revenue in our CCS segment in the third quarter was $4.9 billion, up 102% driven by very strong growth in both our networking and compute and storage end markets. Our CCS segment accounted for 84% of total company revenue. Our networking end market revenue was up 68% higher than our outlook of approximately 60%, primarily driven by growth in 800G networking, complemented by early volumes on ramping 1.16 programs and solid demand in our 400G programs. Revenue in our compute and storage end market was up 242% well above our outlook of approximately 190% driven by ramping volumes in hyperscale AI compute and complemented by stronger-than-expected demand and storage programs.

In our HPS business, revenue was $2.3 billion in the third quarter, representing growth of 70%, driven by ramping 1.60 and 800 G switch programs across multiple hyperscaler customers. our HPS business accounted for 40% of total company revenue. ATS revenue for the quarter was $957 million, up 23% and above our outlook of a mid-teens percentage increase. The strong growth was driven primarily by an acceleration in capital equipment and supported by higher demand in each of our businesses.

ATS segment accounted for 16% of total company revenue. Moving on to segment margins. CCS segment margin was 8.9% in the third quarter, an improvement of 60 basis points, driven by strong operating leverage. Our ATS segment margin was 6.6%, up 110 basis points driven by improved mix and operating leverage. In the third quarter, 3 customers each accounted for 10% or more of our total revenue, accounting for 35%, 17% and 14%, respectively. Turning to working capital. At the end of the third quarter, our inventory balance was $4.5 billion, a sequential increase of $1.1 billion and a year-over-year increase of $2.5 billion. as we support very strong demand with an accelerating growth outlook in 2027.

Cash cycle days for the quarter were 42%, representing a 23-day improvement over the prior year and a 5-day improvement sequentially. Moving on to cash flows. We generated $128 million of free cash flow in the third quarter and $413 million year-to-date. Our capital expenditures were $274 million, primary 4.7% of revenue. compared to 1.2% of revenue in the prior year quarter. Year-to-date, our capital expenditures are $767 million or 5.3% of revenue, and we continue to anticipate spending approximately $1 billion in 2026.

During the third quarter, we completed our previously announced public offering of approximately 11.1 million shares with gross proceeds of $3.45 billion before underwriting discounts and operating expenses. At the end of the third quarter, our cash balance was approximately $4 billion, while our gross debt was $736 million, resulting in a net cash position of $3.2 billion. We had no outstanding draw on our revolver at the end of the quarter. Our gross debt to non-GAAP trailing 12-month adjusted EBITDA leverage ratio was 0.4 turns, an improvement of 0.1 turn sequentially and 0.4 turns versus prior year.

As of September 30, 2026, we were in compliance with all our financial covenants under our credit agreement. Our cash balance, in combination with our revolver provides us with approximately $5.7 billion of available liquidity, positioning us to support strong anticipated future growth. We did not repurchase any shares for cancellation during the third quarter under our current normal course issuer bid. Now moving on to our guidance for fourth quarter. Fourth quarter revenue is projected to be between $6.275 billion and $6.575 billion, representing growth of 76% at the midpoint.

Adjusted earnings per share is anticipated to be between $3.37 and $3.57, representing an increase of $1.58 at the midpoint or 84% growth. Our adjusted earnings per share assumes approximately 127 million diluted weighted average shares outstanding. Assuming the achievement of the midpoint of our revenue and adjusted EPS guidance ranges, our adjusted operating margin is expected to be 8.7%, which would represent an increase of 100 basis points. We anticipate our adjusted effective tax rate for the fourth quarter to be approximately 20%. Next, we'll review our revenue outlook for each of our end markets.

In our CCS segment, we expect revenue in our networking end market to grow approximately 65%, primarily driven by hyperscaler lead growth in 800G programs and the ramping of our 1.6T programs. In our compute and storage end market, we expect growth of approximately 160%, driven by the continued strength in hyperscale or AI compute as well as from on-demand and storage. In our ATS segment, we anticipate revenue to be up in the low 20s percentage range. driven primarily by strong growth in our capital equipment business as well as new program ramps in our other businesses. Our solid fourth quarter guidance reflects our continued momentum to close out the year. This, combined with our strong year-to-date performance, supports our increased 2026 financial outlook.

Our 2026 revenue outlook increases from $20.5 billion to $21 billion, representing growth of 69%. We are also raising our adjusted EPS outlook to $11.55, up from $11.30, representing growth of 91%, reflected in our outlook is a full year adjusted operating margin of 8.4%, unchanged from our previous outlook. Finally, our free cash flow outlook for 2026 is $460 million. compared to our previous outlook of $600 million. The change in our free cash flow outlook expectation is primarily driven by targeted working capital investments to support the accelerating growth in 2027. And with that, I'll turn the call back over to Rob to begin the Investor and Analyst Day portion of our presentation. Rob, over to you.

Thank you, Todd. At our Investor Day a year ago, we talked about Celestica being future ready. Over the past year, that future has arrived rapidly as our growth trajectory has steepened dramatically. Today, Celestica stands at the forefront of driving technology advancements as we are actively helping build the world's leading platforms for advanced AI infrastructure. For the remainder of our call, members of our executive leadership team and I will discuss our operations, financials, strategy and capabilities that position Celestica as a key contributor to the AI investment super cycle. .

Our strategy, technical capabilities and market tailwinds coalesce around 4 distinct pillars that drive sustainable long-term shareholder value creation. First is differentiation. Celestica access a critical launch partner our next-generation platforms. Hyperscalers partner with us from early stage R&D to co-design, commercialize and scale their most complex AI networking and compute solutions. Second is growth. We hold industry-leading positions in AI data center technologies, benefiting from generational secular growth tailwinds. We strategically pursue technologies where we have a competitive advantage anchored by our leadership position in Ethernet networking and full system design capabilities for next-generation AI compute platforms.

Third is visibility, our multigenerational customer engagements and deep alignment with the hyperscaler and Front care AI lab customers enable us to pursue long-term capacity planning with high confidence. And fourth is our disciplined approach to investments. We employ an investment framework to drive long-term earnings growth, while filtering those decisions to our strategy and key financial hurdles on margins and return on capital. These 4 pillars create a powerful foundation for compounding shareholder value over the long term.

Our business is fundamentally stronger today than at any time in Celestica's history, reflected by our accelerating financial performance driven by a step function change in our scale, profitability and earnings power. Over the 3-year period through 2026, Celestica's revenue is expected to have delivered a 38% CAGR, more than doubling our top line. At the same time, margin improvements are driving non-GAAP adjusted EPS to compound at an even faster 67% CAGR, representing a nearly fivefold increase over the same period.

Our growth has been deliberate, strategic and profitable. When leading technology companies choose Celestica for their most complex solutions, they do so because of our core competencies in leading-edge design, commercialization and scaling. These capabilities directly address the critical challenges facing our customers. We manage the entire development to deployment life cycle for hyperscalers and frontier AI labs, our next-generation AI infrastructure, enabling cost optimization, accelerated time to market and reliability of their global manufacturing solutions through seamless execution.

The integration of full stack design and scale manufacturing of highly complex platforms produces a powerful combination that is greater than the sum of its parts and enables a strong competitive moat that is challenging for our competitors to replicate. We'll take a closer look at each of these foundational capabilities, which have helped position us as a preferred multigenerational partner for leading AI data center innovators. A central element of our differentiation is our design leadership. We believe our global team is amongst the best and most capable in the industry with the depth of proven institutional knowledge developed over successive generations of technologies.

Our design-led approach enables customization for optimization through our Joint Design and Manufacturing or JDM model. Increasingly, this has become a preferred approach to product development for our hyperscaler and Frontier AI lab customers. By leveraging our system-level design capabilities, we collaborate directly with our customers to deliver customized solutions. These solutions enable superior price performance and an overall lower total cost of ownership while also ensuring supply chain flexibility and vendor diversification. The commercialization phase acts as the vital bridge between design and mass production, enabling a seamless transition that is critical to the timing and quality of program delivery, separating design and manufacturing across different vendors can leave blind spots, where each can miss critical considerations that have the potential to lead to ramp delays, pool yields and costly redesigns when moving from prototype to mass production.

Celestica's vertically integrated model eliminates this friction. Through our design for manufacturability approach, our design and manufacturing engineering teams work side-by-side from initial development straight through commercialization. This allows us to identify and resolve critical production hurdles early on. We believe this is 1 of the more underappreciated strength of our model relative to competitors within our industry. It continues to be a crucial differentiator that unlocks both scale and speed to market for our customers.

Finally, operational excellence is deeply ingrained in Celestica's DNA. scaling a highly complex product from proof of concept to mass production is a technical discipline in itself. It presents unique operational and engineering challenges, a critical area where we often see industry competitive struggle. Our customers rely on us to continuously innovate our manufacturing and testing processes to deliver their most complex solutions. Our stewardship culture sets us apart from other operators. We take the initiative to own the complete solution end-to-end and approach problem solving as a true technology partner.

We standardize these best practices through the Celestica operating system, ensuring strict adherence to standards of quality and reliability throughout our global network as programs scale. To support the significant anticipated acceleration in demand driven by recent program awards, we have approved plans to significantly increase our global capacity over the next couple of years. adding to our growing network of over 40 sites in 16 countries. In North America, our U.S. presence is centered in Texas where we are adding capacity across multiple sites to enable critical home country manufacturing for hyperscaler customers. The new capacity supports advanced L11 integration for a scale deployment of liquid-cooled AI racks. And to address power bottlenecks we are adding hundreds of megawatts across our network, including at our U.S. sites.

In Mexico, we continue to support demand for near-shoring by investing in upgrades at our campus in Montedey. In Asia, our data center customers are seeking additional capacity at our flagship Thailand campus to leverage our exceptional local capabilities and talent. We are making significant additions to support major production ramps through 2027 and 2028. We are also making targeted investments in Japan and Malaysia to support our networking programs and offer customers greater geographic diversity across the Asia region.

Our regionalized multi-node network strategy delivers a flexible, resilient supply chain that helps customers mitigate geopolitical and operational risk while scaling their deployments. Within our network, we are making targeted investments in new automation lines to wrap at scale production. Today, this includes automating L6 through L10 processes for manufacturing the latest generation networking and compute trays. We are also deploying new lines to automate L11 rack assembly, We believe this is a capability that is unique among competing OEMs and strengthens our value proposition for hyperscalers, particularly in higher cost labor constrained markets.

By automating highly repetitive tasks, such as cabling and fiber optic insertion, we are solving critical bottlenecks and high-sensitivity clean room environments. These applications also extend into several of our ATS end markets with fully automated lines in operation at facilities in Mexico and Europe. We take a phased approach to these investments, verifying operational viability before scaling, which we believe deliver attractive payback periods, faster ramps and repeatable high yields.

As we navigate this high-growth phase, we will continue finding new opportunities to drive scale without compromising the exacting quality and reliability our customers depend on. Our differentiated model allows us to serve a highly specialized role for our customers, where the EMS model is restricted in its design capabilities, and the OEM model leans toward more rigid architectures, Celestica combines critical elements of both these models through delivering an integrated end-to-end capability stack, beginning with collaborative early-stage codesign through to full-scale manufacturing and deployment we operate as a true technology partner. Our ability to customize leading-edge AI infrastructure is supported by our investments in HPS hardware road maps, a flexible software stack and our commitment to open vendor ecosystems.

Furthermore, our operational rigor and end-to-end ownership of the solution ensures a seamless transition from prototype to manufacturing at scale. Ultimately, we believe this differentiated model continues to position Celestica as a preferred partner, closely aligning our development strategies and growth investments with our customers' multiyear road maps. With that, I will now turn the call over to Mandeep, Group President, Global Markets, to discuss our growth strategy. Mandeep, over to you.

Thank you, Rob, and good morning, everyone. Our demand environment today is defined by an unprecedented AI investment supercycle, bringing fundamental changes to both our business and our industry. One of these is our improved long-term demand visibility compared to historical norms. High demand for AI compute capacity has seen it become heavily constrained by power and hardware availability. Because of this, hyperscaler customers are undertaking longer-term commitments and engaging with providers like Celestica in multiyear planning cycles to secure the necessary components and manufacturing capacity. Our supply chain strategy manages overall technology availability rather than basic procurement, collaborating with ecosystem partners early to qualify components and build design flexibility.

Furthermore, our Design Light HPS programs require customers to pursue early alignment on product development with us and our silicon ecosystem partners up to 24 months in advance of mass production. We expect these dynamics, particularly our visibility enabled by design-led time lines to persist, allowing us to pursue our long-term growth investments with confidence. We view the AI infrastructure build-out as a durable, multiyear investment cycle, and we believe we are positioned well to capture this opportunity.

Third forecasts anticipate cumulative data center CapEx will reach more than $3 trillion by 2030, representing nearly $2 trillion in incremental TAM growth from 2026. Constrained AI compute capacity due to surging demand for AI workloads, serves as a strong signal for continued investment. Forecasts anticipate total annual AI accelerator shipments will almost triple from 2026 to 2030 with every AI accelerator deployed, driving investment in high-performance networking, making these core technologies that we support amongst the fastest-growing segments of the broader data center ecosystem.

Importantly, nearly 70% of this capital investment is expected to be concentrated amongst the largest U.S. hyperscalers and leading frontier AI labs, who comprise the majority of our CCS customer base. These indicators align with the multiyear demand signals we are observing with both our current and prospective customers. As we touched on, our core competencies are uniquely suited to serve the distinct demanding requirements of AI infrastructure for hyperscalers and Frontier AI labs. These customers operate at the very cutting edge of compute and networking and share unique characteristics, which are defined by several core dynamics, which drive our commercial model and technical investments.

The first is accelerated upgrade cycles with new, more capable compute silicon now being introduced every 12 to 24 months, while networking platforms are being refreshed approximately every 2 years. The second is leading-edge performance and technical requirements. which necessitate the most advanced design engineering and technical capabilities to enable the latest systems. The third is customization, which requires us to enable bespoke solutions designed to support multiple variants of silicon, compute architectures and use cases and tailored to their unique infrastructure specifications and AI workloads.

The fourth is scale, requiring optimized, highly reliable manufacturing processes and supply chain support for the largest deployers of infrastructure in the industry. Finally, to address these specific challenges, Celestica collaborates directly with hyperscaler and Frontier AI lab engineering teams via our JDM model to target the most critical design and manufacturing challenges. Why has Celestica been able to expand from switch design into full rack scale and system-level solutions in modern AI clusters, networking is the fundamental nervous system. Amongst core hardware technologies within AI data centers, high-speed networking design represents the most technically demanding engineering discipline.

It requires multidisciplinary competencies in the areas of ultra high-speed signal integrity thermal management, optical connectivity, board design, software and testing. Having mastered those challenges at 800G and 1.6T, we believe we have the foundational technical capabilities to engineer adjacent domains required to design complete AI systems across compute, power delivery systems, liquid cooling, cabling, intra and inter rat connectivity, software orchestration and rack design.

The boundary for design optimization is shifting from individual boards up to full rack and complete AI clusters, where considerations regarding networking, optics, power and cooling are becoming increasingly interconnected. This dynamic creates a powerful competitive barrier. Moving up from server assembly and rack integration into designing high-speed networking present significant engineering challenges. It requires a kind of deep institutional knowledge and proof points that we have spent years building. However, moving from high-speed networking into converged system design is a shift that plays directly to the breadth of Celestica's core engineering capabilities and strengths, unlocking a growing set of opportunities where we see massive potential in the coming years.

The value we deliver to leading AI platforms is directly reflected in our engagements with hyperscaler customers, which continue to accelerate. We have earned this position by executing relentlessly across the critical success factors of speed to market, customization for optimization and scale, allowing us to build strong multigenerational relationships with these customers. From 2023 through 2026, we expect Celestica's hyperscaler revenue to grow by approximately 5x, expanding at a 71% CAGR. We are positioned to remain a critical provider to the industry's leading players as we are expanding our engagement to support additional hyperscaler and Frontier AI lab customers in 2027.

Our market position is reinforced by this diversified customer portfolio, consisting of a number of the leading hyperscaler and frontier AI labs in the industry. This is supported by our deep relationships with an ecosystem of the leading designers of high-performance compute and networking chips. We collaborate closely with silicon providers to align early on product road maps, supporting the next generation of chips in development. This enables us to begin our work on designing market-ready solutions well ahead of wider commercial availability and collaborate on supply chain planning, supporting speed to market and availability for our mutual customers.

Our breadth of critical relationships and collaborations have immense benefits that continue to pay dividends, solidifying our role as a critical technology solutions partner within the global AI data center ecosystem. With that, I'll now hand the call over to Steve to discuss our CCS segment.

Thank you, Mandeep, and good morning, everyone. Our Connectivity & Cloud Solutions segment continues to see unprecedented momentum from the AI infrastructure investment super cycle. We are actively designing, industrializing and manufacturing the world's most advanced AI infrastructure for leading hyperscalers and frontier AI model laps. The trajectory of our financial performance reflects the tremendous scale of revenue growth we're driving with CCS revenue expected to reach $17.4 billion in 2026, which roughly quadrupled our business in just 3 years. Our proprietary design capabilities delivered via our HPS offering continues to be an important driving force in our success, and now accounts for roughly half of our CCS revenues.

As we look ahead to 2027 we believe our expanding program wins and technology leadership position CCS to accelerate our growth trajectory. Our team is incredibly excited and energized by the pipeline of opportunities in front of us. which we believe have the potential to sustain our positive momentum in the years to come. Our HPS offering continues to see incredibly strong engagement from our customer base, helping bring leading-edge Ethernet networking and system-level solutions to market. To maintain our technology leadership, we have grown annual R&D spend by more than threefold since 2023 and expect to invest over $300 million in 2027.

The scale of our investment enables us to support a rapidly growing breadth of technical capabilities, proof of concepts and product road maps. We are seeing strong productivity from these investments, which translate directly into revenue opportunities enabling us to engage earlier in product cycles, support more platforms and convert proof of concepts into multiyear programs.

We continue to support a broad portfolio of unique programs in active production and at various stages of development. And our team of over 1,600 HPS engineers across 9 global design centers today is expected to grow to over 2,000 engineers by next year. While HPS built its reputation and high-performance Ethernet networking, we have successfully expanded the scope of our projects into custom ASIC compute and rack scale systems, which opens up opportunities for greater design-led engagement in the market.

Our R&D strategy consists of a structured 4-step innovation life cycle. We first explore the development of new proof of concepts and capabilities to support emerging technologies alongside our key partners. well before the formal RFPs, building expertise on frontier Silicon ahead of industry adoption curves. Next, we evaluate the commercial viability, including potential scale of market adoption through rigorous screening to identify the candidates meriting further investment. Third, we industrialize the design by pairing early R&D directly with factory floor process engineering eliminating handoff friction and derisking the transition from NPI to mass production.

Finally, we commercialized complete HPS solutions like our DS 6000 Series switches which serve as market-ready proof points of our IP and expertise and can be leveraged as reference designs for customized JDM variants. We developed this disciplined innovation formula to derisk development schedules, support high first pass manufacturing yields and dramatically shorten time to market. Modern AI data centers are undergoing fundamental structural changes that are driving our technical investments today. First, advances in technology and greater infrastructure scale are lowering the technical and economic barriers to hyperscalers designing a proliferation of new high-performance custom ASIC chips.

This leads to the second as each new custom chip requires a bespoke rack scale compute architecture and in many cases, a customized networking fabric to support. As the level of customization increases, foundational aspects of the system must be tailored for the specific silicon and performance requirements, requiring a bottoms-up approach to designing each new architecture.

Third, compressed upgrade cycles necessitate that all of this is done faster and more frequently. With silicon upgrade cycles contracting to 12 to 24 months, hyperscalers and chip designers are often developing multiple generations concurrently at various stages making a well-executed development phase and faster time to market more critical than ever. These developments are resulting in a growing number of unique product road maps and architectures, increasing technical demands on design engineering resources, both our customers and design partners like Celestica.

To keep pace, our strategy focuses on enabling core technologies and manufacturing capabilities that serve as foundational building blocks for leading-edge systems with broad applicability allowing us to support a wider range of customized solutions and customers simultaneously. With rising complexity and demand for customized solutions, we believe the value of Celestica's model continues to increase. Our engineers are working on solving today's foundational technical challenges to enable the next generation of leading-edge AI systems. First, as we touched on already, is the shift towards engineering, full rack scale and system-level solutions. We see it becoming more challenging and, in some cases, impractical, to approach the design of networking and compute components independently from the entire rack or system, leading to a deeper scope of engagement.

We are witnessing this in networking fabrics where we are seeing an increasing use of disaggregated dedicated networking racks as well as compute architectures. Next is increasing integration of optical components into the rack design, the networking trays with approaches such as co-packaged optics and eventually into compute trays. We will touch more on this later. The move towards integrated optics is also related to challenges associated with Signal Integrity. As we move towards 400-gig surveys, this tightens electrical loss budgets and restricts copper distances, requiring simultaneous optimization across the entire physical path, driving the need for advances in PCD laminates, surface loan packaging and direct attached copper.

With our deep institutional knowledge in this area, we continue to invest in specialized engineering talent, simulation and lab capabilities to develop and validate next-generation designs. Finally, the areas of power density and thermal management are tightly coupled as rack designs approach 500-kilowatt or more. Here, Celestica is actively engineering solutions to accommodate the transition to 800-volt DC power distribution architectures. Future systems will leverage the use of advanced liquid cooling for components beyond GPUs and switch silicon to include optics and memory. We are actively evaluating next-gen technologies, including 2-phased liquid cooling and advanced thermal interface materials.

By solving these challenges at the system level, we help customers bring high-performance platforms to market faster and effectively scale increasingly complex AI infrastructure. Now diving into our end markets, beginning with our networking portfolio, anchored by our market-leading Ethernet networking business. In 2026, networking revenue is expected to reach $11.8 billion, up 66%. Our momentum this year is being driven by ramping 800 gig switch deployments along with the beginning of mass production in our earliest 1.6T programs.

Looking into 2027, we expect 800G to continue its very strong growth trajectory and 1.6T volumes to accelerate substantially across a broadening base of customers and programs, including a number of variants supporting scale up. As a result, we anticipate that our 1.6T revenue will surpass 800G revenue next year. To address evolving network architectures, Celestica has continued to develop our flagship family of DS Series switches for the 1.6T generation within our HPS portfolio.

Last year, we introduced the DS-6000 and DS 6001, the air-cooled and liquid-cooled variants of our Broadcom Tomahaxic-based switches. This year, we've added 2 new variants. Earlier this year, we announced the DS 6010, a scale-up switch, purpose-built and co-engineered for AMD's Helios rack-scale architecture, utilizing UA Link over Ethernet, which we will dive into further in a moment. Earlier this month, at the OCP Global Summit, we introduced the DS-6005, our first co-packaged optics design built on Broadcom's Tomahawk 6, Davidson Chip, integrating high-density silicon photonics engines, liquid cooling and field replaceable external laser modules in a 1.5 OU form factor, delivering maximum power efficiency for dense clusters.

All DS 6000 series platforms are supported by Celestica's enterprise-grade SONIC software stack, providing open, flexible network telemetry and fabric orchestration. We are seeing expanding investments across every tier of AI networks. Celestica has Ethernet networking solutions for use in all of the different domains within the data center. However, within our portfolio, we primarily support designs at the leading edge bandwidths deployed in back-end networks for scale-out and a fast-growing portfolio within scale up. supporting low latency XPU to XPU connectivity. Back-end fabrics comprised the majority of the AI-driven demand. They link tens of thousands of AI accelerators, demanding maximum bandwidth, ultra-low packet latency and operate on shorter 2-year refresh cycles with the majority of spend dominated by hyperscalers.

Front-end networks have lower bandwidth and less stringent latency requirements relative to back end. though this domain is still an important part of hyperscale data centers as well. We've seen the evolution of the various technologies in networking, shape our strategy and the different road maps we are supporting. Historically, Ethernet and AI back-end networks supported primarily scale out and emerging scale across fabrics and relied on traditional copper interconnects and pluggable optical transceivers.

As the industry transitions into the 1.6T generation, networking requirements shift dramatically. Historically, scale-up fabrics linking accelerators inside a node relied on proprietary closed architectures. One of the more consequential shifts currently underway is the introduction of Ethernet for scale-up networking, which has already meaningfully expanded our destable content within the network. Furthermore, at these higher bandwidths, traditional copper and pluggable optics again reaching physical boundaries and power consumption, port density and signal integrity.

As a result, we are seeing the introduction of new interconnect technologies, including embedding of optics into the switch with co-packaged optics and near package optics as well as new standards supporting form factors like extra dense pluggable optics. Ultimately, we see pluggable and integrated optics solutions coexisting over the long term with each employed for appropriate use cases. We are supporting a growing number of networking road maps, which diversifies our broader product portfolio, including within each of our hyperscale customers and enhances our value as a trusted technology partner that can help accelerate the path from design to deployment.

Independent forecasts validate the significant market expansion underway in Ethernet data center networking. The TAM is projected to reach $173 billion by 2030. Within that, the high-bandwidth segments defined as 800G and higher, which comprises the majority of Celestica's current portfolio and pipeline are forecasted to account for over 90%, a sevenfold increase. Looking at the markets by layers of the networking fabric, back-end AI networking is the fastest-growing expected to account for over 80% of total market spending by 2030, indexing us with the largest and highest growth segments of the market.

As noted earlier, the adoption of Ethernet for scale-up networking is driving a significant new market opportunity. Scale-up Ethernet spending is projected to grow from negligible volumes in 2026 during the early stages of the 1.6T ramps to $55 billion by 2030 when it is expected to represent 40% of back-end Ethernet switch spending and about 1/3 of the total data center Ethernet market.

From an architecture perspective, scale-up fabrics exhibit dramatically higher port attach rates and bandwidth requirements, orders of magnitude higher, up to 10x and 100x that of scale-out networks and front-end networks, respectively. Scale-up fabrics also require ultra-low latency, custom chassis designs and can effectively leverage open source networking operating systems like SONIC. The requirements of these designs play directly to our strengths in system-level engineering on XPU architectures and our JDM model supporting customization, positioning Celestica for market leadership and scale up Ethernet.

One of the latest projects that serves as a valuable showcase of our networking design prowess is our collaboration with AMD on the DS 6010 switch, which enables the scale of fabric for AMD's Helios architecture. supporting high-speed ultra-low latency connectivity of 72 AMD instinct MI-455x GPUs within a single rack. Alongside AMD's innovative architecture design team, our proven track record designing leading-edge switches made Celestica a uniquely qualified collaboration partner for this project. The DS 6010 represents a milestone in engineering and design leadership as the industry's first 204.8 terabit switch powered by 2 Broadcom Tomax chips in a single 1U chassis. making it 1 of the most technically advanced solutions we've ever built.

The open rack wide compatible design is connected via a high-density cable back line utilizes direct-to-ship liquid cooling and supports full Sonic integration. Our specialized manufacturing capabilities support the complex assembly and rigorous validation testing required for this high-performance design. Through this collaboration, we are enabling speed to market for the industry's leading hyperscalers and frontier AI labs deploying Helios. We expect to begin ramping by the end of the year.

As we've discussed, at bandwidth reaching 1.6T and eventually migrating to 3.2T, utilizing 400 gig service traditional copper trace connectivity and standard pluggable transceivers face steep physical limits and signal integrity as well as power and thermal efficiency and form factor density. emerging approaches such as XPO are addressing some of these challenges with a new pluggable form factor, integrated optics designs such as CPO and NPO, our approaches to board design and connectivity that involve migrating the optical engines closer to the switch silicon. a departure from traditional pluggables where the optical components are contained within a transceiver that's connected to the front panel.

We believe we are in the very early stages of the industry's deployment of integrated optics architectures and that these solutions will progressively see increasing traction in the coming years. We are investing in product road maps to support each of these various approaches to integrated optics and next-generation pluggables in order to enable customer flexibility across different architectures.

Optical standardization efforts like the optical compute interconnect consortium will give hyperscalers vendor interoperability while elevating design partners like Celestica as a critical system orchestrator. However, the move towards integrated optics is not without its challenges. Beyond design, the reliable, scalable manufacturing for integrated optics presents new hurdles that are different from traditional copper connected switches. Celestica has strategically invested in developing our capabilities and processes including expanded state-of-the-art clean room space, robotic assembly and inspection systems, advanced optical fiber insertion where submicron tolerances are required and multistage engineering validation testing protocols to support the high-volume manufacturing of these technologies.

With the investments we are making and 1 of the industry's early proof points with the DS 6005 we believe that we are at the forefront in building our capabilities towards commercializing integrated optics at scale with reliable commercial yields. As previously announced, these capabilities have earned us an award with a hyperscaler customer to co-design and manufacture 1.6T rack-scale co-packaged optics networking solution with production scheduled to ramp in late 2027. As we progress through the development process to be ready for 2027 availability, we believe we are 1 of the first teams in the world to have successfully brought up a functional CPO platform with Broadcom's Tomahawk 6, Davidson Silicon and we are well into the testing phase.

We are currently in discussion with other customers exploring proof of concepts and potential future designs. We believe that interest in these solutions will broaden across our customer base as networking technologies evolve in the coming years. This program award serves as a powerful validation of our early R&D investment and capabilities and positions Celestica as an early mover in commercializing Ethernet switches with integrated optics at hyperscale volumes. Celestica's technical capabilities and execution are reflected in our leading market share across our core customer base of hyperscalers and Frontier AI labs, where we are seeing an inflection in our growth even relative to our peers.

Because we enable our customers' leading edge deployments, we maintain the largest share of the highest bandwidth segments in data center Ethernet networking. Independent research outlets recognized Celestica as the world's #1 provider in 800G and higher, accounting for 31% of revenue and 44% of total port shipments. Furthermore, in Custom Solutions, Celestica holds an even more dominant position with a 58% share of total custom Ethernet switch revenue more than double our nearest competitor, positioning us to capture significant growth as hyperscalers increasingly evaluate and bespoke solutions enabled by our HPS offering.

Turning now to our compute and storage end market. For 2026, revenue is projected to reach $5.6 billion, up an impressive 170%, driven by major ramps in our AI compute business and complemented by solid demand in storage. We continue to anticipate exceptional growth in custom ASIC compute programs in 2027, driven by our multigenerational relationship with Google in support of their TPU platform and our production ramp with OpenAI's Hellopino platform. We expect compute and storage to account for a growing share of CCS revenues in 2027 as these program ramps accelerate.

The trend of customization for optimization is driving incredible growth in the deployment of new custom ASICs. The custom accelerator market is projected to exceed $430 billion by 2030, growth of more than 7x and more than double the growth rate of merchant GPUs. Custom ASICs are expected to account for over 1/3 of the AI accelerator market. These forecasts have accelerated considerably during the past year, reflecting why custom ASIC compute represents such a powerful multiyear tailwind for Celestica and validates our strategic focus on this opportunity.

Our hyperscaler and Frontier AI lab customer base are leading this shift as exemplified by the leading platforms in the marketplace today, the fixed cost of a successful custom accelerator program are surpassed by the benefits of lower cost per chip and superior performance per watt due to customization for workload specificity. As an industry leader in system level codesign and manufacturing, Celestica is positioned as 1 of the premier infrastructure partners in this segment of the market. Celestica's role in supporting the launch of OpenAI's Halapino architecture is another important proof point reflecting the evolution of our system-level design capabilities. This program represents the culmination of more than 2 years of development with our HPS team standing as our most comprehensive and complex project to date.

The inference optimized system leverages the first generation of OpenAI's custom ASIC platform, Halapino, which was co-developed with Broadcom. It has been validated through OpenAI's published tests to deliver higher inference throughput per watt than the comparison systems across 3 tested open models. Celestica is helping bring this breakthrough architecture to life. supporting end-to-end delivery from codesign and engineering, software and testing, straight through volume production. Each unit in the dual rack system hosts a 128 Halpino accelerators utilizing a cable back plane because of the incredible density, it is designed to accommodate extreme mechanical loads and advanced liquid cooling infrastructure.

The system is supported by an Ethernet scale-up fabric leveraging 1.6T switches with Broadcom's Tomahawk 6 Silicon, enabling multi-rack multilevel connectivity in the scale-up domain for up to 2,048 accelerators and runs on a customized version of the Sonic operating system codeveloped with Celestica. It is expected to begin ramping production in 2027. Through our multigenerational relationship, Celestica and OpenAI are engaged in the development of future generations of their platform as we look to continue to build on our successful collaboration together. With that, I'll turn the presentation over to Todd Cooper to review our Advanced Technology Solutions segment.

Thank you, Steve, and good morning, everyone. Two years ago, we undertook steps to strategically reshape our ATS portfolio, prioritizing higher-margin programs and durable growth with Tier 1 customers. In 2026, we've seen the benefit of those as we expect a return to double-digit growth for the year with our ATS segment expected to deliver revenue of $3.6 billion, up 13% and for segment margins to be up 110 basis points to a record 6.4%.

We drive distinctive value in an environment characterized by increasing complexity, regionalization and a greater need for engineering support as well as resilience in supply chain management, allowing us to drive faster time to market in a cost-effective manner for our customers. We believe this core value proposition underpins our strategic customer relationships and positions ATS for continued sustainable and profitable growth.

Looking ahead to 2027, demand tailwinds and new program wins across our portfolio are expected to sustain our momentum and propel ATS to another year of double-digit growth. In capital equipment, strong AI-driven demand is necessitating foundries to make investments in new fab capacity with market forecasts calling for wafer fab equipment spend to grow by more than 40% next year and continue that strong growth in 2028. Our strong relationships with the leading wafer fab equipment OEMs and our differentiated capabilities position us for robust growth. In Industrial, we expect solid growth on improving market trends complemented by ramping business in energy generation, storage and metering along with our traditionally strong test and measurement and telematics markets.

Further, early pursuits and data center power infrastructure are also showing promise. supported by our investments in specialized engineering teams to enable solutions for this market. Both Aerospace and Defense and the health tech sectors are expected to see solid base business demand with growth driven by new program ramps in particular, with increased defense demand across the U.S. and Europe as well as in advanced surgical robotics in diabetes care. We believe our strong position market tailwinds and targeted commercial strategy will enable us to deliver another year of solid growth in 2027.

Our commercial strategy is tied hand-in-hand with our engineering-led investments aimed to unlock greater value add for our customers and deeper long-term engagements. With our growing engineering services team of nearly 400 engineers, we are continuously investing in specialized technical talent with dedicated capabilities across key submarkets. By engaging with customers early in their product life cycles, we provide high-value design support, rapid prototyping and automated testing. ATS Engineering Services revenue is growing at more than 30% CAGR since 2023 with new wins seeing an attach rate of 50%, validating the opportunity and our engineering value proposition.

We expect higher engineering-led mix, combined with improving operating leverage will continue to expand our segment margins. We remain disciplined in managing our commercial portfolio, balancing our exposure across proven Tier 1 anchor customers with newer accounts that show the potential for high growth in emerging technologies. We believe this strategy has built a strong foundation that positions ATS for sustained high-quality growth over the long term. I'll now turn the call over to Todd Ankenmann to discuss our financial outlook.

Thank you, Todd. Turning to our outlook for 2027. We see another year of accelerating revenue growth driven by unprecedented customer demand. We anticipate revenue of $40 billion, representing a 90% increase compared to our 2026 outlook, while non-GAAP operating margin is expected to be 9% for the full year, an expansion of 60 basis points. Underlying our revenue assumption is approximately 110% growth in CCS revenue, driven by rapidly growing demand in both networking and compute and storage, while ATS is expected to grow approximately 10%. As we expect to be ramping multiple large-scale new programs throughout the year, we anticipate sustained sequential growth, leading to meaningfully higher revenue in the second half versus the first half of 2027.

We anticipate non-GAAP adjusted EPS of $21.50, up 86%. Our outlook assumes approximately 127 million diluted weighted average shares outstanding for the year. Capital expenditures are expected to be in the range of $1.5 billion to $2 billion, which directly supports the programs we have won to date. Our outlook assumes an adjusted effective tax rate of approximately 21%. Our financial outlook factors in our best assessment of dynamics, including the impact of supply constraints and timing of capacity ramps against a backdrop of higher customer demand. I want to spend a moment diving deeper on the dynamics underpinning our 2027 revenue outlook and our view of the longer-term multiyear growth trajectory. As we've discussed today, constrained supply and long design collaboration cycles have lengthened demand visibility.

These multiyear customer commitments give us strong conviction in our growth investments and the sustainability of the demand we are witnessing today. In fact, our 2028 customer demand forecast based solely on awarded programs are meaningfully higher than our 2027 revenue outlook. When we look across our core product pillars, the anticipated growth drivers are robust and mutually reinforcing. In networking, we expect the already strong momentum we are experiencing to accelerate during the transition to 1.6T bandwidth, which were in the very early stages of today. We expect these ramps to be significantly larger and faster than the 800G upgrade cycle the past 2 years. as we anticipate benefiting from multiple incremental drivers of demand, including Ethernet use and scale up and new customers and ecosystems in our portfolio.

Importantly, recent networking upgrade cycles have shown to be additive to demand, and we expect 800G deployments to continue to grow alongside initial 1.6T rollouts next year. In AI compute, multi-generational program ramps across our existing and new hyperscaler and Frontier AI lab customers continue to grow in scale and are pointing to meaningfully higher volumes in 2027 and 2028. Finally, beyond our committed ramps, we have a significant funnel of pursuits, including a number of potential large-scale engagements with existing and new customers. With our confidence in the sustainability of the demand we are seeing, we are thoughtfully yet proactively positioning the company to capture these opportunities. The achievement of our $40 billion revenue outlook for 2027 would represent more than fivefold growth in just 4 years. Establishing this greater scale continues to enable structurally higher and more resilient operating margins. We expect our non-GAAP operating margin to expand at 9% in 2027, benefiting from design-led wins, cost discipline and scale.

We continue to commit higher levels of CapEx to support the expansion of our manufacturing network. Our plans for $1.5 billion to $2 billion of CapEx next year are intended to enable the scale necessary to support our multiyear growth plans. Additionally, with our 2027 outlook calling for nearly $20 billion of incremental revenue, we are investing meaningfully in working capital to support these large-scale ramps. The combination of these investments will result in a net use of cash in 2027. Over the past several years, we have managed to generate consistent and growing non-GAAP free cash flow, even as we have supported significant growth in revenue.

Next year, however, the level of net cash usage will be a function of the sheer magnitude and velocity of our revenue ramp. Our balance sheet is conservatively financed and well funded. With $4 billion of cash on hand, low leverage and additional $1.75 billion in available liquidity, we believe we are sufficiently capitalized at this time. When assessing potential investments, adjusted ROIC remains an important benchmark for capital efficiency and disciplined execution. Our adjusted ROIC has more than doubled in less than 3 years. and has accelerated meaningfully in 2026 to 57% year-to-date, reflecting the incredible growth and margin improvements in our business and our strong track record of capital stewardship.

The scale, efficiency and profitability profile of our business enabled us to generate consistently and structurally higher returns on capital than at any time in our company's history. We continue to employ our same prudent approach to capital deployment. We believe the investments we are underwriting in 2026 and in 2027 support short payback periods and fast revenue conversion and profit generation. as we phase our investments to align with customer deployment schedules. As such, we expect to continue generating adjusted ROIC well above historical levels.

Our financial growth framework is aimed towards driving long-term compounding of our adjusted EPS in a disciplined and sustainable manner. To achieve this, we evaluate every commercial opportunity through strict strategic filters. We select programs that align with our core competencies and design and manufacturing differentiation, while supporting customers where we see the opportunity for long-term multi-generational engagements. We aim to drive sustainable adjusted operating margin expansion while also managing our mix at a portfolio level.

And we assess investments based on adjusted ROIC hurdles and payback periods across multiple scenarios. This approach allows for risk mitigation, while prioritizing our allocation of capital to the most strategically aligned projects with strong anticipated returns. As a management team, we pride ourselves on being trusted, thoughtful and responsible stewards of capital. Our financial health is foundational to our capital allocation plans. as we aim to support our multiyear growth trajectory while preserving financing and investment optionality. We are committed to maintaining a healthy and flexible balance sheet, managed with investment-grade discipline.

Organic growth remains our top area of investment. As we grow our engineering and commercial resources alongside major investments in our asset base to support higher levels of working capital and CapEx to grow our global manufacturing network. In this hypergrowth environment, we find ourselves in, we believe these are high-return investments and the best use of our capital. We also look to leverage strategic M&A opportunistically to accelerate key capabilities and to grow our presence in existing or adjacent markets. Finally, we remain opportunistic in our approach towards returning capital to shareholders through the intended renewal of our NCIB program. With that, I'd like to pass it over to Rob for his closing remarks.

Thank you, Todd and team. As we close out our prepared remarks, we want to revisit the pillars that are the foundation of our current success and that we continue to build upon for the future. Celestica find itself at the threshold of a truly remarkable opportunity and what I believe is the most exciting period for our company and our entire industry during my tenure as CEO. Our conviction is supported by our industry-leading positions in some of the fastest-growing segments of the AI data center build-out and a multiyear demand outlook was structurally improved visibility with our leading hyperscale and fronter AI Lab customers.

Our differentiated capabilities and technology leadership position us as a critical multigenerational partner. And our disciplined investment strategy has yielded us a proven track record of value creation. We hope today's discussion has provided you with a deeper understanding of our differentiated capabilities and our deliberate strategic positioning to capitalize on the unprecedented growth we see ahead. In closing, I want to extend our appreciation to our more than 34,000 global employees for their incredible efforts that make this all possible and to our valued shareholders for your continued trust.

Thank you all for joining us today. We'll now start the Q&A portion of our call, where we'll also be joined by Gavin Cato, Senior Vice President, CCS Engineering. I will now turn the call back over to the operator to open the line for your questions.

[Operator Instructions] Your first question comes from David Vogt with UBS. Please go ahead.

And I'm sure we're going to get a lot of more color later today at the investor in vet. But I wanted to ask your view of the '27 outlook.

[Technical Difficulty]

Ladies and gentlemen, we are currently experiencing technical difficulties. Please stand by as we resolve the issue. Thank you, ladies and gentlemen. We appreciate your patience. We will now continue the call. Your first question comes from David Vogt with UBS.

Guys, can you hear me? I just want to double check.

We can hear you fine, David. Thank you. Sorry about that.

So I was just asking, so obviously, you gave a lot of color, a lot of detail. We're going to get more at the investor event later today. But can you kind of speak to the demand drivers versus supply chain kind of how you're thinking about how we should sort of think about whether it's silicon, optic supply, some straight supply, your own manufacturing capacity supply in terms of how do we remedy kind of the $40 billion number versus kind of the supply chain dynamics that are in the marketplace? And how confident do you feel in terms of your ability to navigate a lot of the sort of challenges and hurdles that are out there in the marketplace today. And then I'll turn it over since you guys got cut off.

Yes. Thanks, David. It's Todd here. Look, our $40 billion guide took into consideration the ramps, the component availability and the operational execution, David. a 90% year-to-year growth, we feel high confidence that, that number is supported. There are obviously opportunities to exceed that number due to those same factors. But when you look at the fundamental drivers [Technical Difficulty] a few things. In terms of engineering, we continue to build. We've been adding, in general, about 500, 600 engineers a year, that's ramping a little bit more recently.

We are getting them across the global sites. We have 9 different locations. And so we're pulling them from the different sites, primarily Austin and Taiwan. have been ramping up significantly recently as well as Silicon Valley. So we are significantly building the U.S. presence right now in the talent plan. In terms of programs, we are engaged across the '27 programs, 2027 programs, so we're wrapping them up. Most of those are in test phases at this point. we are engaged actively and pretty heavily now in programs that are moving 2028 moving forward.

So it is a forward leaning investment, and we're scaling that investment. So at this point, we probably have 60 to 70 programs that we're actively again. I think did that hit all your points, Scott.

Yes. Yes. you talked about other areas in system maybe elaborate on, yes.

Yes, sure. From a system level, networking is core. We continue to expand into the compute space. So we are seeing additional compute racks coming in. So that rec engagement rack level work is a primary focus for us. We do see opportunities to look at both memory and storage in terms of opportunities that they provide as rack architectures change and we are seeing advancements in optical OCP, the work essentially in all the technology functions that underlie. So you're seeing some diversification in the capabilities in the rack themselves. And so we're covering that ground. .

Your next question comes from the line of Joseph Cardoso with JPMorgan.

Guys, maybe I wanted to change the -- shift the conversation into kind of the cash commentary that you guys have put out there. First and foremost, any way that you guys can put a range around 2027 net cash use and how you're thinking about cash cycle days that compose it and then maybe more medium term, what needs to happen for free cash flow conversion to turn positive again? And then are we on a path toward potentially that towards exiting 2027 or does this level of demand environment kind of keep that bogey in flux? I appreciate any kind of color you can provide here. .

Yes. Thanks, Joseph. It's Todd here. Yes. We reset '26 the free cash flow forecast based on the increased demand outlook in '26 going up $0.5 billion to $21 billion. And when you look at our '27 view of the $40 billion and looking at the free cash flow there, it is going to be a year of investment to support that incremental $20 billion of revenue. As you're aware, we did raise $3.5 billion of equity in August, and we feel they should be sufficient to cover our needs. We have more than enough capital to execute to our current outlook with $4 billion of cash and additional liquidity available. When you look at 2028 and beyond, we feel comfortable based on our current view, that we're confident we'll be able to target to return to positive cash flow in '28 Joseph. .

Your next question is from Karl Ackerman with BNP Paribas. Please go ahead.

You spoke about the rising complexity of next-generation switch designs, including the transition of liquid cooling and integrated optics on the Board and switch silicon. How much of an influence on your improved margin outlook is coming from the rising complexity of these next-generation switch designs? And are you also able to improve margins on your server business as complexities rise there also as well as your ability to support L11 capabilities that offer this full Rail test approach.

Thanks, Karl. This is Steve. Yes, as we think about the different products that we're supporting for our customers and the things that are most attractive to us, we have the ability to be somewhat selective in terms of the things that we invest in and the things that we want to offer to our customers. And margin is certainly a factor that we consider the products and the deal structures that will enable the best margins for us. We have -- sorry. Do you want to...

I'll add to that, Steve. When you think of it, looking at our gross margins, exiting '26, we feel pretty strongly that our margins will continue in '27 in the low 11s. It will be on dependent on mix between networking and compute. and we still see networking being accretive from a margin perspective. When you look at kind of that, that's leading to the continued based on the additional revenue volume, the operating leverage that we're getting on our op margin and in the ability to expand to 9% in 2027.

Your next question comes from Mehdi Hosseini with Susquehanna International Group.

You did mention that 1.6T revenue should exceed 800G. Should I assume that majority of the transceivers are still going to be pluggable or a mix of density plugable and traditional pluggables?

This is Gavin. Good question, Matt. I think what we're going to do in 1.6T is certainly see a lot of the pluggable remain I think you'll see some efforts to prove out CPO capabilities are to prove out and look at NPO type capabilities. But I think you're going to see pluggables remain in the 1.6 time frame. That's just the nature of the business. I think as you start to see 3.2T and 400 gig 30s, I think that's where you're going to see CPO play more predominantly.

Great. I actually have 1 quick follow-up. I know we're supposed to ask 1 question, but given all the disruption. I think the computer and storage growing relatively faster than networking in Tony 7. Is that specific to these ASIC projects that you have? Or is that the combination of networking more of a back-end loaded waiting for more pluggable come to the market. How should I think about the difference in the growth between networking and compute?

This is Steve, Mehdi. I think that what we're seeing is what we're focusing on. We've done a lot of things to diversify the revenue streams that we receive from our customers and where we participate. And we've had this strong industry-leading position in networking. And now we're expanding and taking some of the core competencies and the capabilities that we have and that we've developed to be so successful in networking and taking it into the AI compute space. And so we see a lot of growth just as a factor of the markets that we're addressing the opportunities that we're addressing and where our customers are focused in terms of custom solutions around the switching or the compute ASICs. Thank you.

Yes Todd here. I think the 2 key drivers in compute, looking at the TPU opportunity and then our OpenAI opportunity that we've announced before will ramp in '27, but it will be second half weighted.

[Operator Instructions] Your next question comes from the line of John Shaw with TD Cowen ahead.

I heard that the OpenAI program is going to be H2 ramp up. How about the other new programs that's pretty much networking content heavy and given some of these programs can be pretty material in terms of moving the needle. Do you think the key metrics such as growth, gross margin is going to be different in H1 versus H2?

Yes, John, No, I don't think there'll be different gross margin dynamics in '27 between the 2 halves. There'll be -- it is dependent ultimately on the mix in the individual quarters. But we feel overall, we'll have strong margins in '27 at 11%.

Your next question comes from the line of Daniel Zhu with Bernstein.

Yes. Congrats on a great set of results. The different scale of design that you worked on for OpenAI Holopenia focuses on expanding the scale of domain beyond a similar single rack. And we're now seeing new designs from Broadcom and NVIDIA that operate on similar design principles, is there anything you can share about how the work you've done on hope has been received in terms of trying to win new programs with other potential customers that are trying to solve similar problems that OpenAI is.

Yes. I think -- this is Steve. I think that the work that we've done with OpenAI and the Hellapino program is really serving as a strong proof point of our capabilities. it certainly led to other opportunities. And just the way the ecosystem that OpenAI is engaged with is constructed. We will find a number of other opportunities to engage with other companies using this technology. Certainly, as I said, it's a proof point that we can demonstrate the strong capability that is attractive to other companies as well. So we have a strong funnel of opportunities. We're being somewhat selective about where we engage. But I think from a capability standpoint, certainly is an important program for us.

Yes. I think 1 of the things that we see is the processes and the techniques that we've used to really look at the rack and delivering a full rack solution for OpenAI that translates, whether you're looking at full network racks or you looking expansion of compute across multiple racks. That end-to-end solution is something that I -- from the standpoint of are design capabilities or testing or validation and the way we translate that into a manufacturing capability is certainly differentiated at this point in time. And so it's helping us in those engagements.

Your next question comes from the line of Michael Ng with Goldman Sachs. Please go ahead.

I just wanted to ask about your CapEx outlook for 2027. Just given the strong pipeline that you have for 2028 and the very good ROICs that you're seeing on capital deployed to date. How do you balance investing additional capital with the ROICs and like making sure that you have enough capacity to capitalize on all the demand opportunity they have.

Yes, Michael, we're always looking at our capacity requirements. But when you look at the CapEx forecast for '27 of 1.5 to 2, those improved investments are for awarded programs and already won and support the capacity needs for '27 and '28. We do have a strong process of looking at multiple different scenarios and outcomes around each of the investments. All of them do have strong ROIs and paybacks and support our ROICs continuing to be above our historical levels. .

We will reevaluate with new wins and new things as they come in, the needs of the capacity on a continuous basis where we have a robust process as a leadership team to do that.

Your next question comes from the line of Ruben Roy with Stifel. Ruben.

Yes. Great. Rob or Todd, it's great to see the execution and the commentary touched on 2027 guidance but the stronger outlook extending into 2028. I'm wondering if you could comment on how you're viewing that 2028 dynamic relative to follow-on generations. You talked about what's going on with Google and multiple generations at OpenAI. Is that driving sort of the confidence into 2028 or is there a portion of unawarded opportunities that you're considering? Just trying to trying to understand how you're thinking about the longer-term dynamic there on extensive visibility? .

Ruben, it's Mandeep. Nice to talk to you. Look, we're very confident right now in 2028. The demand visibility is getting longer and longer. We know that material lead times were what was originally driving that but with the capacity build-out plans that are needed to build data centers, it's extending even beyond that, we're having conversations right now with customers about programs that are going into 2029 and in some cases, even 2030, we're being awarded programs right now where the silicon has not yet been developed, but it's being -- programs are expected to begin in 2028. And so it's just leading to very integrated conversations with our customers. The design cycles are very intertwined. Our R&D teams are working together.

And so it gives us a lot of confidence. To the points that have been raised by Todd earlier, the views that we have right now on capacity is for programs that have already been won, but the funnel of opportunities that we are continuing to engage with our customers on provide a meaningful level of upside opportunity, and that's what we're focused on. So we're focused on execution, but we're also focused on driving some additional opportunities, which we think that really have a meaningful impact in 2028 and 2029.

Your next question comes from the line of Thanos Macopolo with BMO Capital Markets.

On scale-up, in 2027, will that primarily be Helios and Helopeno? Or are there programs expected to ramp as well? And then heading into '28 will be the dynamic for scale-up. .

Yes. This is Steve. I think as you've described, that's where the predominant revenue will come from. Are those programs that are wins for us and I think it represents an increasingly interesting space for us, an opportunity for us to expand into providing a broader range of solutions to our customers, scale up is something that's very attractive to us and participating in a bigger way. We have a number of opportunities that we're considering in the funnel, but you've mentioned the ones that we've named specifically. but we're encouraged by what we see in this space.

Yes, Thanos. I mean, we have announced that we won ten 1.6T programs. When you look at those, they are -- 4 of them are scale up and 6 of them are scale out. So I think when you look at those growing through '27, that's going to be 1 of the growth drivers going into the networking space.

Your next question comes from the line of George Notter with Wolfe Research.

This is Karen on for George. Could you guys just talk a little bit about your significant funnel of pursuits, specifically the nature of the new large-scale engagements you guys are pursuing?

This is Steve. Yes, I think as it relates to the AI compute space, the hyperscale market in general and the emerging opportunities that we see with next-generation adopters of these kinds of technologies in these models. I think there's a lot of opportunity available to us. It's more a matter of being selective about where we engage. and being very deliberate about where we can actually bring differentiated value to these customers. We think about growing our business and diversifying across several dimensions.

We've done a good job of adding customers expanding our offerings, both in terms of products and value-added services and participating with higher levels of completion on the products that we do for our customers. So we've diversified in several ways, but I think that we also want to be careful about where we expand and what it means to our business. Certainly, we want to continue to execute as strongly as we have for the current base of customers that we're supporting today and looking to selectively add customers that may help us diversify our portfolio going forward. I think when we look at the funnel, we've got a strong demand -- a strong funnel of strong demand from our current customers. And I would say it's at a historic high level. but it's largely coming from our base customers, our core customers and them recognizing the value that we can bring and directing more of their requirements to us.

I think from a new customer perspective, it's a situation of -- it's not always good to have all your eggs in 1 basket, but if you got a very strong basket and you watch it closely, I think it's an acceptable place to be, and that's largely how I see it.

Your next question comes from the line of Steven Fox with Fox Advisors LLC.

Just I was curious if you could dig in a little bit more into the manufacturing advantages you talked about going forward and how they may be as you ramp them hurt the gross margins and eventually help the gross margins.

Yes, Stephen, it's Mandeep here. I would say 2 things. The first 1 is actually reiterating that the CapEx that we're putting in place by the end of 2026 will really enable the revenue in 2027. What you're seeing is very targeted investments in specific areas we're growing significantly in our campus in Thailand, where we have a very robust capabilities that go back, frankly, decades, We're investing in Malaysia. We're investing in Japan, and we're investing substantially in Texas as well.

And what this is really doing is supporting our customers from a regionalization perspective. and really giving them that global end-to-end solution. As a result of having these focused investments, we're driving a very high level of productivity and our operating leverage is continuing to help us, and we're seeing that scaling benefit. And we're being very thoughtful and disciplined on our -- on driving cost leverage and very good productivity in our sites. And to your point on the margin profile, so an area that we're focused on is really to have not only expanded or increased levels of square footage but increased levels of power availability.

And so we are increasing our overall power offering substantially as we go between 2025 into 2026 and '27 and even for plans that we've already made for 2028. And because of those power, those investments that we're making in both square footage and power it's allowing us to capture the program that many of our competitors have not been able to secure. And those are leading to HPS type of engagements. where we're able to then bring engineering to the table, and that's ultimately leading to strong pricing and good margins as well.

So there's a lot of different factors that we put into place when we're pursuing the wins as Steve talked about. We don't pursue every opportunity that's in front of us. We're quite targeted to try and get the ones that we believe are the best for our portfolio.

Your next question comes from the line of Ruplu Bhattacharya with Bank of America.

Congrats to Mandeep and Todd for their new roles. My question is how should investors reconcile the recent 3.5 billion equity raise that you did with the non proposed NCIB covering up to 2.5% of the float. Is the repurchase primarily intended to offset stock-based comp? Or could this be meaningful capital return and maybe you can weave in like how much revenue can your existing footprint support? And do you have any flexibility for acquisitions? .

Yes. Thanks, Ruplu. It's Todd here. Look, we reviewed our -- we renewed our NCIB program to purchase up to 2.5% we'll be opportunistic on the timing. Our -- the priority and focus of our capital is on the organic growth to support the $20 billion growth we see into '27 and the substantial growth we see beyond that into '28. So that will be the focus of where we're going to do that. It is a Canadian requirement to renew the program on an annual basis. So that's 1 of the things we had to pursue and renew the NCIB program, Ruplu.

And on M&A., we do remain -- on M&A U.S., we do have an active funnel, and we've been very targeting that, largely capability-driven at this stage of the game. and we do have the capability to do both.

And part of my question was how much revenue can the current infrastructure support any guidance on that? .

Yes. We won't give a revenue figure specifically tied to capacity because it's unique to the mix of the production and the individual customer programs. But we feel we've got the capacity needed with the investments in '27 to support our requirement for '27 and '28.

Your next question comes from the line of Todd Coupland with CIBC..

Hopefully, this question wasn't asked before I got back on. So I was curious, with the lab starting to ramp more materially in '27 and '28. How would you characterize their ability to make the spend that you're building into their forecast? In other words, do they have the financing in place now? Or is that required? Just frame that up for us for '27 and '28. .

Yes. We're always looking at our customers are well funded. We see no financial distress across our portfolio. We engage regularly on their funding and liquidity, and we see no concerns that they have the liquidity to support the outlooks. We do look at their third-party funding support their contractual back when necessary. But at this point, we feel our customers are well funded to have liquidity to support the business as we see it.

One thing I'll just add to that, Todd, is like, as you know, because of the lead times of materials, we're having conversations with our customers on what their demand outlook is 3, 4, 5 quarters out in a very detailed way. And then we're bringing that inventory in. When we're bringing that inventory in, it's assessed on the credit that they have in the moment. And so we do have very open conversations. We have no concerns with some of the largest companies in the world. We do know that other customers are growing very rapidly from a smaller base. And so we have closely tied with them on what their credit is and we don't have any concerns at this time.

Your next question comes from the line of Robert Young with Canaccord Genuity.

Over the last few quarters, you've noted several times that you're carrying very high levels of utilization in the factory, you have not really given historical quantitative numbers around that. But given this enormous amount of revenue that you're adding in 2027 and the capacity expansion, how confident are you that you can keep the utilization that high? How important is that to maintaining the operating margins and getting to that 9% number. Maybe if you can give us just some historical context around how you've been able to manage these ads to capacity and keep utilization high, whether that's important. .

Rob, it's Mandeep here. Look, high utilization isn't by accident. We're deliberate and when we try and drive it. And we've been able to do that successfully now for many, many years. When we're adding on capacity, we want to make sure that the white space is being used, and it's being used to the highest returns. So back to a point that Steve was making earlier, we do compare different opportunities in our funnel knowing that we can't bring on everything into our factories, and we want to make sure that we get the best yield for it. we are running at a high level of utilization, specifically in the factories that I mentioned. When we're adding on capacity, it's because they run out of space and -- but we don't want to add on just more space and then go and fill it.

It's tied to a business case, and so we want to know that we can fill that capacity in a relatively short period of time. In Asia, we're able to get new capacity online sometimes in as little as 12 months. takes a little bit longer in the U.S., but at the same time, the road maps that we have with customers match up to us. But we do expect high utilization in 2027, and we are working towards high utilization in 2028 as well.

Your final question comes from the line of Paul Treiber with RBC Capital Markets.

Just a follow-up to 1 of Mandeep's previous comments just in regards to competition in regards to power availability and manufacturing capability, which have been helping drive wins. The -- is supply availability become more important than design and engineering capabilities for next-generation wins? Or is it very balanced between those 2. And then can you comment on the competitive ecosystem overall, particularly for full rack solutions, just given increasing complexity and scale, are you seeing fewer competitors that can pursue these programs. .

Yes. This is Steve. Some important questions you've raised here and something that we're always being very mindful of is just -- what is it that will drive the outcomes that we're looking for from our customers. And certainly, the design piece and the ability to support leading-edge technologies, making the investments as we've outlined in R&D and building these strong ecosystem partnerships with key technology providers is really critical to our ongoing success, we feel. So this is a priority.

We also look at what capacity that we can make available in the time frames required. And we're fortunate, I would say, we're not in a situation of having to compete per share on a regular cadence on a regular basis, much like you would expect in a more common or commoditized product space. We are dealing with more complex, more bespoke kinds of solutions for our customers. And with that, there's a fair amount of co-investment. And with that, the customers and ourselves are aligned on maximizing the utilization and getting the maximum output from those investments. So I think in that regard, that's the second thing. And I believe you may have added a third part to that question.

Yes. Just the overall competitive ecosystem and increasing complexity and scale, are you seeing fewer competitors.

Yes. So certainly, where we believe we've had a lot of success in this space. It's going to invite competition. We do have additional competitors. But as I mentioned, we're kind of in a space where our share is derived by a number of factors, and it's something that we usually are negotiating with our customers on a regular basis. As I said, they have a mutual interest in getting the maximum return on the investments they've made with us. but also where our products are in the life cycle is also a factor. For example, a customer may choose to move an older product away from us to ensure that we can have focus and capacity available for next generation of new products. So I think from a competitive standpoint, I think we're in a very strong space.

I'll just add 1 thing, Paul, to your point on the competitive landscape and the prepared remarks that we made in the prepared remarks, we really are focused on building competitive boats in the areas that we're focused on, on the networking side. the complexity of the products are continuing to grow as you go to every subsequent generation. And so there are fewer competitors today that we are going up against on the 1.6 and on the 800G programs than they were on earlier generations. And we think that as we -- as the industry moves towards things like 3.2 and a scale up becomes more broad-based we think that our offering is going to be very, very well received because of many years of investment.

On the enterprise side and then on the compute side, as Gavin talked about, the solutions that we have within OpenAI are really a good proof point of where a lot of the industry is going, and that's the integrated rack. and being able to be vertically integrated and having proficiency in both compute and networking being able to put it all together and then liquid cool that rack. And then in addition to that being able to bring services to it is difficult. And so as we continue to have proof points there, we do believe that not all competitors will stay for the journey. There certainly are competitors, but we do think that we're building more and more proof points, which are being well received by our customers.

We have reached the end of our Q&A session. I will now turn the call back to Rob Mionis, CEO, for closing remarks.

Thank you all for joining us. Obviously, the folks hosting the call. We're not using Celestica designed networking or compute hardware. We look forward to seeing most of you, if not all of you this afternoon. And then if we didn't get to your questions, we'll certainly try to get to them this afternoon. Thanks for your continued support. Have a great day. .

This concludes today's call. Thank you for attending. You may now disconnect.

Hello, everyone. Thank you for joining us, and welcome to the Celestica Second Quarter 2026 Financial Results and Conference Call. [Operator Instructions]

I will now hand the conference over to Matthew Pallotta, Head of Investor Relations. Please go ahead.

Good morning, and thank you for joining us on Celestica's Q2 2026 Financial Results Conference Call. On the call today, we have Rob Mionis, Chair of the Board and Chief Executive Officer; and Mandeep Chawla, Chief Financial Officer.

Please note that during the course of this call, we will make forward-looking statements, including statements relating to the future performance of Celestica, our business outlook, guidance for the third quarter of 2026, our 2026 annual outlook and anticipated trends in our industry and their anticipated impact on our business. These are based on management's current expectations, forecasts and assumptions as of July 27.

These forward-looking statements are subject to risks and uncertainties that could cause actual results to differ materially from our expectations and their potential impact on our results cannot be reliably predicted at this time. For identification and discussion of the material assumptions, risks and uncertainties, please refer to our public filings with the SEC and on SEDAR+ as well as the Investor Relations section on our website. We undertake no obligation to update these forward-looking statements unless expressly required to do so by law.

In addition, during this call, we will refer to various non-GAAP financial measures. We have included in our earnings release found in the Investor Relations section of our website, a discussion of those non-GAAP financial measures and a reconciliation to the most comparable GAAP measures. Unless otherwise specified, all references to dollars on this call are to U.S. dollars. All per share information is based on diluted shares outstanding. All references to comparative figures are a year-over-year comparison.

With that, I will now turn the call over to Rob.

Thank you, Matt, and good morning, everyone, and thank you for joining us on today's call. As we cross the midyear mark, we continue to see sustained and accelerating momentum across our portfolio. In the second quarter, we delivered revenue of $4.7 billion, and adjusted EPS of $2.54, both exceeding the high end of our guidance ranges. This performance was backed by an adjusted operating margin of 8.2%, a new high for the company. These results reflect strong operational execution across both segments alongside record demand from our CCS customers. .

Within CCS, we continue to expand with leading hyperscale customers through close alignment on multiyear capacity road maps. Rapid progression in technology upgrade cycles in networking and AI compute are driving broad demand, and we are in the process of ramping a number of significant new programs expected to launch in the second half of the year and into 2027. At the same time, our ATS segment continues to see strong profitability improvements alongside solid revenue growth during the quarter. Before I provide an update on our longer-term business outlook, I would like to turn the call over to Mandeep who will take us through our Q2 results and Q3 guidance. Mandeep, over to you.

Thank you, Rob, and good morning, everyone. Revenue in the second quarter was $4.70 billion, up 62% and above the high end of our guidance range as strong demand and solid operational execution led to higher-than-expected revenue in our ATS segment and both of our CCS end markets. Our non-GAAP operating margin was 8.2%, up a robust 80 basis points with solid margin expansion in both segments, resulting primarily from improved operating leverage. Our adjusted earnings per share was $2.54, exceeding the high end of our guidance range and an increase of $1.15 or 83%.

Moving on to some additional metrics. Adjusted gross margin was 11.5%, down 20 basis points, largely due to mix in our CCS segment. Our second quarter adjusted effective tax rate was 20%. Lastly, our strong earnings growth and disciplined working capital management led to adjusted ROIC of approximately 55%, higher by 20 percentage points compared to the prior year.

Turning now to our segment performance. Revenue in our CCS segment was $3.81 billion, up 84% compared to the prior year period. driven by very strong growth in both our communications and enterprise end markets. In the second quarter, the CCS segment accounted for 81% of total company revenue. Our communications end market revenue was up 62% higher than our outlook of approximately 50%, primarily driven by growth in our 800G networking switch programs complemented by continued strong demand in our 400G programs. Revenue in our enterprise end market was up by 167%, well above our outlook of approximately 130% and driven by the accelerated ramp of an AI ML compute program with a hyperscaler customer and supported by stronger-than-expected demand in storage programs.

In our HPS business, revenue was $1.9 billion in the second quarter, representing growth of 58% driven by ranking 800G switch programs across multiple hyperscaler customers. Our HPS business accounted for 41% of total company revenue in the second quarter. ATS revenue for the quarter was $888 million, higher by 8% and above our outlook of a mid-single-digit percentage increase. The solid performance was driven by revenue growth in each of our businesses. Our ATS segment accounted for 19% of total company revenue in the second quarter.

Moving on to segment margins. Our CCS segment margin was 8.7% in the second quarter, an improvement of 40 basis points driven by strong operating large from significantly higher volumes. Our ATS segment margin was 6.3%, up 100 basis points year-to-year, driven by operating leverage and higher engineering-driven product mix. In the second quarter, 3 customers each accounted for 10% or more of total revenue, accounting for 32%, 17% and 14%, respectively.

Turning to working capital. At the end of the second quarter, our inventory balance was $3.4 billion, a sequential increase of $0.7 billion and higher by $1.5 million compared to the prior year, driven by continued scaling to support significant growth and update program ramps in our CCS segment as anticipated. Cash cycle days for the quarter were 47%, representing a 19-day improvement over the prior year and an 8-day improvement sequentially.

Moving on to cash flows. We generated $147 million of free cash flow in the second quarter and $285 million year-to-date. Our capital expenditures were $264 million or approximately 5.6% of revenue compared to 1.1% of revenue in the prior year quarter. Consistent with our prior outlook, our full year 2026 capital expenditure guidance remains unchanged at approximately $1 billion. Our capital expenditure investments are intended to support the significant anticipated growth from currently ramping and awarded programs. They are underpinned by strong demand visibility and close alignment on multiyear capacity planning with our key customers.

At the end of the second quarter, our cash balance was $536 million, while our gross debt was $740 million, resulting in a net debt position of $204 million. We had no draw outstanding on our revolver at the end of the quarter. Our gross debt to non-GAAP trailing 12-month adjusted EBITDA leverage ratio was 0.5 turns, an improvement of 0.1 turn sequentially and 0.4 turns versus prior year. Our cash balance, in combination with our recently upsized revolver, provides us with approximately $2.3 billion of available liquidity.

As of June 30, we were in compliance with all financial covenants under our credit agreement. We did not repurchase shares for cancellation during the second quarter. While we continue to approach share repurchases on an opportunistic basis, our highest priority at this time is to reinvest in our business to support the unprecedented growth we anticipate over the coming years. We maintained a disciplined approach to capital allocation, deploying to the highest return opportunities in order to drive long-term shareholder value.

Now moving on to our guidance for the third quarter. Third quarter revenue is projected to be between $5.25 billion and $5.55 billion, representing growth of 69% at the midpoint. Adjusted earnings per share is anticipated to be between $2.88 and $3.08, representing an increase of $1.40 at the midpoint or 89% growth compared to the prior year. Assuming the achievement of the midpoint of our revenue and adjusted EPS guidance ranges, our adjusted operating margin is expected to be 8.4%, which would represent an increase of 80 basis points. We anticipate our adjusted effective tax rate for the third quarter to be approximately 20%.

Finally, let's review our revenue outlook for each of our end markets. In our CCS segment, we expect revenue in our communications end market to grow approximately 60% driven primarily by hyperscaler led growth in 800G programs as well as the commencement of mass production ramps in our first 1.6 terabit programs. In our enterprise end market, we expect growth of approximately 190% driven in large part by the continued ramp in hyperscaler AI ML compute as well as improved demand in storage. In our ATS segment, we anticipate revenue to be up in the mid-teens percentage range, driven primarily by strong demand in our capital equipment business, as well as by new program ramps in our other businesses.

With that, I will now turn the call back over to Rob to provide an update on our 2026 annual financial outlook and additional color on the latest development in our business.

Thank you, Mandeep. We are once again raising our full year 2026 annual outlook driven by our strong first half performance and strengthening second half customer forecast supported by improved component supply. Our revenue outlook increases from $19 billion to $20.5 billion, representing very strong growth of 65%. We are also raising our adjusted EPS outlook to $11.30, up from $10.15, representing growth of 87%. The reflected in our outlook is full year adjusted operating margin of 8.4% higher than our previous outlook of 8.1%.

Finally, we are increasing our free cash flow outlook for 2026 from $500 million to $600 million, which includes approximately $1 billion in planned CapEx. Our outlook maintains a prudent view on component availability and the timing of new program ramps. Looking beyond 2026, our visibility is very strong. This is driven by multiyear global capacity alignment with our key customers, along with improved component supply planning to effectively manage extended component lead times.

For 2027, as a result of exceptionally strong customer demand, and new program wins, we expect our revenue growth rate to accelerate as compared to the 65% growth expected in our 2026 annual outlook. Furthermore, as we continue expanding our adjusted operating margins, we anticipate our adjusted EPS growth rate to outpace our revenue growth. We look forward to providing a detailed full year outlook for 2027 later this year.

Now moving on to discuss the business outlook for our segments and beginning with CCS. Based on our updated 2026 annual outlook, we now anticipate CCS revenue to grow by approximately 85%. In Communications, we expect strong momentum to continue through the rest of the year, driven by continued growth in 800G Ethernet switch programs alongside the launch of mass production for 1.60 programs with 2 hyperscaler customers beginning in the third quarter. In 2027, we expect that the combination of continued growth in 800G volumes alongside the acceleration of 1.60 ramps across our customer base will drive very strong growth. Within our enterprise end market, our outlook for the second half of 2026 remains exceptionally robust as we see strengthening demand in our AI compute portfolio.

We expect to continue this strong growth into 2027, driven by the ramp of multiple compute programs. Additionally, Later this year, we expect to commence initial deliveries of custom racks for our digital native customer whom we are pleased to share is OpenAI. In our role as a strategic systems partner Celestica will collaborate with OpenAI and Broadcom in support of their multigenerational customer accelerator road maps, beginning with a Jalapeno accelerator with mass production planned in 2027. As OpenAI's custom and accelerated road maps advance towards future generations, Celestica will continue to provide its expertise to support their multi-gigawatt scale deployments.

Now moving on to our ATS segment. We are updating our full year outlook, which now calls for revenue growth of approximately 10%. Growth is accelerating in the second half of the year, largely driven by sequential growth in our capital equipment business, where stronger wafer fab equipment demand tailwinds are anticipated to continue into 2027. This momentum will be complemented by continued growth across the rest of our portfolio, supported by new program ramps. We also remain positive on the trajectory of strengthening ATS segment margins. We anticipate that this momentum will continue as the portfolio benefits from both improving mix and leverage from returning revenue growth.

Looking ahead, our demand pipeline remains at record levels across key customer programs, we continue to see improving multiyear visibility into AI infrastructure investments, which reinforces our confidence in our growth trajectory through the remainder of 2026 and into 2027. Combined with strong operational execution and disciplined capital allocation, we believe we are well positioned to expand margins, deliver EPS growth and return strong long-term value for our shareholders. With that, I will now turn the call over to the operator to open the line for questions.

[Operator Instructions] Your first question comes from the line of Tim Long with Barclays.

Yes, one question, and then I'll have a follow-up after. Maybe I know you don't want to get too into the 2027, but the acceleration of growth is obviously pretty meaningful. So we're looking at, I don't know, $13 billion or so of incremental revenues. Could you just maybe at a high level, talk about how much those new programs are contributing. I think as the OpenAI that you mentioned in 2 others, AMD and 1.6 CPO switch. So if you could just give us a little bit of color on kind of new versus existing? And then I also have a follow-up right after that.

Sure, Tim. What we're seeing in 2027, first is we see scaling of 800G programs and also the scaling of the 1.60 switch programs and this is complemented on top of very strong growth or I should say, continued growth of 400. So 400G, as we go from '26 to '27 is remarkably resilient. Then on top of that, half of 2027, we have ramps of next-generation AI ML compute programs. And then to your question, we have our new customer deployments with OpenAI and AMD, both are several billion-dollar opportunities that we see in 2027, specifically on the AMD Helios scale-up platform. The pipeline on that program continues to grow. And again, that's a multibillion dollar program in 2027. On top of that, ETS is also really coming into its own. Its margins are improving and the growth fueled by capital equipment is really adding fuel to the fire for 2027.

Okay. Great. And if I could just follow up on gross margin. It seems like a solid performance in the quarter. Maybe just a little bit how we should think of that as the complexion of the business changes. A lot of these newer programs are pretty scale and some of them, I think they're mostly HPS. So with just a high level on profitability and how HDF can impact that over the next year.

Yes. Tim, it's Mandeep here. So we're pleased with where our gross margin is coming in at right now, up a little bit sequentially, and we do believe that we should be able to maintain being around the mid-11s as we go through 2026. As we look into 2027, still dialing in some of the specifics to your point, we do have some positive benefits coming through. We're going to be seeing accelerated growth in HPS, the ramping of network switches, which are primarily in HPS. And that definitely helps us. We're still working through the mix impact as we look at the rest of the portfolio.

But what I'll point you back to then is operating profit. We have a lot of confidence right now that we will see operating profit expansion for yet another year as we go into next year.

Your next question comes from the line of Joseph Cardoso with JPMorgan.

Maybe as a follow-up to the revenue guidance, it's great to see the increase here for this year and next year as well. maybe just given the context of the broader supply constraints we're seeing in the industry, both perhaps direct and indirect. Can you maybe talk to how those have tracked as the quarter or quarter today? And to what extent are they acting or still acting as limiters to the guidance? Are you seeing it kind of much materialize worse? Or are you seeing it more easier to navigate?

Thanks, Joe. Yes, we performed better than we expected this quarter. The team has really been outstanding and navigating a very tight supply chain. Demand across the AI data center infrastructure continues to outpace the global component supply. However, in our '26 commitments and our '27 outlook, we are appropriately hedged. And frankly, we find it manageable. We're continuing to work closely with our customers and our suppliers to secure even improved capacity as we move into 2027. And it's an ongoing process. But ultimately, the positive side of this is that the supply tightness that we're seeing really reflects the strength in the multiyear durability of the underlying demand pipeline that we see. .

And frankly, the unprecedented levels of demand visibility is enabled by extended lead times. So this supply chain constraint is really helping us do a better job of planning with our suppliers and with our customers, and it's giving us confidence in our longer-term outlook.

Just one thing to maybe add to that would be if you just look at it holistically through '26 and through '27, we have the capacity in place to execute the demand that's there for 2026. We have the capacity plans in place to execute the demand in 2027. However, the constraint is really around materials. And so we've taken that into our consideration when providing the figures that we're providing, but the demand does exceed the revenue figures that we've been sharing.

Your next question comes from the line of Ruben Roy with Stifel.

Rob, to start with, can you maybe spend a minute on how you're thinking about 1.6 ramps. There's been a lot of discussion about pace of those rents, et cetera. Obviously, 800 gig is continuing to do well for you guys and you're talking about continued expansion in '27. So I'm just wondering if you could maybe talk about what customers are thinking in terms of 1.6 ramps and if you think there's a crossover coming in '27 or if not, if you can give us how you're thinking about the mix, that would be helpful.

Sure. So 1.6 really starts ramping coming into its own in the back half of '26 and really surging in 2027. And again, we have 10 active 1.60 programs in the mix, and they'll start really picking up in 2027 on top of some very strong growth, year-over-year growth in 800G. On top of that, the pipeline for 1.60 in terms of sales pipeline is continuing to grow, and we're probably will end up booking some additional awards here in the coming quarters. And there's some programs still in development that will also start kicking in, in the latter half of 2027.

I know I have one question, but if you could maybe just comment on CPO as well. I know you have the one win. Is that sort of on track with the way you're thinking about timing? And any additional customer discussions around CPO.

The samples for that CPO win will come out in the first half and mass production for that program will start in the second half of 2027. And we continue to work with that customer on next-generation technologies to kind of further the technology road map.

Your next question comes from the line of Michael Ng with Goldman Sachs.

I just wanted to ask for a few more details on the 2027 revenue growth guidance improvement. Could you just talk a little bit about what changed most significantly to raise the growth outlook? Was it more on the communication side or the enterprise side, any specific programs or deals that kind of give you that confidence and visibility to increase the outlook at this stage of the year?

Sure. So just some 90 days ago, our AI/ML compute demand has significantly increased for 2026 and also leading into 2027. That has given us confidence the demand pipeline for the next-generation racks for AMD next-generation racks also continue to increase. And then our base demand across networking for 800G programs has also just dramatically increased for 2027. As Mandeep mentioned earlier, we do have the capacity to support that demand. And hence, we raised our outlook for 2027.

Michael, maybe I'll just add on to Rob's comment. We talked about this a little bit over the last 6 months or so, but the capacity planning discussions we're having are allowing us to have some very forward-looking conversations with our customers. we're having demand conversations in 2027, for 2028 and in some cases, even for 2029. And to Rob's point, the demand is continuing to strengthen. The forecast visibility in many cases, is widening, but really, what's changed in the last 3 to 6 months is you guys start learning the material.

And with extended lead times now with some products being well above 52 weeks, we really needed to solidify the demand outlook for next year. And so with those orders now in place, it gives us just that one extra level of confidence to be able to share the numbers that we did.

Your next question comes from the line of Ruplu Bhattacharya with Bank of America.

I have one question, but as you may have guessed, it's going to be in 2 parts. The first question is, Rob, you're guiding very strong growth for fiscal '27 much stronger than we had expected. What are some of the -- what are 2 or 3 of the key assumptions that are embedded in this outlook that investors should monitor over the next 6 months or so. I'm trying to get at what are some of the risk factors that investors should keep in mind. And the follow-up for Mandeep, because of all the political uncertainty, are you getting requests for more regional manufacturing? And do you see this as an opportunity to expand in North America versus in Southeast Asia and other places?

In terms of the risk factors, they always continue to be the same. It's largely at this stage of the game around continued material supply and continuity of supply. That being said, as Mandeep mentioned earlier, the lead times for components are quite long and our suppliers -- I mean our customers have placed non cancel, not refundable orders for long lead silicon, which gives us confidence that we've also have supply agreements on hard to find components such as memory and PWBs things along those lines. So we think we've effectively managed the risk on the downside.

In terms of the upside, just unprecedented demand increases across the board, specifically in AI ML compute and also in networking, we feel like we're gaining share across the board, driven by our ability to execute and execute well at scale, and that's also enabling us to provide solid 2027 guidance. Again, for 2027, we feel that it's continuing to unfold, and we're refining it so we're not placing a ceiling on the upside at this time, but the guidance that we gave, we felt very comfortable with this space.

And Ruplu, just to also address your question around the operational footprint, and I'll tie it back to dollars as well because I'm sure that this is top of mind as well. We still think the $1 billion of CapEx for this year is the right number around a little bit just based on the timing of payments. We think the $1.5 billion is a good placeholder still for next year. We'll give a number in October because if that number is to go up and it may, it will really be reflective of the demand that we're seeing in 2028. So then when you look at where that capacity is being deployed, it really is quite global in nature.

We are adding capacity in Thailand, which is one of our flagship locations supporting a number of the highest growth customers, just a fantastic execution track record in Thailand. But we're making expansions in Japan at the request of our customers as well and moving some switching business into Japan, but we're making significant investments in the interstate. In particular, in Richardson, Texas, we're expanding our existing footprint in Fort Worth, Texas, we are already bringing online some new capacity and more capacity will come online next year as well. And so we're finding that our customers right now are looking for an equal amount of growth in Asia as well as in the United States.

Your next question comes from the line of David Vogt with UBS.

I have more with thematic philosophical question for you guys. When you think about the demand that you're seeing that I think you called out, Rob, as unprecedented demand over the next several years. Can you help us understand -- I understand all the programs that are ramping 800, 1.6 CPO. But are the bands wider in terms of what you're hearing from your customers? And what I mean by that is -- are you seeing sort of a wider range of potential outcomes from your customers given the supply chain uncertainty? And outside of CapEx, maybe, Mandeep what kind of capital support or working capital support are your customers asking for at this part of the cycle given how strong demand is and potentially some limitations on capital at some of your potential customers going forward?

I would say earlier in the year, as the demand started picking up, our customers' demand signals are very, very dynamic. Just to use an analogy, I think all our customers through a bunch of spiriting on the wall, and they try to figure out what would stick or not. And then after the supply chain, the broad supply chain, try to understand its true capacity and where it could scale. The demand signals then normalized and strengthened around the theory of constraints, if you will. So right now, what we're really just seeing is a very constant and growing sort of demand signals from our customers. And hence, we've been working methodically with our suppliers to expand it. .

So I think the noise has calmed down quite a bit. Our customers, again, continue to want more than the supply chain could supply, but that is, again, not perishable demand. It's just giving us more visibility into the future when we're able to responsibly scale to achieve it.

And David, to your question around the engagement with our customers. Look, we're servicing the best customers in the world right now. We have very open and detailed conversations around balance sheet strength on both sides, ensuring that the commitments that we're making are supported by strong credit to be very kind of blunt about it. To be equally blunt, we're not a bank. So we don't go and fund our customers' ability to grow. And so we feel comfortable that when we're signing up for programs that we have the right contractual terms in place to protect both sides.

We do work very collaboratively with our customers long-term forecasting and increasing levels of visibility is where it really starts. In some cases, we have binding forecasts that go out 12 to 18 months in order to secure the materials. We do work with our customers on specialized equipment or nonrecurring expenses to be covered by them, and we'll cover the majority of the rest of it. But really, we take a dynamic approach by customer based on their individual situations, but we have no concerns across our customer base at this time.

Your next question comes from the line of Thanos Moschopoulos with BMO Capital Markets.

On MD Helios, there's obviously been a lot of programs announced. Could you give us some color in terms of at this point, and you're involved in what may represent potential opportunities and to what extent do you expect revenue to start kicking in later this year?

Sure, Thanos. So again, for the AMD Helios were the R&D design and manufacturing partner for the scale of networking switch. And this swissforms the interconnect backbone for the entire RAC. We are shipping samples towards the end of 2026, and the ramp begins in earnest in the first half of 2027. And again, this RAC has interest and demand and orders from several customers and the pipeline is growing. At this stage of the game, in 2027, we view this as a multibillion-dollar pipeline and growing.

And should we assume that you're involved in most of those programs? Or is it a day hyperscaler to determine the configuration of the RAC?

No, we're involved in all of these programs. .

Your next question comes from the line of Robert Young with Canaccord Genuity.

I just wanted to dig deeper on the CCS margin outlook. It looks like the non-HPS elements of that business are tracking at margins approaching HPS. I was curious what the driver is. And if you could also discuss the margin profile for $400 million, which is remaining at higher levels of demand than you expected and then what it looks like in 800 and 1.6 terabyte going forward. Is it a similar margin profile? Or is there something to understand there?

Rob, nice to talk to you. Look, the CCS business is performing very well from a margin perspective, record margin in the last quarter, and we think that there continues to be an opportunity to see expansion there. You're hitting on a couple of different points, but how I would maybe bring it up a level is are we growing in HPS or are we not and the vast majority of our switching programs are HPS, meaning a lot of our design is incorporated in there, a lot of our know-how. And that does allow us to have that reflected in price.

And so as we see an acceleration of networking growth next year, that will clearly be a benefit. We're still working through mix impacts. Non-HPS business is not as margin-rich as HPS product. So we're working through that piece right now. But then the other area that we have a very clear line of sight to is the operating leverage. We will continue to be very disciplined on our OpEx spend and we do believe that we're going to see another year of very good operating leverage coming through on the back of really double-digit productivity across the majority of our site and so CCS is performing very well.

Just lastly, I'll talk about ATS. ATS is really benefiting us as well. We're really happy to see that the business is returning to growth. We see that growth accelerating in the back end of this year. We see growth next year as well. But from a margin perspective as well, ATS just has its highest margins in its history at 6.3%. And we think that based on the mix of our portfolio, that there's an opportunity for expansion there as well as we go into next year. So really multiple contributors right now to the margin, sorry.

Your next question comes from the line of George Notter with Wolfe Research.

This is [indiscernible] on for George. Can you talk a little bit more about what you're seeing in the scale up landscape? Like are there any other scale-up program opportunities beyond Helios? What you guys are seeing over the next year?

Yes. Still, obviously, there is Helios, but there's also the OpenAI RAC that we're doing with respect to OpenAI timing, we're starting to deliver samples in 2026. And then that program will ramp in 2027 well into 2028 as well. We review this revenue opportunity as again, as a multibillion-dollar revenue opportunity from this. Regarding broader scale-up opportunities, broadly speaking, we have numerous engagements and discussions with folks to go into scale up. Scale for us really started with 1.6. And as we continue to win awards for integrated RAC systems, we expect this to be a huge opportunity for us.

Yes. And to add, from a customer logo perspective, to Rob's point, we're going to be seeing a lot of scale up demand with OpenAI as well as AMD, but we're also winning a number of networking programs in one of our largest customers. that is also in scale up. And so we find this to be really encouraging. The TAM is expected to grow in scale up between now and the next 3 years quite substantially. And that's tracking for us because we're seeing it come through in the winds that we have.

Your next question comes from the line of John Shao with TD Cowen.

I just wanted to ask about the turnkey versus consignment considerations especially for new programs in 2027. So any changes on how you recognize the revenue given some of your peers are choosing more consignment and any implications when it comes to OpenAI program given that it's more end to end?

John, so the answer is, maybe I'll start with OpenAI. I think it's fair to assume right now that, that will have the dynamics for program. And so therefore, the material costs will not be flowing through our revenue. As you know, the vast majority of our networking has the silicon included and the vast majority of our compute does not have the [indiscernible] included. And those dynamics are really holding as we go into next year. So no big change in the accounting side that's impacting our revenue.

Your next question comes from the line of [indiscernible]

I guess my question is on the fiscal year '26 outlook. You've raised both revenues and margin outlook for the year. When I look at the incremental margins on that $1.5 billion revenue, I got about 12% of operating margin of incremental. And I was wondering how should we think about the drivers behind that margin improvement? Is that your cost of component supply that is improving? Or are you able to capture more value out of customers? And as a follow-up, if you were to have more supply and low, how should we think about the incremental margins on this upside?

I'm just trying to answer that at a little bit of a higher level, which is we are seeing margin expansion quarter-to-quarter and it's being driven by 2 main things. The first one is that CCS margins are accretive to the company and CCS is growing at a record rate this year. CCS is going to be growing close to 85% year-over-year. And then we're going to see a similar level of growth next year. just naturally as CCS becomes a bigger part of the pie margins do have that benefit. Within CCS, it's the 2 things that I had talked about.

One is we're seeing an increasing level of HPS or ODM content that allows a strong pricing because of the value add that we're bringing. But then we're seeing operating leverage. Our factories right now are running at a high level of utilization and it is not by accident. It's intentional because we know that you can drive very strong productivity in that type of environment. And so we are seeing the benefits of operating leverage to your point on if revenue was to continue to accelerate. While we don't expect to add an SG&A dollar for every revenue dollar. So I would say the operating leverage flow-through should be there. The rest of it would really be mix-dependent.

Your next question comes from the line of Paul Treiber with RBC Capital Markets.

You mentioned you're gaining market share across Board. Can you elaborate further on market share momentum within both enterprise and communications. And then specifically with programs becoming larger and larger, are you seeing the competitive intensity of those programs increasing and how you're working to retain share going forward?

Sure, Paul. So broadly speaking, our customers are valuing continuity and surety of supply. One of the things that we do very, very well is that we are able to deliver high complex products to our customers reliably at scale. So in many cases, where products were dual source, and that's very common in our industry as products get awarded initially. We're usually the ramp partner. And then as they get mature, sometimes they get offloaded a second sauce for us to ramp the next generation.

In many cases, what we're seeing is our customers are asking us to pick up, take back some share that was potentially order to other providers because we can more reliably produce those products at scale, if you will. So we're seeing that across the board. And while we're able to rise to the occasion is that material is already pipeline. So we don't have to get back in line. We're actually able to pick up that supply chain and just put it through our proven manufacturing processes. And we're seeing that certainly in enterprise and also in communications.

One thing I would add to that, Paul, is one of the things that we're hearing from our customers is the amount of value that is placed on a transition from designing a program to ramping a program. It is not easy. One of the differentiators we think that Celestica has is we have a very sizable design engineering organization, which we've been vesting in for over a decade. That organization will be close to 2,000 engineers by the end of this year. And so being able to design the product in collaboration with our customer and then being able to work very collaboratively with the manufacturing side of the business to ramp it successfully.

We believe at Rob's point has led to some market share gains. Some people do design on people do manufacturing, we do both and that's part of our secret sauce. Thanks for the question, Paul.

There are no further questions at this time. I will now turn the call back to Rob Mionis, CEO, for closing remarks.

Thank you. Our strong second quarter performance and momentum carrying us into the second half of the year gives us confidence to raise our full year 2026 outlook and continued exceptionally strong cost demand and strategic investments in our infrastructure are enabling us to accelerate our 2027 revenue growth past the 2026 65% mark. Again, thank you for your support, and we are looking forward to update you next quarter.

This concludes today's call. Thank you for attending. You may now disconnect.

Thank you for joining Celestica's Annual Meeting of Shareholders. I will now hand the meeting over to Rob Mionis, Chair of the meeting.

Good morning. I am Rob Mionis, Chair of the Board and Chief Executive Officer of Celestica, and I will act as Chair of this Annual Meeting of Shareholders. I am joined by Mandeep Chawla, Chief Financial Officer; Doug Parker, Chief Legal Officer and Corporate Secretary; and Pam White, Senior Director of Communications.

All of the director nominees have joined the meeting as well. Before we proceed with the business of the meeting, I would like to welcome everyone joining us today. 2025 was an outstanding year of disciplined execution and exceptional growth for Celestica. I would also like to take a moment to recognize two individuals. First, I want to formally recognize retiring Chair, Mike Wilson for his exceptional leadership over many years. On behalf of the entire Board, I want to thank Mike for his unwavering commitment to Celestica and its shareholders.

We also saw the departure of Luis Muller from the Board this past year, and I would like to also thank Luis for his many years and contributions to the Board. As was the case last year, this meeting is being held as a hybrid meeting. Our goal is to replicate the experience shareholders attending virtually would have if they were all meeting in person.

As in past years, we expect the vast majority of votes will have been cast in advance of the meeting by proxy. That said, registered shareholders and duly appointed proxy holders will be allowed to vote in person or online in accordance with the instructions to be provided. Given the hybrid format of the meeting and in order for us to expediently undertake discussion on any matter proposed for a vote, we will pause at certain points during the meeting to provide an opportunity to vote or ask questions.

However, [ we would not ] cover registered votes or duly appointed proxy holders participating online, who have specific questions on a formal item of business to submit each question now by clicking on the messaging icon and then clearly identifying the applicable item of formal business as well as your name and contact information.

Such questions will be addressed prior to voting on the applicable motions. Following the business of the meeting, I will provide a brief update on the company and together with Mandeep, we will be available to respond to any questions. If you have a question about the company, but not specifically related to a formal item of business to be voted upon at the meeting, please wait until the general Q&A session following the conclusion of the meeting -- or if you are participating online, you may submit the question along with your contact information at any time and will be considered for the Q&A session.

If we are unable to address your question during the Q&A session, our team will follow up with you after the meeting. At this meeting and during management's presentation, we will make statements containing forward-looking information and refer to certain non-GAAP financial measures. I bring to your attention the cautionary note regarding such forward-looking statements, which for those who are attending in person has been printed and made available to you and for those participating online is available by clicking on the highlighted documents drop-down icon at the top right of your screen and on our website.

I now call the meeting to order. With the consent of the meeting, Doug Parker will act as Secretary of this meeting and Roxanne Persaud of Computershare Investor Services, Inc., will act as scrutineer, also known as the Inspector of Elections for this meeting. The Secretary will now report on certain procedural matters.

Mr. Chair, I can report that the notice of this meeting, proxy statement, proxy card and the consolidated financial statements of the company for the financial year ended December 31, 2025, as well as the independent auditor's report thereon have been mailed or made available to the requisite recipients in compliance with applicable requirements. The scrutineer has provided a preliminary report on the attendance, and I can confirm that the requisite quorum of shareholders is present in person or represented by proxy at this meeting. Accordingly, the meeting is properly constituted for the conduct business.

Thank you, Doug. The voting at today's meeting will be conducted by ballot. If you have already submitted a proxy, it is not necessary to also vote again since your vote will be recorded in accordance with the proxy instructions. If you are a registered holder or a proxy holder attending here in person and you have not already voted or if you are a registered holder and would like to change your vote and you have not already received a ballot, please put up your hand when requested to do so and the scrutineers will provide you with a ballot.

The ballot should be completed by marking an x in the appropriate spaces and must be clearly signed. If you are a registered shareholder, please put your name on the ballot. When you have completed and signed the ballot, please so indicate the scrutineers will come and collect it. If you are a registered holder or a duly appointed proxyholder participating online, and you have not already voted by proxy or you would like to change your vote, you can vote when prompted.

For those participating virtually, the online polls will be open for all items of business to be voted on at the same time. This will allow you to vote on each item immediately or if you prefer, you may wait until the conclusion of the discussion on all items prior to casting your vote. Once the online polls have been opened, the items of business to be voted on and your available voting options will be visible on your voting panel, accessible at the top of your screen.

To submit a vote, please click on one of the voting choices displayed on your screen. Once the discussion has concluded on all items of business, I will allow for a moment and then declare the voting closed on all matters of business. I now declare the online polls open for all items of business.

The first item on the agenda is the election of directors. I now ask the Secretary to propose the nominees.

I hereby nominate each of the following 9 persons to serve as a director of the company until the close of the next annual meeting of the company's shareholders or until the director's successor is duly appointed, subject to the provisions of the company's bylaws. Kulvinder Kelly Ahuja, Rob Cascella, Chris Colpitts, Francoise Colpron, Jill Kale, [ Laurette Konar ], Amar Maletira, Rob Mionis and David Reeder. These nominees have accepted their nomination.

As the company did not previously receive timely notice of any further nominations of persons for election as directors of the company, as required by the company's advanced notice bylaw, I declare the nominations closed. Is there any discussion regarding the election of the nominees from those here in attendance in person?

Thank you. Let us now also pause to account for any delay in the broadcasting of the online meeting to allow for questions to be submitted by shareholders participating virtually.

Mr. Chair, I confirm that we have not received any questions from shareholders participating online specifically on this item.

Whether you are participating online or in person, please follow the same voting instructions I provided at the start of the meeting. The next item of business is the appointment of the independent auditor of the company and the authorization of the Board of Directors to fix the independent auditor's remuneration. May I have a motion?

I move that KPMG LLP be appointed the independent auditor of the company until the close of the next annual meeting of the company's shareholders and to authorize the Board to fix the remuneration to be paid to the company's independent auditor.

Thank you. Is there any discussion of the matter from those here in attendance in person. Thank you. Let us now also pause to account for any delay in the broadcasting of the online meeting to allow for questions to be submitted by shareholders participating virtually.

Mr. Chair, I confirm that we have not received any questions from shareholders participating online specifically on this item.

Whether you are participating online or in person, please follow the same voting instructions I provided at the start of the meeting. The next item of business is the advisory resolution on the company's approach to name executive officer compensation or as it is commonly referred to, say-on-pay vote. Shareholders are being asked to approve on an advisory basis and not to diminish the role and responsibilities of the Board of Directors -- the compensation paid to named executive officers as disclosed in the company's proxy statement delivered in advance of the meeting.

I hereby move that the say-on-pay resolution appearing on Page 39 of the company's proxy statement be approved.

Thank you. Is there any discussion of the matter from those here in attendance? Thank you. Let us now also pause to account for any delay in the broadcasting of the online meeting to allow for questions to be submitted by shareholders participating virtually.

Mr. Chair, I confirm that we have not received any questions from shareholders participating online specifically on this item.

Whether you are participating online or in person, please follow the same voting instructions I provided at the start of the meeting. We will now proceed to the next item of business being the presentation of the company's consolidated financial statements and the independent auditor's report thereon. These were included in the annual report that was mailed or made available to each shareholder who elected to receive the report.

Copies of the financial statements are also available on our website and under our profile on SEDAR and EDGAR. We will now proceed with the process for completing the online voting on all items of the business of the meeting.

Pam, just to double check. Given the delay in the broadcasting of the online meeting have any further questions come in from shareholders participating online, specifically on any of the motions.

Mr. Chair, I confirm that we have not received any questions from shareholders getting online specifically on the motions.

Thank you, Pam. For those of you participating through the virtual meeting platform, who have not voted on all of the items of business, please do so now. We will now pause to allow for your time to vote.

[Voting]

That concludes the voting at today's meeting. I would also ask that the scrutineer compile a preliminary report regarding the results of voting on all business matters. May I have the scrutineers preliminary report on the votes conducted by ballot at this meeting.

Mr. Chair, the scrutineer has now reported to me that based on solely on proxies submitted prior to the commencement of this meeting, all matters put to a ballot at this meeting have been passed with the requisite shareholder support.

Thank you, Doug. I declare that the individuals nominated are elected as directors. The scrutineers' report also shows that based solely on proxies submitted prior to the commitments of this meeting each elected director received votes in excess of the thresholds established under Celestica's majority voting policy, as described in the proxy statement.

I also hereby declare that KPMG LLP has been appointed the independent auditor of the company and that the directors have authorized to fix the independent auditor's remuneration. I hereby declare that the advisory resolution approving named executive compensation to have passed.

As there is no other business that may properly come before this meeting, I declare the meeting terminated. Before we close, I'd like to take a few minutes to share some brief perspectives on our business. As we recently shared in our first quarter results, we've had a very strong start to the year and remain confident in our trajectory.

We've built a track record of consistent outperformance while strengthening our position the trusted technology partner to many of the world's most influential companies. Our ambition is clear, to lead and accelerate market advancements. And we deliver this through innovation, technology leadership and focused execution across our segments.

In our Connectivity & Cloud Solutions, or CCS segment, we are capturing significant growth as we support hyperscalers building next generation of data center infrastructure. Our technology leadership is driven by deep engineering expertise, spanning hardware and software focused on solving complex system-level challenges.

This enables us to design and develop leading-edge platform solutions across our technology stack. We are a recognized market leader in 400G and 800G Ethernet switches, and we are delivering the next generation of 1.6T switching solutions.

This leadership and our ability to execute complex next-generation networking designs at scale is reflected in the significant new program wins we've announced in previous quarters. Our Advanced Technology Solutions, our ATS segment, is also performing well. Here, we are focused on our engineering-led strategy. By investing in specialized market-focused teams and global design centers, we have become a trusted technology partner capable of solving our customers' most complex challenges.

Strong progress reflects the dedication of our global team. Our culture is grounded an assured commitment to excellence, and it is the strength of our people that continues to drive our performance. Thank you to all of our employees for your commitment and contributions. And to our shareholders, thank you for your continued trust and confidence in our team and our vision.

The opportunity is ahead of us is significant. We are helping set the pace of innovation across our markets, and we are just getting started. I would now be happy to take any questions.

There are no questions. This concludes today's meeting.

Hello, everyone. Thank you for joining us, and welcome to the Celestica Q1 2026 Financial Results and Conference Call. [Operator Instructions] I will now hand the conference over to Matthew Pallotta, Head of Investor Relations. Matthew, please go ahead.

Good morning, and thank you for joining us on Celestica's Q1 2026 Financial Results Conference Call. On the call today, we have Rob Mionis, President and Chief Executive Officer; and Mandeep Chawla, Chief Financial Officer.

Please note that during the course of this call, we will make forward-looking statements, including statements relating to the future performance of Celestica, our business outlook, guidance for the second quarter of 2026, our 2026 annual outlook and anticipated trends in our industry and their anticipated impact on our business. These are based on management's current expectations, forecasts and assumptions as of April 27. These forward-looking statements are subject to risks and uncertainties that could cause actual results to differ materially from our expectations and their potential impact on our results cannot be reliably predicted at this time. For identification and discussion of the material assumptions, risks and uncertainties, please refer to our public filings with the SEC and on SEDAR+ as well as the investor relation section on our website. We undertake no obligation to update these forward looking statements unless expressly required to do so by law.

In addition, during this call, we will refer to various non-GAAP financial measures. We have included in our earnings release found in the Investor Relations section of our website, a discussion of those non-GAAP financial measures and a reconciliation to the most comparable GAAP measures.

Unless otherwise specified, all references to dollars on this call are to U.S. dollars. All per share information is based on diluted shares outstanding and all references to comparative figures are a year-over-year comparison. Let me now turn the call over to Rob.

Thank you, Matt, and good morning, everyone, and thank you for joining us on today's call. We kicked off the year with solid results in the first quarter as revenue surpassed $4 billion with adjusted operating margin of 8%, a new high for Celestica. This performance drove adjusted EPS of $2.16 for the quarter, which exceeded the high end of our guidance range.

Our awarded backlog and the opportunity pipeline with both existing and new customers are the strongest they have ever been during my tenure as CEO. We continue to see exceptionally strong and accelerating demand from our hyperscaler customer base, complemented by a steadily strengthening outlook in our ATS segment. This momentum underpins our expectation for sustained growth in both revenue and adjusted EPS throughout 2026, while our outlook for 2027 has strengthened compared to just 90 days ago.

I'll discuss our latest 2026 outlook in a moment, but first, I'll hand it over to Mandeep to walk through the Q1 details and our Q2 guidance. Mandeep, over to you.

Thank you, Rob, and good morning, everyone. Revenue in the first quarter was $4.05 billion, up 53%, just above the midpoint of our guidance range, driven by very strong demand in our CCS segment. Our non-GAAP operating margin was 8.0%, up 90 basis points, driven by solid margin improvement in both of our segments. Our adjusted earnings per share was $2.16 in the first quarter, exceeding the high end of our guidance range and an increase of $0.96 or 80%.

Moving on to some additional metrics. Adjusted gross margin was 11.3%, up 30 basis points driven by improved mix and strong productivity. Our adjusted effective tax rate for the quarter was 19%. And finally, strong profitability and disciplined working capital management led to an adjusted ROIC of approximately 50%, up more than 18 percentage points versus the prior year.

Moving on to our segment performance. Revenue in our ATS segment for the quarter was $806 million, flat year-over-year and higher than our guidance of a low single-digit percentage decline. The performance was driven by higher revenue in HealthTech offset by tougher comps due to previously communicated portfolio reshaping in our A&D business and softness in capital equipment. Our ATS segment accounted for 20% of total company revenue in the first quarter.

Revenue in our CCS segment was $3.24 billion, up 76%, driven by very strong growth in both our communications and enterprise end markets. The CCS segment accounted for 80% of total company revenue in the first quarter. Revenue in our communications end market increased by 69%, better than our outlook of low 60s percentage growth primarily driven by strong demand and ramping programs for our 800G networking switches across our largest hyperscaler customers.

Our enterprise end market revenue was higher by 101% driven by the planned ramping of a next-generation AI/ML compute program with the hyperscaler customer. This result was modestly lower than our outlook of 100 high-teens percentage increase as the timing of the planned ramp was partially gated by select component constraints.

Our HPS business generated revenue of $1.7 billion, representing growth of 63% and accounted for 42% of total company revenue. The growth was driven by ramping 800G switch programs with multiple hyperscaler customers.

Moving on to segment margins. ATS segment margin was 6.0%, up 100 basis points, driven by improved mix and higher profitability as a result of our portfolio optimization activities. CCS segment margin in the first quarter was 8.6%, an improvement of 60 basis points, driven by strong mix and operating leverage from higher volumes.

During the first quarter, we had three customers that each accounted for at least 10% of total revenue, representing 35%, 15% and 15% of revenue, respectively.

Moving on to working capital. At the end of the first quarter, our inventory balance was $2.67 billion, a sequential increase of $485 million and higher by $885 million compared to the prior year as we support significant growth in our CCS segment. Cash cycle days during the first quarter were 55, representing an improvement of 14 days versus the prior year and 6 days better sequentially.

Turning to cash flows. We generated $138 million of free cash flow in the first quarter. Our capital expenditures were $230 million or 5.7% of revenue, an increase of $135 million sequentially and $193 million versus the prior year. Consistent with our prior communication, our full year 2026 capital expenditure guidance remains unchanged at approximately $1 billion to enable significant growth in our CCS segment. This investment is supported by awarded programs and our increased level of visibility to a multiyear capacity alignment with our key customers.

At the end of the quarter, our cash balance was $378 million, while our gross debt was $719 million, resulting in a net debt position of $341 million. We had no draw outstanding on our revolver at the end of the quarter.

Our gross debt to non-GAAP trailing 12-month adjusted EBITDA leverage ratio was 0.6 turns, an improvement of 0.1 turn sequentially and 0.5 turns versus the prior year period. As of March 31, we were in compliance with all financial covenants under our credit agreement.

Subsequent to the end of the quarter, we amended our credit facility, increasing our revolver by $1 billion to $1.75 billion. In addition, we received more favorable terms regarding certain covenants and interest rates and the maturities of both the Term Loan A and revolver were extended to 2031. The upsize revolver on the amended facility, along with our cash balance, provides us with more than $2 billion of available liquidity and which we believe is sufficient to meet our current operating needs.

During the quarter, we repurchased approximately 73,000 shares for cancellation under our normal course issuer bid at a cost of $20 million. We continue to be opportunistic with respect to share repurchases.

Now moving on to our guidance for the second quarter of 2026. Second quarter revenue is projected to be between $4.15 billion and $4.45 billion, representing growth of 49% at the midpoint. Adjusted earnings per share are anticipated to be between $2.14 and $2.34, representing an increase of $0.85 at the midpoint or 61% growth compared to the prior year. Assuming the achievement of the midpoint of our revenue and adjusted EPS guidance ranges, our non-GAAP operating margin is expected to be 8.0%, which would represent an increase of 60 basis points. We anticipate our adjusted effective tax rate for the second quarter to be approximately 21%.

Finally, let's review our revenue outlook for each of our end markets. In our ATS segment, we anticipate revenue to be up in the mid-single-digit percentage range, fueled by program ramps across our HealthTech and industrial businesses alongside strengthening market demand driving a return to growth in our capital equipment business.

In our CCS segment, we expect revenue in our communications end market to grow by approximately 50%, driven by ongoing hyperscale ramps in multiple 800G programs, complemented by continued strength in 400G programs. In our enterprise end market, we expect growth of approximately 130%, supported by the continued ramp in an AI ML compute program with a hyperscaler customer as well as ramping volumes in storage.

With that, I will now turn the call back over to Rob to provide an update on our 2026 annual financial outlook and additional color on the latest developments in our business.

Thank you, Mandeep. Driven by the continued strengthening of our demand pipeline and enhanced visibility as we progress through the year, we are raising our full year 2026 annual financial outlook. We are raising our revenue outlook from $17 billion to $19 billion, representing very strong growth of 53%. This latest outlook reflects accelerating demand in the second half of 2026, fueled by production ramps for awarded programs.

We are also raising our outlook for adjusted EPS and from $8.75 to $10.15, which, if achieved, would represent growth of 68%. Included in our forecast is an adjusted operating margin outlook of 8.1% higher than our previous outlook of 7.8%. Finally, we are reaffirming our free cash flow outlook of $500 million which fully incorporates our $1 billion in planned capital investments in 2026. This outlook represents our high confidence view for 2026 and while the supply environment remains highly dynamic, our outlook is informed by a measured assessment of component availability.

Building on my earlier remarks, our longer-term customer demand outlook has continued to solidify over the past 90 days, bolstered by additional new program wins and enhanced forecast visibility. We expect to grow revenue by over $6.5 billion in 2026 based on our latest outlook. And now we expect to grow revenue significantly more than this in 2027. As the year progresses, and our visibility continues to improve, we will continue to update our outlook.

Now moving on to our businesses and beginning with our CCS segment. Based on our latest 2026 annual outlook, we now anticipate approximately 70% revenue growth in our CCS segment. Even as demand across our customer base remains exceptionally strong and continues to accelerate, the unprecedented growth throughout the data center ecosystem has created a tighter supplier environment, effectively pacing our growth. As we navigate extended lead times and supply chain constraints for certain advanced components, we view our overall demand as durable and cumulative, underpinning our sustained growth runway as supply and capacity aligned through the second half of 2026 and into 2027.

Moving on to our end markets. In communications, accelerating volumes from the ramping of multiple 800G Ethernet switch programs are driving very strong and sustained growth. In addition, we are expecting to begin mass production on 1.6T switch programs with two hyperscaler customers, which will contribute to additional growth in the second half of the year.

During the quarter, we also secured two important new program wins, further bolstering our networking demand pipeline into 2027 and 2028. First, we announced in March, we are collaborating with AMD on the design and manufacturing of a scale-up networking switch for the Helios rack scale AI architecture. This collaboration leverages our leading-edge Ethernet networking expertise to support deployments of the Helios platform. Development is in progress, and we expect initial units to be available by year-end.

Additionally, we have secured a landmark program award for the design and manufacturing of a 1.6T, co-packaged optics Ethernet switch with a hyperscaler customer. This win serves as a critical validation of our ability to execute complex next-generation networking designs at scale. We expect mass production to commence in 2027. Within our enterprise end market, the growth outlook remains very strong into 2027. As anticipated, volumes for our next-generation AI/ML compute program with a hyperscaler customer continue to scale through 2026.

We continue to expect a strong momentum to sustain in 2027, supported by the launch of mass production for our digital native rack scale program as well as higher demand from our hyperscaler programs driven by ramps in next-generation platforms.

Moving on to our ATS segment. We are increasing our full year revenue outlook which now calls for mid- to high single-digit percentage growth. The strengthening growth outlook relative to prior quarters is bolstered by a reacceleration of customer demand in our capital equipment business, as the market forecast for wafer fab equipment spending is strengthening in the second half of 2026 and into 2027. We also continue to expect solid growth in both our Industrial and HealthTech businesses. supported by new program ramps.

We are also very encouraged by the strengthening margin profile of our ATS segment portfolio as we see the benefits of our strategic portfolio reshaping activities. We expect that these dynamics, along with improving operating leverage will continue to lead to stronger segment margins as we progress throughout the year.

The fundamentals and visibility supporting our long-term demand outlook through 2026 and into next year are stronger than ever. The intensifying need for leading-edge AI compute and networking infrastructure is driving an unprecedented level of customer demand. Today, as we scale our global operations to meet this accelerating demand, our primary focus remains on execution. We are confident that we are incredibly well positioned to execute and deliver on the growth opportunity in front of us.

With that, I will now turn the call back over to the operator to begin with the Q&A.

[Operator Instructions]

Your first question comes from the line of Samik Chatterjee with JPMorgan.

Maybe for the first question, you did mention the next-generation programs that you have with your enterprise customer on AI/ML compute programs and there's been a lot of discussion around share. So maybe if you can just dive in a bit into what you're seeing relative to your market share with your customer on that front, particularly on these next-generation program. And what level of visibility are they providing you, particularly as the demand sort of increases on that front? Are they looking for additional vendors or suppliers to help them with that ramp? And I have a follow-up.

Samik, yes, based on the long lead times that there is for silicon these days, the visibility that we have into next-generation programs is quite long. And we are supporting that customer on all the future generation programs, ones in production now and several are in the pipeline to start ramping in subsequent periods. We haven't seen a significant change in market share shifts or things along those lines, and we continue to execute well.

In the first quarter, by the way, it is not that you asked, we did have a little bit of a component issue relative to our AI/ML compute issue. It wasn't a demand issue. It was actually a material supply issue. That has been resolved and will be catching up in subsequent quarters and the program remains on track.

Got it. Got it. Interesting. And maybe for my follow-up, the announcement of the win around the CPO Ethernet switch. Just trying to get a bit more visibility in terms of what it means in terms of your content related to maybe a sort of a typical 1.6T switch, what does the content for Celestica look like? And what would be the margin profile? And what are you seeing in terms of engagement beyond this one hyperscaler that you won in terms of CPO ethernet switches?

Yes. Good question. So this is not just another switch award. We actually believe it's the first major production scale deployment of co-packaged optics using Broadcom's Tomahawk 6 Davisson module and designing that directly into our system. And winning this award validates our multiyear investment in developing this capability ahead of the market. We talked about that in the last call.

The transition to 1.6 requires managing unprecedented thermal and signal integrity challenges. And the fact that we're able to do this further demonstrate that we've actually moved up the value chain and we are now a sophisticated codesign partner for the most advanced hyperscalers.

It also sets us up well for the 3.2T adoption where we think this will kick in. So we think, at this stage of the game, we're a market leader in terms of margin profile, because of the high value add, this will be on the higher end of what we typically do.

And are you seeing more hyperscalers interested beyond this one hyperscaler that you won?

I think each of the hyperscalers will have their own adoption for CPO. Some are choosing to do CPX before CPO. But at the end of the day, we'll think the major adoption will be at the 3.2T node.

[Operator Instructions] Your next question comes from the line of Michael Ng with Goldman Sachs.

You talked about the CCS revenue growth of 70% for 2026 and the 2027 outlook getting better relative to the last 90 days. I was wondering if you could talk a little bit about your 2027 CCS revenue growth expectations relative to where you guided us last time? And any way to size or help think about the order contributions from some of the key programs that you mentioned between scaled [ Helios ], CPO and digital native?

Yes. Mike, it's Mandeep here. Yes, thanks for the question. Look, we're very pleased with the accelerated growth that we're seeing across the business. And we have no indications right now that it's slowing down. When we're looking at [indiscernible] that we shared, we're growing by about $6.5 billion this year, and we think we're going to grow significantly more than that, which means the floor would be somewhere around $25.5 billion but we do see revenue higher than that. And that's on programs that we've won.

As you know, we're seeing really great networking dynamics this year. 400G continues to be strong. 800G is accelerating materially and we're ramping -- we're seeing some contribution towards the back end of the year from the 1.6T programs. When you start looking at 2027, 800G demand continues to be strong. 1.6T is now ramping across the programs that we've already started plus additional programs that will be coming online. And then we have the very large program, the rack scale system that we're doing for the digital native customer.

And then in addition to that, as Rob talked about, on the AI/ML compute programs, we're continuing to see strong growth, including next-generation programs. And I could go on with other things as well.

But with the programs that we've won with the material that we are in line to secure with the capacity that is coming online, we have strong confidence in 2027, and we'll give more specificity around the number as we go through the year.

Your next question comes from the line of Tim Long with Barclays.

Yes, I was hoping you could talk a little bit about the HPS business off to a pretty good start, $1.7 billion in the quarter. I'm curious if you can talk about kind of the progression of the HPS business. I know that's been part of the CapEx increased investment is more HPS centers as well. And particularly into next year, I think some of the -- I think all 3 of these large programs, the digital native, AMD and CPO should be HPS as well. So if you could just give us a little color on how you see the glide path for that piece of business in the mix.

Great. Tim. Look, the HPS business and specifically, the design work that we're doing with our customers is really underpinning the majority of the growth that we're seeing or a significant portion of it. You've already mentioned a few of them. We're seeing very good traction on the switching side. The CPO program that Rob just talked about is an HPS program. The 1.6T programs are predominantly HPS as well as our 800G and 400G programs.

And so we have a very strong competitive position when it comes to networking. As we grow this digital native program into next year, it really starts to extend ourselves into compute as well. And all of this is underpinned by investments that we're making. You'll notice that our R&D spend is up materially, and it's where we want to be. We have about 1,350 design engineers right now. That's much higher than we had this time last year. And by the time we end this year, we're going to have even more. And the benefit of that is that they're working on not this year's programs, but they're really working on programs for next year and the year after. And so because of the programs that we've been winning in the pipeline that we have, we're very comfortable making the investments that we are in R&D, which is specifically HPS.

Your next question comes from the line of Karl Ackerman with BNP Paribas.

Two questions for me please. First, as you think about the drivers for upwardly-revised outlook for 2026 and 2027, I was hoping you could bucket the relative growth drivers between some of the new switch programs versus the AI servers. And how that might influence your margin trajectory throughout the balance of '26 and '27.

Karl, I'll get started and Rob can feel free to add on for anything that I may miss. Look, as we look into 2027, what we are seeing right now is very strong growth coming in particular from CCS, although ATS will also be growing. I'll just quickly touch on ATS because often we miss it on the call. It's very nice to see performance that's underway right now in ATS, which is a return to growth and strong margin, and that's really being underpinned by capital equipment.

Capital equipment, we believe now is entering into a very nice cycle where there is a very strong order book from our major customers, and that will take us through this year and next year. So ATS will be contributing to the growth. But the growth is overwhelmingly may be driven by CCS, and it is being driven by both communications and enterprise. We're not going to just put up the details yet. Again, more to come. But just to talk about the specifics a little bit more.

On the enterprise side, we are growing a new storage program that we've just launched and it did ramp. We have the AI/ML compute program that is going strong through this year. And then the next-generation program, 2 programs to be specific, are going to be ramping in '27. And then on the networking side, as we talked about, we have a couple of 1.6T programs that are coming online towards the end of this year, and then we have a lot more that are coming online next year. And then we never want to lose sight of the digital native win, which is a completely integrated rack system, which has compute and networking in it, highly complex, not easy to do at scale, and that will be entering production next year as well. And so those are really driving the growth across CCS.

And I would just add, Karl, that in terms of 1.6T we have 10 active programs. They'll be ramping heavily in 2027. On top of that, 800G remains strong. And as Mandeep mentioned, we have the digital native. We have the Helios Rack, so lots of new programs and also our AI/ML compute program that has very strong end customer demand for those programs that will end up ramping very nicely into 2027 as well. So overall, a very strong demand environment in 2027.

[Operator Instructions] Your next question comes from the line of Mehdi Hosseini with Susquehanna Financial Group.

I want to follow up on the Helios project. To what extent -- actually, can you size the opportunities associated with Helios over the next 18 months? And to what extent some of those opportunities are now embedded into your calendar year '26 revenue target?

Yes. With respect to Helios, the program is in development right now, and we'll be shipping samples this year. We view the overall market as a multibillion-dollar market. And as we get into next year, I think revenue will probably be based more by silicon availability than by end market demand. At this stage of the game, things are on track with the development and it has a lot of market interest.

Your next question comes from the line of David Vogt with UBS. Please go ahead.

So just maybe, Mandeep, I have a question regarding the revenue ramp versus the trade-off on gross margin. So obviously, TPU sounds like it was capacity constrained this quarter, and that's going to ramp stronger as we go into Q2 and the second half of this year, and it certainly sounds like strength in calendar '27. How should we think about the gross margin progression of the business, particularly with TPU ramping and also the -- what sounds like a really strong start to the T&C relationship potentially in the second half of this year into next year?

Yes, David. Look, we're pleased that we saw gross margin expansion in the quarter on a year-over-year basis. We think that the gross margin dynamics that we're seeing are going to play out in a similar way. There are going to be mix impacts along the way. But we definitely are looking to maintain, if not grow our gross margins as we go forward. Two things I'll talk about.

The first one is that longer term, there are some headwinds on the gross margin side, and it really has nothing to do with pricing because we're not -- we have capacities at a premium. Our execution is a real differentiator. And so we have choice on where we apply our focus. And so we're not giving up price in the marketplace. But the reality is that there's some input costs that are going up materially, whether it be memory or whether it be silicon. And so those are some headwinds that we're working through on the gross margin side.

But more specifically, what I'll talk about is the operating profit. We're seeing very nice operating leverage right now in the company, and we think that there's more opportunity in front of us as well. We're pleased to be able to raise the full year to 8.1%. That's up from the 7.8% that we had just 90 days ago. It's a reflection of mix, but it's a reflection of operating leverage because we'll be very disciplined on the operating expenditure side. And although we are very pleased to make investments along the way to fund growth. And so as you look into 2027, there will continue to be an opportunity on, I would say, operating leverage, and we are very focused on ultimately translating that into EPS growth. We are a management team that is very focused on long-term sustainable growth in EPS. And so the levers that are being pulled are driving that outcome.

Your next question comes from the line of George Notter with Wolfe Research. Your line is open.

I guess I wanted to come back to the CPO win that you announced here, is that an existing customer for networking infrastructure? Or is that a brand new customer? And then also, I'm wondering if it's an existing customer, is it likely that, that cannibalizes an existing revenue run rate for you? And then anything you could say about the potential size of that deal would be great. And then also, you said 2027 would be the ramp. Is it any more specifics there? Is it early '27? Is it late '27? Any help would be great.

George, yes, it's an existing customer. We have multiple 1.6T awards with this customer and programs. That's why they felt comfortable in giving us this one as well. We don't feel it cannibalizes the current programs, that current demand that we have on those programs is actually going up. And with respect to timing, this will be second half of 2027.

Your next question comes from the line of John Shao with TD Cowen.

So regarding Google's new Boardfly architecture for TPU v8, could you maybe talk about implications to your business? And do you see an ongoing trend towards more complex data center interconnect -- and as a result, you're going to be there to benefit as ODM?

Yes. As the architecture becomes more critical and as our hyperscale customers continue to innovate, what's becoming more and more evident is that systems level manufacturing design is becoming more and more important for these customers, especially in liquid cooling, advanced rack scale infrastructure, thermal management. So these customers are looking for us to continue to innovate with them and design with them on multiple nodes moving forward.

So we're pleased to be able to continue to support them as they continue to innovate and advance their architecture through all the versions that you mentioned.

Your next question comes from the line of Ruben Roy with Stifel.

Rob, I wanted to come back to the supply commentary. You characterized the environment as highly dynamic, and it sounded like there was a specific issue around AI/ML in Q1. So I'm just wondering if you could maybe expand on how you're thinking about supply relative to the guidance and the demand dynamic as you think about both Q2 and the second half of the year. Are there caps on sort of what you're able to ship? Is it broader based than the AI/ML program that you talked about for Q1? Any detail on that would be helpful.

Good question. We are experiencing more component shortages now than 90 days ago. 2 main factors. One is the demand really continues to grow. And as a result, the suppliers are a little bit behind on adding capacity. The good news is that all of our key suppliers are currently in the works of adding capacity, and we expect the situation to improve.

The constrained commodities right now on allocation are custom silicon and memory. That's probably not a newsflash for you. We are seeing challenges in PCBs, the 40-plus layer ones, power components, optical components. The positive news is that we have commitments from all our suppliers to secure the outlook that we just gave. So we think the outlook that we just gave factors all these in. We think it's prudent. We think it's conservative, and that's why we marked it as high confidence.

We have capacity in '26 coming online ourselves also in 2027. So if things continue to improve in the back half of the year, we'll have the opportunity to get it out the door and also into next year. But we think we factored it all in, but it is more constrained now than it was 90 days ago, and the lead times are extended. And one last positive thing is with the extended lead times, we are getting unprecedented visibility from our customers' end item demand. So that's actually a very positive thing for us as we do long-term planning.

Your next question comes from the line of Ruplu Bhattacharya with Bank of America.

Rob, Mandeep, you're projecting strong growth for next year. I think you said significantly more than $6.5 billion year-over-year in fiscal '27. How should we think about free cash flow in that context? Looks like you did not take up the free cash flow guide for fiscal '26. What are some of the puts and takes impacting free cash flow for fiscal '26 and '27, and how should we think about the working capital as you see this strong growth?

And just another clarification. I think, Rob, the AMD opportunity is a scale-up opportunity, whereas the one you announced in the press release that the switch -- CPO switch is a scale-out opportunity. Can you help us size how much of that $6.5 billion plus growth next year is from scale-up and scale-out and which is a longer-term opportunity?

Ruplu, I love you because you find a way to bind two questions in one. That is experience talking there. .

Funding, I'll let Rob talk about some of the program specifics that you brought up. So look, we're very happy that we were able to raise the revenue outlook for 2026 by $2 billion and yet maintain our free cash flow outlook of $500 million. And then just as a reminder, that's including funding $1 billion of CapEx.

The balance sheet is incredibly strong, but we are -- we believe we are very disciplined when it comes to capital allocation. We'll assess our needs as we go through -- as we get closer to 2027 and as the forecast starts to solidify. But I'll say just a few things. Again, the balance sheet is very, very strong. We have ample capacity to target to use it as we need to. We're very comfortable investing to support customer growth. But one thing you'll know is that we will always remain very disciplined. And so we are a free cash flow-focused management team. But at the same time, the growth is really unprecedented right now. And so if that requires additional funding, we have no hesitation to do that.

And regarding scale-up versus scale-out, correct. The CPO win is scale-out and the Helios Rack is a scale-up opportunity. We are -- have been very strong on scale-out, as you noted. So we view scale-up as a huge opportunity. But frankly, we're supporting both very strongly in our system-level architectures.

Your next question comes from the line of Robert Young with Canaccord Genuity.

Back to the component constraints. I think if you look back to the supply constraints of the pandemic, you especially took a lot of share because of its execution, its relationships. And so I'm curious if you'd see the current issues as strengthening your hand competitively. And then I was wondering if you could talk about relative impact on compute and networking or if it's relatively a similar supply chain impact.

Yes. One of our key strengths is execution. So once we get the parts and securing the parts, wherever it happens to be in the quarter, we're able to execute that at scale very quickly. So in the land of very dynamic supply constraints, we do advanced planning very well. And typically, we gain share through that environment because we find that peers or competitors have a hard time executing through turbulent times.

So that is an opportunity. We haven't necessarily seen that as of yet as a result of the supply chain constraints. But frankly, the constraints are just starting right in the thick of things right now. I think it will get better or at least more dynamic as the year gets along. And with respect to compute versus networking, on the networking side, it's not very memory-centric. So there, the constraints are really around PCBs or silicon, but we think we have a good handle on that. On compute, it's all about memory, but our customers have very strong presence with the main memory suppliers, and we have secured at least in the near term, what we think we need to do to support our customers.

Your next question comes from the line of Thanos Moschopoulos with BMO Capital Markets. Please go ahead.

[Operator Instructions] Your next question comes from the line of Paul Treiber with RBC Capital Markets. Please go ahead.

Just you mentioned unprecedented visibility, long-term visibility. And then also you mentioned in the slide deck, capacity alignment with customers. Can you elaborate on both of those? And specifically, what's fundamentally changed to give you far better long-term visibility than in the past? And is there anything contractual that gives you better visibility than you may have had in the past?

Yes. I would say a quarter or so ago, our key customers, the main hyperscalers had very, very strong demand. They loaded in the system just to try to figure out where the constraints would be. At this stage of the game, I think the entire supply chain knows where the constraints will be and we've aligned our physical capacity with our supply chain capacity with our CapEx plans. And that's the current state that we're in.

And because the lead times are very long, specifically custom silicon lead times are so long, we've not just done this for '26 or '27. We're actually dipping in it to 2028 and actually booking awards right now that ship into 2028. So the unprecedented visibility is really just making sure that we're aligning long-term material supply with long-term capacity agreements. And a lot of the orders that we're placing supported by our customers are NCNR. So there are contractual terms of protecting us through ups and downs. So we feel very comfortable in our long-term growth trajectory at this stage of the game. It's all about execution, and that's what we do very well.

Thank you for your question. Your next question comes from the line of Todd Coupland with CIBC.

Thanks, and good morning, everyone. So I wanted to ask about the switch ramps in the second half of the year. I'm wondering if they are coming in as expected or have some of them been pushed into 2027, which gives you the higher confidence in that year. Can you just talk about the dynamics between '26 and '27?

I would say the switch ramps that are coming in the back half of the year are going as planned. In the back half of the year, we have 2 customers ramping 1.6T and several more coming online in 2027. We've secured the silicon that is required to complete the development processes, and we also have commitments to support the ramps and 800G remains very strong throughout the year. So I would think things are going as planned on the networking side.

Thank you for your question. Your next question comes from the line of Michael Ng with Goldman Sachs.

I was just wondering if you could give us some updated thoughts on capital intensity and the outlook for CapEx beyond this year. Obviously, you've got a lot of new program awards and some new wins. Should we expect CapEx to ramp up beyond that $1 billion that you laid out for this year in 2027? How are you thinking about additional capacity that you need to put in to support these new wins?

Mike, it's Mandeep here. So $1 billion of CapEx this year, as you're aware, we will be doing that or more next year. And I think the way to think about it right now would be -- I'm just going to give you a rough number of $1.5 billion as a placeholder for now, and we're very comfortable with that.

The way that -- to build on the comments that Rob shared just recently, we're having very long-term conversations with our customers, a, as it relates to supply because we need to get in line on their behalf and lead times have extended, in some cases, beyond a year for certain types of materials, but then really around capacity. And the majority of our capacity investments right now are in Southeast Asia, Thailand specifically and in the United States, and I would say Texas specifically.

And we have a number that are going to come online this year. We have some that are coming online next year. And then there's other ones that we're evaluating still, but that's to support growth in 2028 and 2029. And so when we make our capacity decision, it's tied to a business case. It's tied to program level specifics. It's tied to programs that we are winning or we intend to win by the time we make that decision to go forward with the CapEx. And the business cases are strong, strong ROI, strong paybacks. And so when you -- when we continue to have business cases like that brought forward, we don't hesitate to invest. And so right now, I would say we will see an elevated level of CapEx spend into 2027 and be able to get 1 year at a time.

[Operator Instructions]

Your next question comes from the line of Thanos Moschopoulos with BMO Capital Markets.

Sorry about that earlier. With respect to your large digital native customer, is that still on track to ramp in early 2027? Or has there been any change in thinking on the timing for that?

Thanos. Yes, that's still on track. We're shipping sample systems this year and production should start in the first quarter -- late in the first quarter next year, that is on track. We've been having very close meetings with all the key suppliers there to make sure we've secured the material across the supply chain and things look very positive at this stage of the game.

Thank you for your question. There are no further questions at this time. I will now turn the call back to Rob Mionis for closing remarks.

Thank you all. Thank you for your support and for joining us this quarter. We have great momentum across our business, and we look forward to updating you on our progress next quarter. Have a great day.

This concludes today's call. Thank you for attending. You may now disconnect.

Ladies and gentlemen, thank you for joining us, and welcome to the Celestica Q4 2025 Financial Results and Conference Call. [Operator Instructions]

I will now hand the conference over to Matthew Pallotta, Head of Investor Relations. Please go ahead.

Good morning, and thank you for joining us on Celestica's Q4 2025 Financial Results Conference Call. On the call today, we have Rob Mionis, President and Chief Executive Officer; and Mandeep Chawla, Chief Financial Officer.

Please note that during the course of this call, we will make forward-looking statements, including statements relating to the future performance of Celestica, our business outlook, guidance for the first quarter of 2026, our 2026 annual outlook and anticipated trends in our industry and their anticipated impact on our business.

These are based on management's current expectations, forecasts and assumptions including that there are no material changes to tariffs or trade restrictions compared to what is in effect as of January 28. These forward-looking statements are subject to risks and uncertainties that could cause actual results to differ materially from our expectations and their potential impact on our results cannot be reliably predicted at this time.

For identification and discussion of the material assumptions, risks and uncertainties, please refer to our public filings with the SEC and on SEDAR+ as well as the Investor Relations section on our website. We undertake no obligation to update these forward-looking statements unless expressly required to do so by law.

In addition, during this call, we will refer to various non-GAAP financial measures. We have included in our earnings release, found in the Investor Relations section of our website, a discussion of those non-GAAP financial measures and a reconciliation to the most comparable GAAP measures. Unless otherwise specified, all references to dollars on this call are to U.S. dollars. All per share information is based on diluted shares outstanding and all references to comparative figures are a year-over-year comparison.

Let me now turn the call over to Rob.

Thank you, Matt, and good morning, everyone, and thank you for joining us on today's call. We delivered very strong results in the fourth quarter, driven primarily by growth in our CCS segment across both our communications and enterprise end markets. This led to revenue and adjusted EPS both exceeding the high end of our guidance ranges, while adjusted operating margin of 7.7%, once again marked the strongest performance in company history.

I'd like to briefly review our performance for this past fiscal year. Overall, 2025 was another exceptional year for the company. For the full year, we achieved revenue of $12.4 billion and adjusted EPS of $6.05 representing growth of 28% and 56% year-over-year, respectively.

Our adjusted operating margin of 7.5% marked the second consecutive year of 100 basis points improvement, driven by growth in AI-related demand for data center technologies, strong operational execution and improved operating leverage. We surpassed our annual outlook for each of our key financial metrics, further building on our positive momentum generated over the last several years. Looking back, our financial results reflect a consistent progression marked by a sustained annual improvement across revenue, adjusted operating margin and adjusted EPS.

As we look ahead, we anticipate the strong momentum to continue with revenue growth expected to accelerate in 2026. Furthermore, our optimism continues to strengthen regarding the significant pipeline of growth opportunities that lie ahead for our businesses, particularly in our CCS segment, which we believe will sustain this growth trajectory in 2027.

Before I provide an update on an annual outlook for each of our businesses, I would like to hand the call over to Mandeep to discuss our financial performance during the quarter and our guidance for the first quarter of 2026.

Mandeep, over to you.

Thank you, Rob, and good morning, everyone. In the fourth quarter, revenue of $3.65 billion was up 44% and above the high end of our guidance range, driven by a very strong demand in our CCS segment. Our non-GAAP operating margin was 7.7%, up 90 basis points driven by strong margin improvement in both of our segments. Our adjusted earnings per share was $1.89 in the fourth quarter, exceeding the high end of our guidance range and an increase of $0.78 or 70%.

Moving on to some additional metrics. Adjusted gross margin was 11.3%, up 30 basis points, driven by higher volumes and stronger productivity. Our adjusted effective tax rate for the quarter was 19% and lastly, as a result of strong profitability and disciplined working capital management, we achieved adjusted ROIC of 43%, up 14 percentage points versus the prior year.

Moving on to our segment performance. Revenue in our ATS segment for the quarter was $795 million, 1% lower and in line with our guidance of a low single-digit percentage decline. The decline in revenue was driven by lower volumes in our Capital Equipment business and previously communicated portfolio reshaping in our A&D business, partly offset by stronger demand in our other end markets.

Our ATS segment accounted for 22% of total company revenue in the fourth quarter. Revenue in our CCS segment was $2.86 billion, up 64%, driven by very solid growth in both our communications and enterprise end markets. The CCS segment accounted for 78% of total company revenue in the fourth quarter.

Revenue in our communications end market increased by 79%, above our guidance of a high 60s percentage growth, primarily driven by strong demand and ramping programs for 800G networking switches across our largest hyperscaler customers. Our enterprise end market revenue was higher by 33%, which was above our guidance of a low 20s percentage increase driven by the acceleration in the ramping of a next-generation AI/ML compute program with a large hyperscaler customer.

Our HPS business generated revenue of $1.4 billion in the fourth quarter, representing growth of 72% and accounted for 38% of total company revenue. The strong growth was driven by ramping volumes in 800G switch programs with multiple hyperscaler customers.

Moving on to segment margins. ATS segment margin in the quarter was 5.3%, up 70 basis points, primarily driven by improved profitability in our A&D business. CCS segment margin in the fourth quarter was 8.4%, an improvement of 50 basis points, driven by strong operating leverage. During the fourth quarter, we had 3 customers that each accounted for at least 10% of total revenue, representing 36%, 15% and 12% of revenue, respectively. For the full year 2025, we also had 3 customers that accounted for at least 10% of revenues at 32%, 14% and 12% of revenue, respectively.

Moving on to working capital. At the end of the fourth quarter, our inventory balance was $2.19 billion, a sequential increase of $141 million and higher by $427 million compared to the prior year, as we support continuing revenue growth in our CCS segment. Cash cycle days during the fourth quarter were 61, an improvement of 8 days versus the prior year and was 4 days better sequentially.

Turning to cash flows. In the fourth quarter, we generated $156 million of free cash flow, resulting in total annual adjusted free cash flow of $458 million in 2025, which was an increase of $152 million compared to the full year in 2024 and above our most recent annual outlook of $425 million.

Our capital expenditures for the fourth quarter were $95 million or 2.6% of revenue bringing our total capital expenditures in 2025 to $201 million or 1.6% of revenue. Since we last spoke at our Investor and Analyst Day in October, we have continued our discussions with key customers in our CCS segment in order to align on long-term capacity planning.

As a result of these discussions, we are meaningfully increasing the scale and scope of our capital investment plans in 2026 and 2027 in order to build out the revenue-enabling capacity required to support the strengthening demand we see ahead. We now anticipate that our capital expenditures for 2026 will be approximately $1 billion or 6% of our current annual revenue outlook. Importantly, we anticipate to be able to fully support this increase in capital expenditures through operating cash flow. The investments we are making in new capacity, which we expect will come online throughout 2026 and 2027 are a response to record bookings, accelerating growth in the scale of our existing engagements and meaningfully improved long-term demand visibility with our hyperscaler customers.

We view our investments in new capacity as highly strategic aligning our global footprint with a multiyear capacity road maps of our key customers in support of their large-scale investments in data center infrastructure and AI capabilities. These investments will include a combination of capacity additions at our larger sites, new customer-driven investments in the United States and upgrades to manufacturing capabilities, including investments in power.

We are undertaking significant new investments in Texas in support of growing customer demand for U.S. capabilities in the areas of R&D, manufacturing and advanced assembly. At both our Richardson campus and new site in Fort Worth, we are adding a total of over 700,000 square feet of footprint, with expanded power availability. This incremental capacity is expected to come online in 2027. Also, in order to facilitate greater engagement on R&D and design, we plan to establish a new HPS design center in Austin.

Our CapEx plans also include large-scale investments in our manufacturing capacity and capabilities across the rest of our global network. In Thailand, we continue to add new capacity to support very strong demand from multiple customers. We are adding over 1 million square feet in additional footprint with upgrades, including expanded power availability, advanced liquid cooling manufacturing and testing capabilities.

We expect this new capacity to come online towards the end of 2026 and into 2027. Elsewhere in our network, we are upgrading and retooling sites to add new manufacturing lines in locations such as Mexico and Japan in support of customer demand for greater geographic diversification. Allowing them the flexibility and optionality to derisk their global supply chains within our network. We are also excited to announce our plans to establish a new HPS design center in Taiwan.

Overall, we are very encouraged by the strong alignment and close collaboration on capacity planning we have with our customers, which underpins our confidence in making these investments.

Turning to our balance sheet and capital allocation. At the end of the quarter, our cash balance was $596 million. Our gross debt was $724 million resulting in a net debt position of $128 million. We had no draw outstanding on our revolver at the end of the quarter, leaving us with approximately $1.3 billion in available liquidity. Our gross debt to non-GAAP trailing 12-month adjusted EBITDA leverage ratio was 0.7 turns, an improvement of 0.1 turns sequentially and 0.3 turns versus the prior year period.

As of December 31, we were in compliance with all financial covenants under our credit agreement. During the fourth quarter, we received regulatory approval to launch our new normal course issuer bid, which permits us to, at our discretion, purchase up to approximately 5% of our public flow until November 2, 2026. We will continue to be opportunistic towards share repurchases as our approach remains unchanged.

During the quarter, we repurchased approximately 132,000 shares under our normal course issuer bid for $36 million. For 2025, our repurchases totaled 1.36 million shares at a cost of $151 million or an average cost of approximately $111 per share.

Now moving on to our guidance for the first quarter of 2026. First quarter revenue is projected to be between $3.85 billion and $4.15 billion, representing growth of 51% at the midpoint. Adjusted earnings per share are anticipated to be between $1.95 and $2.15 representing an increase of $0.85 at the midpoint or 71% growth compared to the prior year. Assuming the achievement of the midpoint of our revenue and adjusted EPS guidance ranges, our non-GAAP operating margin for the first quarter is expected to be 7.8%, representing an increase of 70 basis points. We expect our adjusted effective tax rate for the first quarter to be approximately 21%.

Finally, let's review our revenue outlook for each of our end markets. In our ATS segment, we anticipate revenue to be down in the low single-digit percentage range as growth in our HealthTech and industrial businesses are being offset by market-related softness in our Capital Equipment business and portfolio reshaping in our A&D business.

In our CCS segment, we anticipate revenue in our communications end market to grow in the low 60s percentage range, primarily driven by ongoing ramps in multiple 800G programs with our hyperscaler customers. In our enterprise end market, we expect a very strong growth in the 100 high-teens percentage range, supported by the progression of a next-generation AI/ML hyperscaler compute program.

With that, I will now turn the call back over to Rob for an update on our 2026 annual financial outlook and to provide additional color on the latest developments in our business.

Thank you, Mandeep. Given the strengthening demand forecast across our portfolio, we are raising our 2026 annual financial outlook. We are increasing our revenue outlook to $17 billion and raising our adjusted EPS outlook to $8.75, representing year-over-year growth of 37% and 45%, respectively.

This represents our high confidence view for 2026, which we will continue to refine and update as the year progresses. We are also maintaining our free cash flow outlook of $500 million. This demonstrates the inherent cash-generating power of our business, allowing us to organically fund a significant increase in capital investments while continuing to generate cash to fund other investment opportunities.

Since our Investor and Analyst Day this past October, the velocity and scale of awarded programs and growth opportunities for Celestica continues to expand. As Mandeep discussed, we have responded by significantly increasing our capital investment plans in order to grow our global footprint in alignment with our customers' multiyear requirements. These investments are intended to provide us with the necessary scale to support the accelerated growth we anticipate in 2026 and which we believe will be sustained in 2027.

In undertaking these investments, we have closely collaborated on demand planning with our largest customers, which has informed our decisions on the location, capabilities and scale of the new capacity we are developing. These investments are targeted to strategically support our customer base and their program-specific requirements over the long term.

On this note, we are proud of our decade-long partnership with Google and are excited to continue supporting the acceleration of leading AI data center architecture. Celestica remains closely aligned with Google on the development of complex data center hardware and systems. As a preferred manufacturing partner for Google's Tensor processing unit, or TPU systems, Celestica is committed to making long-term investments in both capacity and capabilities both in the United States and across our global footprint, which includes our planned investments to expand manufacturing capacity in 2026 and 2027.

These investments are designed to support the scaling of production for current and future generations of Google's custom silicon TPU systems as well as leading-edge networking technologies. Based on our latest outlook, we anticipate full year revenue growth of approximately 50% in our CCS segment, supported by strong demand and new program ramps across both end markets.

In communications, demand from hyperscalers is driving strong volumes for our 800G programs, while 400G remains highly resilient. We continue to expect mass production for our first 1.6T switching programs to begin ramping in the latter part of the year. Over the past 90 days, we have continued to add to our pipeline of newly won business in networking, adding to an already robust view of demand into 2027. We are pleased to announce that we have secured the design and manufacturing award for the 1.6T networking switch platform with a third hyperscaler customer. This HPS engagement is expected to ramp production beginning in 2027 with design work already underway. This new program award, along with strengthening demand forecast from our largest customers and a significant funnel of opportunities gives us confidence and optimism regarding the growth trajectory of our networking businesses. In our enterprise end market, demand signals remain solid.

As anticipated, we saw a meaningful ramp in our next-generation AI/ML compute program with a hyperscaler customer during the fourth quarter, and we continue to expect that volumes will accelerate into 2026. Looking towards 2027, we continue to anticipate strong demand from our hyperscaler and digital native customers, driven by ramps in next-gen AI/ML compute programs.

Now moving on to our ATS segment. We are maintaining our outlook for revenues to remain approximately flat to up in the mid-single-digit percentage range for the full year 2026, consistent with the targets we shared at our Investor and Analyst Day in October. We continue to expect growth in our Industrial and HealthTech business, supported primarily by the ramping of new programs. We anticipate this growth will be at least partially moderated by lower volumes in our Capital Equipment business in the near term.

As we progress through 2026, we anticipate overall ATS revenues to be higher in the second half of the year, led by a recovery in Capital Equipment volumes as broader market growth tailwinds come into effect. We also expect year-over-year growth to improve as we lap the impact from the strategic portfolio reshaping activities we undertook in A&D during the first half of 2025.

Overall, we expect 2026 to be another year of transformational progress in the growth and evolution of our business. We are experiencing an unprecedented level of demand supported by the sustained large-scale multiyear investments from our largest data center customers. We believe our company is uniquely positioned as a critical enabler of the AI/ML revolution, helping to solve the most difficult challenges in the data center from advanced liquid cooling solutions throughout the rack to the transition to next-generation networking platforms.

It's our ability to deliver these complex system-level solutions that allows us to win new mandates and solidify our leadership in the technologies of tomorrow. Today, our team is intently focused on our operational execution as we scale our global footprint to meet this growing demand.

With that, I will now turn the call back to the operator to begin the Q&A session.

[Operator Instructions] And your first question comes from the line of Ruplu Bhattacharya from Bank of America.

So it looks like you've taken up both the top line and the bottom line guide for fiscal '26. If we take the midpoint of the guidance literally, then there seems to be a slowdown coming in fiscal second half and also some loss of operating leverage. I mean the revenue guidance is 51% year-on-year for fiscal 1Q, but the full year is 37%, so implying some slower growth in the remaining 3 quarters. Likewise, in EPS it's 71% for the first quarter, but full year is 45%.

So EPS is definitely growing faster than revenue and there is leverage in the model, but it looks like some operating leverage decline in the remaining 3 quarters. So can you just clarify for us, is there something specific that's causing this slowdown? Or should investors just chalk this up to conservatism in the guide?

First of all, welcome back. We're always very happy to work with you. So thank you for the coverage. Yes, look, we're very confident on our 2026 outlook. And as we said in our commentary and Rob mentioned, it's our high confidence view. Our customer forecasts right now for 2026 are higher than the $17 billion that we are guiding and what's also really nice to see right now is that the demand outlook with our customers is actually extending beyond sometimes our typical fourth quarter outlook.

Similar to past outlooks that we've had, Ruplu, we're taking a pretty pragmatic view. Our views on next quarter and the quarter after that are typically going to be very much dialed in and we're going to share with you what that visibility exactly looks like. But when we look beyond the 2 quarters, we're just being pragmatic. We're focusing on securing supply. We have no concerns at this time, but we just want to make sure that the supply base can also ramp as fast as we are ramping, and then we take into account the macro uncertainties, which, as you know, there's a lot of them. But as we go through the year, we are working towards a higher number, and we'll look to be updating the numbers as we go.

Okay. If I can ask a quick follow-up. I want to ask about risk management. So you obviously have a lot of opportunity in both your white box switching business and the custom ASIC server business. One thing you mentioned is you're increasing CapEx to fund the growth. Can I ask if you're concerned about any potential funding for future AI-related projects? And is there any risk to programs materializing? And have you taken that into account?

And also, you've kept free cash flow at $500 million. Given that CapEx is going up and you're probably going to need more working capital to support revenue growth, can you just tell us like -- is there a risk to the story here? And what is giving you confidence to maintain the free cash flow guide? And again, congrats on the quarter.

I'll start off, I'll let Mandeep finish up. With respect to programs materializing, the build-out that we're doing is based on both businesses. We had a record bookings year in 2025 and we're really just building out to support those bookings. So there's very little risk in those programs materializing. They have been in the development cycle right now, and we're doing proof of concepts with respect to validation testing and they're well underway to ramping in 2026.

In terms of risks to the entire story, Mandeep talked about it. We view it more as uncontrollable, like geopolitical risks. There's always an opportunity of tightening supply chain. But frankly, our suppliers realize now that we have a lot of leverage these days given our scale. And we're also a design agent, which is giving us some leverage in the supply chain. We also have a lot of opportunities, as Mandeep mentioned. Demand continues to well outstrip our ability to provide it in the very short term. We have very strong demand from networking with respect to 400G, 800G and 1.6T ramps that are happening later on this year. And on top of this, we have some very strong demand for AI/ML compute. And within the enterprise market, we're also seeing signs of very significant growth. So overall, we see more opportunities than risk at this time.

I'll talk about cash generation. And look, we're very comfortable with our ability to invest. And frankly, we're willing to invest even more as we go through the year. That's what's in front of us. We think we'll generate at least $500 million of free cash flow this year. That's after paying for $1 billion of CapEx. I know that those on the call already are aware of this. We generated positive free cash flow every quarter for almost 7 years now.

And it's because we are very focused on generating strong positive free cash flow every quarter. And so with the growth plans that we have in front of us, we don't see that being a risk. And this is even going beyond the fact that we have an incredibly healthy balance sheet. And so we think that we can fund these with cash generation and not have to even use the balance sheet. Thanks for your questions.

Your next question comes from the line of Samik Chatterjee with JPMorgan.

Maybe if I can start with the CapEx investment and the ramp here, I know you provided us an update at the Investor Day and you mentioned that activity really ramped with customers again since then and engagement did ramp. I'm trying to think like when you are sort of going ahead and doing this investment, should we think about this as something that drives revenue in 2027 itself? Or are these sort of programs as well as the ramp sort of more to address customer demand in 2028, 2029. Just trying to get a sense of what kind of program visibility customers are giving you already to drive this significant investment from you? Just trying to get a sense of that? And I have a follow-up.

Samik, yes, the capacity that -- the CapEx that we're investing in now, as I mentioned earlier, is based on booked business. With respect to 2026, we do have the capacity to grow beyond our current high confidence outlook. So the investments we're making are enabling additional capacity for 2027 and 2028 based on booked business. Now as we continue to win in the marketplace, we'll further evaluate our capacity expansion plans, and then there will be an opportunity to expand our revenue outlook for '27 into '28. But right now, the investments we're making in '26, which also will have a follow-on effect into '27 is really just on the backlog of business that we have right now.

Got it. Okay. And then maybe for the follow-up, the outlook that you're sharing for CCS to maintain these sort of strong growth rates into 2027. Just wondering, does that sort of incorporate the digital native customer and the ramp with that customer? -- And any updates in terms of over the last sort of 90 days, anything -- any updates in relation to your timing or sort of how you think about the magnitude of that ramp in 2027?

Yes, Samik. So we are seeing accelerating growth happening within CCS. If you go back to our commentary from 3 months ago versus today, 3 months ago, we were saying that when you break down the numbers that CCS would be growing by about $3.5 billion in '26. And then when we put a 40% growth rate on that, it was implying about a $5 billion of CCS growth in 2027. We're now updating those numbers and going off of a higher base. So now what we're implying is that 2026, CCS will grow probably closer to $4.5 billion, so about $1 billion higher than what we talked about 3 months ago. And because we're saying that we're seeing very strong trajectory continuing, we're now seeing CCS grow close to $7 billion in 2027 and that's off of a higher base. And so the demand outlook is very robust.

Your question on the digital native customer, that continues to progress just as we would have expected it to. We still expect it to be a meaningful contribution in 2027. We are actively working on the design aspects of the program. And we do believe that, that program will still ramp into the '27, and that's included in the numbers that we're sharing.

[Operator Instructions] Your next question comes from the line of Thanos Moschopoulos with BMO Capital Markets.

Can you speak to how we should think about the margin trajectory? Just given the mix shift dynamic where you've got some enterprise becoming the larger part of CCS mix, would that imply that there might be some compression in CCS margins as the year progresses and into '27? Or are there offsets to that?

Yes, we're seeing tremendous amount of growth happening right now in enterprise. We are really pleased with the trajectory that we are -- that's already underway. You saw that we had a very nice growth number in the fourth quarter, and that's accelerating as we go into Q1. We expect that program to continue to grow all through 2026. And then just as a reminder, we've already won the next generation of that program. And so we would expect those programs to actually ramp into 2027.

So our outlook for enterprise continues to be very healthy. We are seeing very strong operating leverage. And so we don't necessarily expect a large mix headwind, if you will, from growing of the enterprise business. We do make more money on networking in general. But with the leverage that we're getting and the very disciplined cost management, we still think that the enterprise business is going to be able to generate very strong profitability.

And so for that reason, it's embedded in our numbers. 2026, we're giving an outlook right now where margins expand by 30 basis points. And what I would just say is that that's our -- that's the floor of our expectation. We would be looking to do better than that hopefully.

I would also add, Thanos, that networking is also very strong in 2026 and going into 2027. In 2026, we see 400G very resilient, the 800G very strong and we see 1.6T ramping in the back half of the year. So we have all 3 major programs running concurrently, which is helping the operating leverage and also helping the mix.

Your next question comes from the line of Michael Ng with Goldman Sachs.

Great. My question is just around the CapEx. Encouraging to hear about all the visibility your partners are giving you. I wanted to ask whether the capital intensity in the business has changed at all? Or does the $1 billion CapEx support 2% to 2.5% revenue over time, kind of implying a path to $40 billion to $50 billion of revenue over time. Is that a fair way to think about it? Or has the capital intensity in the business changed at all?

Michael, I'm not going to help you back into that number, but I completely understand the way that you look at it. What I would say is that we have in the last number of years been investing the majority of our CapEx dollars into growth CapEx. We spent probably $70 million to $80 million on maintenance and that's not going to change very much. And so as a percentage of revenue, we expect that our maintenance CapEx is going to be very predictable and not a huge driver. And so therefore, the delta is really on growth CapEx.

To the point that Rob has made, we are making this sizable investment to tie to programs that we've already won that are going to be generating, we believe, material revenue in 2027 and 2028. Should those wins continue, and we would expect that they would, we have no hesitation in increasing our CapEx. But you almost want to think of it almost like at a project level. We are building -- we're making these investments to support specific wins at this time. At a certain point, we would expect the CapEx to moderate because, again, the vast majority of it is growth. And so when we get back to a maintenance level, we would be back to what we would normally expect.

Your next question comes from the line of Karl Ackerman from BNP Paribas.

So I know you have deep engagements on the 400-gig and 800-gig switch programs, but could you speak to the opportunity you have to address multi-rack scale-up XPU networks, such as optical circuit switches and co-packaged optics-based switches perhaps in terms of the breadth of customer engagements.

Yes, sure. So we see increasing activity and increasing R&D expenditures. Some of it's a little premature to talk about now, but to do more AI/ML and networking integrated -- fully integrated systems, both supporting scale-up and scale-out fabrics. Based on a proof point with our digital native and some other early engagements that we have with other providers, we see this as a major growth driver for our business moving forward.

As these AI models continue to grow and GPU to GPU interconnects become more and more important, scale-up will be as much as an opportunity as scale-out. So we see this as a major growth opportunity for us. And we're well underway in capturing a lot of that -- those growth opportunities and we hope to have more to share with you in coming months.

Karl, what I would just add to Rob's comment is that when you look at the 1.6T wins that we've already had to date, they are both being used for scale-up and scale-out. And with the funnel of opportunities that we have in front of us that diversification continues, and we would expect that we would continue to grow in that area.

In addition, I think to the question that you raised on co-packaged optics, we are starting to see conversations with our customers increase in this area. We still believe that in terms of mass adoption, it's going to be more towards 3.2T, which are programs that we're working on in our R&D group. But we haven't seen customers looking for mass adoption of CPO as of yet. And so that's pretty similar to what we said a few months ago.

Your next question comes from the line of Tim Long with Barclays.

I did want to just talk about a few comments on the call you guys made about new programs and new program wins. Could you talk a little bit about kind of -- you talked about some strong backlog and visibility and wins as well as, obviously, the capacity expansions. You obviously got a lot of large switching and AI/ML and digital native rack wins. Can you talk about the outlook for the next few years. What we should expect to see from newer programs where they could be centered? Would this more be around new switching customers or new applications or use cases from some of the existing customers? Anything you could give us on that would be helpful.

Yes. The visibility, Tim, that we're seeing with our customers at this stage is unprecedented. We're certainly into '27 and many customers were talking into 2028. Our customers now are viewing us less as a supply chain partner and more as a technology leader. And part of that process is aligning on our technology road maps, which is informing our investment decisions. And these investment decisions are enabling and informing all the future products moving forward. And those products are more in the lines of a fully integrated rack systems supporting in scale-up and scale-out. Also staying on the leading edge of switching 3.2T samples are due in probably towards the end of 2026 and we're already starting to work on that. Mandeep alluded to some of the proof of concepts that we have co-packaged optics. So where it's only going to be ready for when that hits down the 3.2 cycle as well. So broadly speaking, our portfolio is getting broader and deeper with our customers moving forward.

Your next question comes from the line from David Vogt with UBS.

So I have a question about sort of the scope of work and the economics of the digital native customer. Can you kind of update us on where we stand in terms of what that relationship looks like as we go into '26 into '27? And then Mandeep, on the CapEx numbers, that $1 billion, can you help us parse through how much of that CapEx is tied to sort of the existing customer base and the expansion of programs and projects with your largest customers versus incremental customers like the [ DNC ] or any other incremental customers that you see in the pipeline for '26, '27?

Yes, I'll start off. With respect to the digital native customer, we have a very tight engineering-driven relationship with our customer. In 2026. So we're going to be shipping them largely samples and getting ready for the ramp that should be starting in the early parts of 2027. At this stage of the game, the program is on track, and we're just getting ready for the ramp working with them and the silicon provider and all the ecosystem partners, but it's -- the relationships is a solid relationship.

Yes. And just to add on to that in terms of question for CapEx and how it's kind of being allocated. So geographically, now you're aware of how we're allocating it. We're putting in significant investments in areas like Thailand, Richardson, Texas as well as Fort Worth. And it's really to support multiple customers. And so we are largely investing in programs that we've won across the major hyperscalers. Those are very -- we have a high confidence view to work with these customers sometimes for well over a decade. But with our digital native customer, we are willing to make investments as well. And so some of the investment is going towards enabling the ramp in 2027. But I would say the vast majority of the expenditures are tied to programs with our hyperscalers.

Your next question comes from the line of Paul Treiber with RBC Capital Markets.

Just a question, just in light of the new program win momentum that you're seeing, can you speak to how the returns and expected returns on those programs compare against existing programs? And really, what I'm going to add is also, are you seeing competition changing the returns on new programs versus what you saw in the past?

Yes. Paul, I'll take the first part of the question, and I'll let Rob talk about the competitive intensity that's happening in the marketplace. Look, the approach that we take when we make investments with our customers is really a holistic view. We look at it on a global basis. We want to ensure that we're generating strong profitability. But more importantly, we want to make sure we're supporting our customers in the geographies that they need. And so we will look at investments at the customer level on a global basis. But of course, we want to ensure that specific investments tie out on their own as well.

I know you know this, which is we're a very ROIC-driven company. We're focused on strong profitability, but just as much we're focused on a very disciplined level of investment. And so we'll make sure that business cases hold. And so from a returns perspective, what I would just say is that we continue to focus on expanding our ROIC. We continue to focus on expanding our margins while generating very strong top line growth. And so those are always factors whenever we're looking at business cases.

In terms of competitive intensity, and I would say as time goes on, the programs that we're bidding on and winning are becoming more and more complex. In many cases, some of the business that we decided not to play the pricing game on in 2025 have come back to us in 2026 because others could not execute on it. So when we look at our competitive moat, we have some fantastic engineering to be able to design these complex products. But even more so a very few of our competition can produce these products at scale. And when you combine those 2 together, it's really giving us a lot of tailwinds in '26 and also moving into 2027. That combination is proving to be very powerful for us.

Your next question comes from the line of Ruben Roy with Stifel.

Rob, maybe you could follow up where you left off there, and I had a question on the 1.6T win, the new win at a new hyperscaler. Are you seeing a shift towards HPS design-led solutions and away from cost plus? You've got the design center that you talked about in Austin. Just wondering if that's something that's happening as you move towards these more complex switching technologies and how you see that playing out from a margin perspective as you think about 2027, '28 time frame?

Yes. Certainly, thanks for the question. 1.6T and even as we move into 3.2T, the complexity that's required, the engineering complexity that's required on these things is moving more towards HPS engagements. So on the networking side, we see that increasing over time. And the density and the complexity is only going to increase at every node.

On the AI/ML compute side, we like to play really on the HPS and JDM design-oriented AI/ML compute. And we also see as that gets more and more sophisticated, more opportunity for us to play in that area, and we have several projects in the pipeline to improve those engagements on the HPS side as well.

Your next question comes from the line of Steven Fox with Fox Advisors.

First of all, congratulations on reaching a point where people are complaining about 37% growth. I thought that was great. In terms of my question, there's been a bunch of confusion around with your largest customer, how the supply chain works on those AI/ML compute programs and where you are sort of positioned versus their other suppliers? Is there any -- can you just sort of clarify how you're playing there? What kind of competition you see? And then it looks like you're also expanding directly to support some more programs on that. So anything on that would be helpful.

Certainly. I would chalk this up, you can't believe everything you read. What I can emphatically say is that our partnership with Google has never been stronger or more integrated. We have absolutely no indication there are new entrants into that market. As you know, these are very complex products to manufacture, especially at scale. And we have been doing it for a very long time with this family of products. As a preferred partner with Google on these leading edge compute programs. We have a joint commitment to each other moving forward, not just for the current generation, but for future generations of their TPUs. And we've been supporting this technology for generations and we hope to continue to do so going well into the future, which is warranting a portion of the investments moving forward. And I would also add that the capacity expansion that we're making certainly is in support of Google, but it's also in support of growth from other hyperscalers and digital native support.

Your next question comes from the line of John Shao with TD Cowen.

So within your guidance, how much do you bake in a price increase of key components or materials? At this point, are you still comfortable with the supply chain? Do you think this is going to be any source of potential margin compression given right now, we're getting this inflationary environment in the supply chain?

Yes. So we factored in inflation and pricing into the numbers that we've already shared. Just as a reminder to everyone on the call, when we have networking, we have it on a turnkey basis, which is our typical approach, meaning it includes the silicon, where on the compute side, it typically does not. And so where there is a lot of price inflation, it's happening on the silicon side. So you're not going to necessarily see our growth in our enterprise numbers being driven by that.

On the networking side, we're growing in terms of overall volume. But yes, there is inflation happening at the silicon side, which we're able to pass on to our customers. And so are we seeing margin compression? No, not right now. But if silicon becomes a much larger part of the bill of materials, then perhaps it will, but that's not in our line of sight at this time. But there is a little bit of contribution in our revenue growth year-over-year coming from just the fact that ASPs are going up, but the vast majority of the growth is due to units.

Your next question comes from the line of Todd Coupland with CIBC.

I wanted to ask about the 1.6 programs in the second half of the year. And at this point, what are the range of outcomes and gating factors for those programs to start to ramp this year? Just talk about that a little bit.

Yes. We have 10 active 1.6T programs in the pipeline right now. And 5 of them will start ramping in the back half of the year and certainly into 2027. And several -- the balance of them are in the development pipeline and will be ramping later in '27 into '28. The gating factors really is just completing the development cycle as planned and things are on track. Silicon is on track. So I just think it's business as usual in terms of supporting our customers' ramps.

Todd, if I could maybe add to that. When we look at our -- the overall switching demand that's out there right now, what we're really encouraged by is there's been a tremendous amount of growth happening in 800G that happened in 2025, and that's continued in 2026 and 400G continues to hold. So 400G will be a strong contributor in 2026. 800 will continue to grow.

And then you got 1.6 coming on as well towards the end of the year. And so the dynamic that's really been playing out in the last couple of years is that the next-generation technology is not necessarily cannibalizing the previous generation. And so this is one of the reasons that we have a lot of optimism on the networking space exiting '26 even and going into '27.

Your next question comes from the line of Atif Malik from Citi.

We got a couple of questions from investors on this yesterday. In your press release, you called out Google TPUs as a preferred manufacturing partner versus sole source. Is that a new disclosure? And then just as a follow-up, if some of your hyperscalers were to adopt more TPUs, do they all go through you guys? Or there are other entities like Broadcom and others that can participate in the TPU rack [ trade ] business?

On the first one, Atif, no, I don't think it's a new disclosure. We're not sole sourced or single sourced on the TPU programs nor have we -- frankly, nor I think we've ever said that. For [ BCP ] purposes, most, if not all of our hyperscaler customers remain a second source. But we are a primary source for them on TPU programs and continue to do so. With Google and with all of our hyperscalers, share is largely awarded on performance. Our performance has been very strong. And as a result, they make the decisions accordingly.

And then to the question that you were raising about as Google's TPU gets adopted beyond just Google itself, how does that play out. Right now, our view is that those -- that increased level of demand for their types of products will flow through their supply chain. And as their preferred manufacturing partner, we would expect to be able to support them with that. And so right now, it's wonderful to see that their product is being adopted in the marketplace, and we do expect to be able to support them with that growth.

Your final question comes from the line of Robert Young with Canaccord Genuity.

On the third hyperscaler, 1.6 win, how was this one? Was it an extension of 800? Was it tied to your Tomahawk ASIC experience. And, like, is it part of a rack integration with another outside vendor? Or is that being done by the hyperscaler? Just some context around that? And then if you could also talk about how you expect operating margins to evolve as you move into 1.6 terabytes programs and how that might differ between -- I think you have 2 full rack and then 2 stand-alone if I understand the large programs. Now how would the margin structure differ and evolve?

Rob. Yes, on the third 1.6T, so with this hyperscaler, we were predominant share on the 400G. We were predominant share and one on 800G, and this is just an extension of going to the 1.6T. The engagement started with a design win that we're happy with the performance with this switch. It's based on the 400 and 800. And we were awarded the mass production for this switch as well.

Yes. And then in terms of the margins, Rob, what I would just say is that we approach our switching portfolio in a similar way even as we go into the next generation, we typically make more money during the ramping and the development cycle of a program. And then as it gets to mass production, we try to offset that pricing with operating leverage. And so we do expect 1.6 programs to be as profitable as we've seen on some of our past switching programs.

One interesting dynamic, though, is that more and more of our switching portfolio should be moving towards HPS. We have some of our switching portfolio today in EMS. And just typically as we embed more of our engineering, that leads to better pricing. And so we are happy with the way that the margin profiles look like for 1.6 products.

And is there any context on between the full rack deployment and stand-alone?

Yes, it's integrated. And so we take a look holistically when we are doing this for our customers. As you mentioned, it's integrated. So there's 1.6 switches, but then there's also compute and then there's the integration activities, so we do testing for them. And then at certain points, we may be able to do services as well. If you look at it on a holistic basis, and we ensure that the value that we're bringing on the switching side, which has the most engineering that we have is getting captured in overall price.

There are no further questions at this time. So I will now turn the call back to Rob Mionis, CEO, for closing remarks.

Thank you. And thank you again for joining us this morning. 2025 was an exceptional year for Celestica, characterized by record financial results. We're excited to build on this momentum in '26 and as we raise our annual revenue outlook to $17 billion. The strategic investments we are making provide us with the capacity to support our customers' multiyear AI road maps and our deep partnership with industry leaders like Google and our expanding global footprint in Texas and Asia reinforces our confidence that our growth trajectory will be sustained into 2027 and beyond.

We look forward to updating you on our continued progress next quarter, and thank you again for joining the call.

This concludes today's call. Thank you all for attending. You may now disconnect.

Ladies and gentlemen, thank you for joining us, and welcome to the Celestica Q3 2025 Financial Results Conference Call and 2025 Investor and Analyst Day. [Operator Instructions].

I will now hand the conference over to Matthew Pallotta, Head of Investor Relations. Please go ahead.

Good morning, and thank you for joining us on Celestica's Q3 2025 financial results and investor and analyst day conference call. On the agenda for today's call, we will begin with our third quarter financial results, followed by our 2025 Investor and Analyst Day. At the conclusion of the prepared remarks, we will open up the lines for Q&A.

Joining us on today's call to provide prepared remarks will be Rob Mionis, President and Chief Executive Officer; Mandeep Chawla, Chief Financial Officer; Jason Phillips, President of our Connectivity and Cloud Solutions segment; and Todd Cooper, President of our Advanced Technology Solutions segment. They will also be joined by Steve Dorwart, Senior Vice President and General Manager of Hyperscalers for the Q&A portion of our call.

Please note that during the course of this call, we will make forward-looking statements, including statements relating to the future performance of Celestica, business outlook and anticipated trends in our industry and their anticipated impact on our business, which are based on management's current expectations, forecasts and assumptions.

These forward-looking statements are subject to risks and uncertainties that could cause actual results to differ materially from our expectations. For identification and discussion of these material assumptions, risks and uncertainties, please refer to our public filings with the SEC on SEDAR+ as well as the Investor Relations section on our website. We undertake no obligation to update these forward-looking statements unless expressly required to do so by law.

In addition, during this call, we will refer to various non-GAAP financial measures. We have included in our earnings release found in the Investor Relations section of our website, a discussion of those non-GAAP financial measures and a reconciliation to the most comparable GAAP measures. Unless otherwise specified, all references to dollars on this call are to U.S. dollars. All per share information is based on diluted shares outstanding and all references to comparative figures are a year-over-year comparison.

Let me now turn the call over to Rob.

Thank you, Matt, and good morning, everyone, and thank you for joining us on today's call. We are pleased to have the opportunity to speak with you today and to share some of the exciting developments in our business and our plans for the future. Before diving into the Investor and Analyst Day portion of our call, Mandeep will begin with a review of our third quarter results and provide our guidance for the fourth quarter.

Mandeep, over to you.

Thank you, Rob, and good morning, everyone. In the third quarter, we once again saw exceptionally strong demand in our CCS segment, which drove very strong overall performance across our key financial metrics. Revenue of $3.19 billion was up 28% and above the high end of our guidance range, driven by a very strong demand in our communications end market.

Our non-GAAP operating margin was 7.6%, up 80 basis points, driven by higher margins across both of our segments. This once again represented the highest quarterly non-GAAP operating margin in the company's history. Our adjusted earnings per share for the quarter was $1.58, exceeding the high end of our guidance range and an increase of $0.54 or 52%.

Moving on to some additional metrics. Adjusted gross margin was 11.7%, up 100 basis points, driven by higher volumes and improved mix in both segments. Our adjusted effective tax rate for the quarter was 20%. And finally, strong profitability and disciplined working capital management resulted in adjusted ROIC of 37.5%, up 850 basis points versus the prior year period.

Moving on to our segment performance. Revenue in our ATS segment for the quarter was $781 million, down 4%, slightly lower than our guidance of a low single-digit percentage decline. The lower performance year-over-year was primarily driven by portfolio reshaping in our A&D business as discussed in past quarters.

Our ATS segment accounted for 24% of total company revenue in the third quarter. Revenue in our CCS segment was $2.41 billion, up 43%, driven by very strong growth in our communications end market. The CCS segment accounted for 76% of total company revenue in the quarter. Our communications end market revenues increased by 82%, above our guidance of low 60s percentage growth. The growth was driven by very strong demand in data center networking, primarily for ramping 800G switch programs across our largest hyperscaler customers, complemented by solid demand in our optical programs.

Revenue in our enterprise end market was lower by 24%, which was in line with our guidance of a mid-20s percentage decline due to a technology transition in an AI/ML compute program with a hyperscaler customer. Our HPS business generated revenues of $1.4 billion in the third quarter, representing growth of 79% and accounted for 44% of total company revenue. The very strong growth was driven by accelerating volumes in our ramping 800G switch programs.

Moving on to segment margins. ATS segment margin in the quarter was 5.5%, up 60 basis points, primarily driven by improved profitability in our A&D business. CCS segment margin in the third quarter was 8.3%, an improvement of 70 basis points, driven by a higher mix of HPS revenues and benefits from operating leverage. During the quarter, we had 3 customers that each accounted for at least 10% of total revenue, representing 30%, 15% and 14% of revenue, respectively.

Moving on to working capital. At the end of the third quarter, our inventory balance was $2.05 billion, a sequential increase of $129 million and a year-over-year increase of $226 million. Cash cycle days during the third quarter were 65, an improvement of 1 day versus the prior year and sequentially.

Turning to cash flows. We generated $89 million of free cash flow in the third quarter, bringing our year-to-date free cash flow to $302 million. Capital expenditures for the third quarter were $37 million or 1.2% of revenue, while capital expenditures year-to-date were $107 million and also 1.2% of revenue. We anticipate capital expenditures to increase in the fourth quarter and for total annual CapEx to be approximately 1.5% of revenue.

Turning to our balance sheet and capital allocation. At the end of the quarter, our cash balance was $306 million, while our gross debt was $728 million for a net debt position of $422 million. We had no draw outstanding on our revolver, leaving us with approximately $1.1 billion in available liquidity.

Our gross debt to non-GAAP trailing 12-month adjusted EBITDA leverage ratio was 0.8 turns, an improvement of 0.1 turns sequentially and 0.3 turns versus the prior year period. As of September 30, we were in compliance with all financial covenants under our credit agreement. During the quarter, we did not repurchase any shares under our normal course issuer bid, and our year-to-date repurchases stand at $115 million.

Looking forward, we will continue to be opportunistic towards share repurchases. And as such, we are in the process of renewing our NCIB program, which is set to expire on October 31. We expect to receive the necessary regulatory approval and to commence the new program in November.

Now moving on to our guidance for the fourth quarter. Similar to last quarter, we highlight that our guidance figures assume no material changes in tariff or trade restrictions compared to what is in effect as of October 27, as any changes to these policies and their potential impact on our results cannot be reliably predicted at this time.

Fourth quarter revenue is projected to be between $3.325 billion and $3.575 billion, representing growth of 36% at the midpoint. Adjusted earnings per share are anticipated to be between $1.65 and $1.81, representing an increase of $0.62 at the midpoint or 56%. Assuming the achievement of the midpoint of our revenue and adjusted EPS guidance ranges, our non-GAAP operating margin would be 7.6%, an increase of 80 basis points year-over-year. We expect our adjusted effective tax rate for the fourth quarter to be approximately 20%.

Finally, let's review our end market outlook for the fourth quarter. In our ATS segment, we anticipate revenue to be down in the low single-digit percentage range as growth in our Industrial and HealthTech businesses are being offset by lower volumes due to portfolio reshaping in our A&D business and market-related softness in our capital equipment business.

In our CCS segment, we anticipate revenue in our communications end market to grow in the high 60s percentage range, supported by continued strong demand for our data center networking switches, including ongoing ramps in multiple 800G programs. In our enterprise end market, we expect to resume growth in the fourth quarter with a low 20s percentage increase in revenue, driven by the ramping of a next-generation program for hyperscaler application in AI/ML compute.

Based on our guidance for the fourth quarter and strong year-to-date performance, our latest 2025 financial outlook now calls for revenue of $12.2 billion, up from $11.55 billion previously, reflecting year-over-year growth of 26%. Our adjusted EPS outlook has increased from $5.50 per share previously to $5.90 per share, implying growth of 52%. Our non-GAAP operating margin of 7.4% remains unchanged. We are also increasing our free cash flow outlook for 2025 from $400 million to $425 million.

And with that, I'll turn the call back over to Rob to begin our 2025 Investor and Analyst Day. Rob, over to you.

Thank you, Mandeep. In the rest of our time this morning, we would like to provide you with a view of where our business stands today and the strategy that got us here. Importantly, we'll then share our view on where we are headed as a company, including our significant market opportunities and the investments we are making in our operations and technology road maps.

Celestica is our global technology platform solutions company. As our fundamental value proposition, we leverage vertically integrated capabilities and provide customized solutions enabling our customers to deploy leading technologies at scale, achieving rapid speed to market. Our goal is clear: to lead and accelerate market advancements in our focused technologies. We achieved this by proactively investing in next-generation technology road maps and the advanced capabilities required to deliver those technologies to market.

Celestica leverages our comprehensive vertically integrated capabilities to deliver leading technology platform solutions. We offer complete end-to-end capabilities, starting with design and engineering through manufacturing and supply chain management to software and aftermarket services. The depth of our system-level capabilities and expertise is best reflected in our technology solutions for the data center across networking and AI/ML compute. However, we leverage a set of competencies and strengths across a range of markets and technologies.

Customers have the flexibility to leverage all or any combination of our capabilities to build tailored platform solutions for their entire product life cycle. So let's look at where we stand today. As Mandeep shared, we are currently delivering the strongest financial performance in the company's history. We will dive deeper into both of our segments shortly to discuss the fundamental factors driving our performance and our plans for the years ahead.

However, first, I would like to take a brief look back at how we arrived here. When I took the CEO role in late 2015, my first action was to solidify the core leadership team. However, the real inflection came in 2018 when we began executing a comprehensive transformation to reshape our business. This was a fundamental shift. We aggressively ramped our investment in design engineering and technology road maps for the data center, while we deliberately disengaged from low-margin, low complexity programs that offered limited opportunity for differentiation and value add.

By 2020, we introduced our 400G switch, marking a pivotal moment in our HPS business, establishing our presence in the high-performance Ethernet switch market. Since then, we have rapidly grown our hyperscaler portfolio, reshaped the margin profile of our business and entrenched our position as a technology leader and key enabler of AI infrastructure solutions for the world's largest data center customers.

Yet the changes made to date may seem modest in comparison to the opportunities that lie ahead. We are currently navigating the most rapid period of change in our company's history, and the pace of that change continues to accelerate, driven by the massive investments in AI infrastructure by our customers. Celestica's culture is rooted in the pursuit of progress, and we are incredibly excited and motivated by the opportunities in front of us.

During this period, we have seen accelerating momentum in the growth of our business, and we are capitalizing on this strength. Based on our 2025 outlook, we are on track to deliver our strongest performance on record. There are a number of key drivers supporting the sustained improvement in our performance. The first driver is capturing share in high-growth markets with the cornerstone being our presence in AI data centers.

Next, demand for our HPS product offerings are rapidly shifting our entire portfolio toward higher complexity engagements, where our design collaboration and value add are critical to our customers' success. In addition, growing volumes are fueling improved operating leverage, and we relentlessly drive productivity and efficiency across our global network with a strong focus on operational excellence.

Our global network operating across 16 countries is essential part of our value proposition. Our customer-centric network strategy provides a reliable and consistent supply chain solution, allowing our partners to derisk their geographic exposure, a capability that's essential in the current climate of geopolitical and trade uncertainty. We have been and continue to make significant expansions and upgrades to our network funded by operational cash flow to support the growing demand and program ramps with our AI data center customers.

Demand for North American capacity remains strong, especially within the United States. To accommodate this, we are continuing to deepen our footprint in Texas. We're expanding square footage and increasing our power envelope at our Richardson site with capabilities to support the production of thousands of additional advanced AI racks annually.

We are adding to our global design engineering network with a new hub in Austin for closer customer collaboration. We are also in the process of finalizing plans for an additional large-scale manufacturing site in the state to support continuing growth with one of our largest customers.

We are equally committed to supporting our customers by investing for growth in Asia, where we continue to make significant additions to our largest campus in Thailand. We are seeing incredibly strong demand from hyperscaler and digital native customers for tight capacity with significant production ramps planned to commence through 2027 and into 2028 across networking and compute. This proactive and regionalized investment strategy ensures we retain a flexible and reliable network positioned to meet our customers' evolving needs and accommodate the significant growth in demand from our customers.

Operational excellence is ingrained in our company's DNA and an integral pillar of our competitive differentiation and value proposition. The Celestica operating system is our standardized framework that ensures unmatched consistency in quality, reliability and on-time delivery across every one of our global sites as we deliver hundreds of highly complex programs simultaneously.

One of the core components is our culture of accountability, emphasizing execution and safety. This is reflected by our performance in these areas, tracking well above industry benchmarks, including 0 critical excursions to customers. We pride ourselves on our customer-first mindset, evidenced by our history of deeply entrenched collaboration in design and engineering, where we have positioned ourselves to act as an extension of our customers' teams.

Another operational priority is our continuous investment in our advanced manufacturing capabilities, including automation and testing, which are critical to enabling product road map development and speed to market in deploying new technologies.

Next, I want to briefly detail a case study that will help bring to life our competitive differentiation enabled by our core success drivers. In this instance, a hyperscaler customer approached us to collaborate on the design of our first of its kind rack-scale liquid cool 1.6T networking solution needed in order to accommodate the increased density and power requirements of the latest generation AI networking platforms. This highly customized design was intended to be integrated into the new state-of-the-art data center architecture.

The customer required an accelerated road map to allow the solution to be early to market, leveraging Broadcom's Tomahawk 6 SC silicon, making speed to market a key consideration. In addition, the customer required multi-node manufacturing capabilities in Asia and the U.S. to support the delivery of the program. As with many of our key engagements, managing complexity was a defining factor.

Celestica was awarded the program earlier this year based on a strong working relationship with the customer and their confidence in our industry-leading design engineering. They also valued our advanced manufacturing capabilities, specifically our ability to operationalize highly complex production lines for liquid cooled racks at scale and to do this faster and more seamlessly than other potential partners.

After receiving initial Tomahawk 6 samples earlier this year, we quickly stood up an operational prototype for the 1.6T switch and believe we were the first team anywhere to have done so. The program is scheduled to begin mass production next year.

As this example and our discussions today will illustrate, Celestica's success is driven by the unique combination of 3 core factors. First, we occupy industry-leading positions in markets with strong structural tailwinds and higher barriers to entry, supporting multiyear runways for growth. Our largest and fastest-growing market presence is within AI data centers, supporting high-performance networking and custom ASIC AI/ML compute platforms.

Second, with these focused markets, we seek to accelerate market advancements through technology leadership. Our early-stage investments in R&D and next-generation product road maps support our ability to remain at the leading edge of technology transitions and enable our customers' speed to market in deploying new technologies. We separate ourselves by helping our customers address complexity and by solving the hardest challenges effectively.

Third, we are steadfastly focused on maintaining best-in-class operational execution. Our global footprint, combined with the rigorous processes of the Celestica Operating System, ensures we can manufacture and deliver the highest complexity products without compromising quality and reliability. Fundamentally, these 3 factors serve as a foundation of our success and our unwavering confidence in the opportunities ahead.

I'd like to now turn the call over to Jason to walk us through our CCS segment. Jason, over to you.

Thank you, Rob. It's great to be here with you this morning. The past several years have seen our business on an incredible growth trajectory. In the midst of what is potentially the most significant secular investment cycle in a generation, we are in the incredible position of supporting the world's largest data center customers in their massive infrastructure buildouts to enable the growth in applications of artificial intelligence. This year, we are tracking towards $9 billion of revenue, more than doubling the size of our business from just 3 years ago.

Alongside this growth, our business mix has shifted towards higher complexity customized programs within our HPS portfolio, helping drive strong profitability. Double-clicking on our HPS portfolio, we are expecting to deliver approximately $5 billion in revenue for 2025, an incredible 80% growth, which speaks to the tremendous uptake from customers for our offerings.

We take a long-term view towards our investments and product road maps, investing early to help customers accelerate deployment of leading technologies. We have consistently grown our investments in R&D over the years. We increased our spend more than 50% this year, and we expect at least a 50% increase in 2026 in support of new program wins that will ramp beginning over the next 2 years.

Our design engineering talent is an important differentiator for our business. We have scaled our team today to more than 1,100 dedicated design engineers, supporting both hardware and software solutions across 7 global design sites and growing, and we expect to add several hundred additional resources in the immediate future. Our recognized leadership in design has been critical to winning the many new programs, which are driving our growth.

Next, we'll take a look at some of the key technology developments and design challenges we are seeing in the data center and where our team is making investments to address those opportunities. Importantly, as a platform solutions company, we are aiming to address these challenges at the systems level. In AI/ML compute, we are making investments in our rack-scale capabilities to support applications for both training and inferencing workloads, which we will touch on more shortly.

To stay ahead of the latest advances in liquid cooling, Celestica is building proof of concepts for our next generation of solutions, which includes innovations in single-phase, dual phase and emerging liquid metal technologies. The rapid advances in switching silicon bring with it increasing complexity in designing the latest networking platforms, particularly challenges in addressing power density and signal integrity.

Celestica is addressing these by driving innovations in our switch designs for 200-gig SerDes solutions to support 1.6T platforms and are planning early-stage investments for 400-gig SerDes to support 3.2T platforms. We're also staying close to the advances in optical technologies that will increasingly be utilized in high-performance networking such as linear pluggable optics, co-packaged optics alongside other interconnect technologies such as co-packaged copper.

As an example, our latest 800-gig and 1.6T switch designs support LPO for optimized power efficiency. And we are in the early stages of our product road maps for our future generations of switching solutions that will accommodate CPO technology.

We also see scale-up networking, which supports high-speed direct connectivity between accelerators as an emerging multibillion-dollar new market opportunity that is being unlocked in large part by the move towards open standards, supported by industry initiatives like UALink and Ethernet for scale-up networking. We are already on the path through recent program wins towards productizing our first solutions for scale-up Ethernet, which will leverage Broadcom's Tomahawk 6 silicon.

We look ahead at emerging technologies by proactively investing and collaborating with our ecosystem partners to define future product road maps. This enables speed to market and establishes Celestica as an essential partner for our customers' next-generation deployments.

As noted, Celestica looks to address technology leads at the system level. And accordingly, we have invested in developing capabilities to support full platform solutions. Our customers engage with us to support a wide array of programs and technologies and leverage the depth of our capabilities to varying degrees. However, we have increasingly observed our hyperscaler and digital native customers seeking bespoke solutions for rack-scale systems, making our full suite of capabilities across multiple technologies increasingly essential.

Critically, we are seeing this demand in both networking and AI/ML compute, where customers are seeking solutions for both custom ASIC and emerging merchant silicon platforms. Beyond designing the hardware for base technologies in networking, compute and storage, our engineering teams are supporting customers in orchestrating, testing and optimizing their rack-scale solutions, including the customization of software platforms as well as aftermarket services.

Our ability to deliver system-level solutions of this kind requires our breadth and depth of capabilities in all of these areas with the ability to integrate them into a seamless solution for the customer. Software is an increasingly critical component of our comprehensive solutions. To support this, we are making focused investments in our capabilities, having grown our global team to nearly 400 dedicated software engineers.

We believe that open-source network operating systems, namely SONiC, are positioned for continued market adoption driven by the desire for vendor diversity, cost effectiveness and sustained improvement and innovation. We have a long history of working with SONiC and our proficiency with this platform is well respected in the industry.

Our Celestica solutions for SONiC customizes and hardens SONiC features, providing customers bespoke solutions with an open-source base as well as related support services. But our software capabilities go beyond SONiC too, as we offer customers the optionality to leverage third-party solutions. And for hyperscalers using a proprietary network operating system, our software knowledge allows us to provide critical support with switch abstraction interfaces, ensuring silicon interoperability across the fabric and to assist with network operating system debugging and testing of customized hardware.

Our ability to deliver a diverse range of leading solutions is significantly enhanced by our ecosystem of partners across both silicon and software. Leveraging these relationships, we work closely and proactively with our technology partners, aligning years ahead of time on next-generation product roadmaps, enabling us to be early to market and deploying leading-edge solutions for our customers.

And our technology partners attest to the critical part we play in productizing and delivering these leading-edge solutions to the market. Broadcom's President and CEO, Hock Tan, highlights the significance of our capabilities and execution by recognizing Celestica as their preferred provider for the most technically demanding data center platform solutions. The strategic relationships between Celestica and industry leaders like Broadcom are a powerful testament to the importance of our role in enabling these critical technologies.

Now let's take a deeper dive into our market opportunities. As mentioned, we are witnessing a generational secular investment cycle in data center infrastructure, driven by artificial intelligence and cloud adoption. Many of the indicators from the companies leading this investment across silicon designers, hyperscalers and the emerging leaders in large language models point towards a multiyear runway of continued growth in data center CapEx.

Annual data center CapEx is forecasted to surpass $1 trillion by 2028, with commentary from leading voices in the industry suggesting this could prove to be conservative. These companies regularly highlight constraints in compute capacity driven by the increasing demands from both training and inference. All of these companies have signaled their commitments to continue to grow their investments in AI infrastructure, which aligns with the forecasts and planning discussions we are having with our customers.

Within our portfolio, hyperscaler customers have continued to be the primary driver of our revenue growth over the past several years. Demand remains incredibly strong and is supported by solid visibility based on program awards that are expected to begin ramping over the next 2 years. Furthermore, we're unlocking the next wave of expansion with our digital native customer portfolio, which is poised to ramp meaningfully starting in 2027 with the delivery of our first HPS rack-scale custom AI system, which we initially announced in January. Our broad portfolio exposure to AI-driven investments from the largest and most established players in the sector ensures our business is exceptionally well positioned to capitalize on this opportunity.

Moving on, we'll discuss the opportunities across our markets. Our communications portfolio is anticipated to generate $7 billion in revenue in 2025, an exceptional 78% growth. Our portfolio is anchored by our networking solutions with our 800-gig switch programs comprising the largest share and our most significant growth driver in 2025.

We anticipate that continued growth in 800-gig and multiple ramps in 1.6T beginning in 2026, including strong engagement in scale-up Ethernet, support a robust multiyear growth outlook for our networking business with our existing customer base alone. In addition, we also have a growing funnel of opportunities for both scale-up and scale-out applications across a diverse set of new and existing customers. We believe that our technical expertise and recognition as a market leader in networking places us in a solid position to continue to grow our portfolio.

At the Open Compute Project Global Summit earlier this month, we announced the latest additions to our growing family of high-performance Ethernet switches as part of our HPS portfolio, the DS6000 and DS6001. The DS6000 series utilize Broadcom's Tomahawk 6 silicon and are designed to support port speeds of up to 1.6T with routing optimized for AI/ML workloads across both scale-up and scale-out networking. Notably, the DS6001 incorporates direct-to-chip liquid cooling technology. We anticipate availability later in 2026.

These latest designs are a testament to our continuous innovation and our commitment to accelerating market advancements through technology leadership. We believe our optimism regarding our networking business is well founded. Our market share leading portfolio is supported by a number of company-specific and market-level tailwinds, which position us to continue to succeed in this fast-growing market.

The latest forecast suggest that the TAM for high-bandwidth Ethernet networking is projected to reach $50 billion by 2029. Within this market, the 800-gig and higher segments are projected to grow even faster than the overall market at an impressive 54% CAGR driven by upgrade cycles led by hyperscalers and leading large language models' providers to keep pace with the demands of the latest AI workloads. Based on the engagement we've seen already, we think that the adoption of scale-up Ethernet is going to be a really meaningful opportunity and additive to the growing overall Ethernet TAM.

In our case study earlier, we highlighted the increasing technical challenges with each successive new generation of networking technology, which we have proven highly capable of navigating. However, there are a number of additional highly favorable dynamics that we believe make this market an incredibly important opportunity. We'll touch on a couple of those now.

One of the core challenges in building AI data centers is scaling of the infrastructure. While the increasing computational power of accelerators or nodes is driving requirements for greater bandwidth, a critical compounding dynamic is the nonlinear growth in connectivity required as the number of accelerators within a cluster scales. The largest fully operational cluster today is believed to consist of roughly 200,000 accelerators. However, commentary from leading silicon companies suggests that multiple hyperscalers plan to deploy clusters consisting of up to 1 million accelerators within the next couple of years.

This rapid scaling in compute capacity required to support leading AI models requires huge increases in networking infrastructure, including high-bandwidth Ethernet switches, which comprise the majority of our communications portfolio.

The build-out of AI data centers is fundamentally shifting the share of spend towards back-end networking, which is expected to grow more than twice as fast as front-end spending. Back-end networking connects the compute clusters used for training models, while front-end connects the network to the external world, primarily for inference traffic. The unique demands of back end align with our competitive strengths, in particular, more intense performance requirements where factors like high bandwidth, low latency and sustained high utilization are an absolute necessity. The back end also necessitates shorter refresh cycles required for the network to keep pace with the increases in computational power.

Since our switch revenues are predominantly comprised of back-end shipments, we have meaningful exposure to the highest growth segment of the market. Our customers tend to be early adopters, and we help them accelerate their deployments of the newest switching platforms early on in upgrade cycles. This is reflected in our leading market share on the highest bandwidth and Ethernet switching for the data center across each of 200-gig, 400-gig and 800-gig platforms. Today, our cumulative market share across all of these speeds as measured by total ports shipped is 41%, more than double the next largest competitor's volume.

As the technical complexity rises with each generation of the technology, designing high-performance switches becomes increasingly challenging and fewer and fewer competitors can do it effectively. Managing this complexity and helping customers achieve speed to market with new technologies are what we really excel at, allowing us to secure the strongest share in the earliest stages of every new upgrade cycle.

We see custom solutions for high-performance AI networking platforms being widely adopted by our leading hyperscaler customers. This model offers the benefits of vendor diversity, cost effectiveness and highly tailored solutions, which become more pronounced as their infrastructure deployments scale. Consequently, we believe hyperscalers and increasingly large digital native customers will continue to leverage these solutions.

In this segment of the overall market, Celestica's share leadership is even more pronounced as we account for the majority of the total spend, 55%, having grown our share meaningfully over the last couple of years. Securing mandates and consistently executing on high complexity customized solutions for the largest customers in the industry reflects our competitive advantage and continues to validate our market strategy.

Moving on to our enterprise market, which includes our AI/ML compute and storage businesses. Our portfolio revenue is projected to be about $2 billion in 2025, and we expect to see meaningful growth in 2026, significantly exceeding our previous peak revenues from 2024, as we ramp the next-generation AI/ML compute program.

Looking further ahead, 2027 is expected to be another transformative year as we plan to ramp production for the rack-scale custom AI system with a digital native customer. The design work for this program is well underway, and we expect to receive initial XPU deliveries in the second half of 2026 to support early test deployments with full-scale production expected to commence in 2027.

The scale and scope of the custom solution, including design, manufacturing, orchestration and deployment for a leading-edge system of this nature is incredibly complex. But as we've highlighted, these are the kinds of challenging engagements where Celestica truly thrives. We anticipate this program will serve as a landmark proof point, showcasing our full suite of system-level capabilities.

Shifting to the market outlook, the TAM for accelerated compute is expected to grow to nearly $500 billion by 2029. Some of the tailwinds driving this growth are particularly favorable for our business. Given that our compute business is focused almost exclusively on solutions supporting custom ASIC platforms, we are positioned to benefit from the highest growth segment of the AI server market. Overall, the constraints on capacity we spoke about earlier, currently being experienced at the largest hyperscaler and digital native customers continue to highlight the clear requirement for more compute infrastructure and the strong demand in this market.

As mentioned, Celestica's market strategy is focused almost exclusively on the custom ASIC segment, which is forecasted to grow roughly sixfold over the next several years. An increasing number of the largest data center players in the market continue to pursue development of custom ASIC platforms, and we are seeing this trend within our own customer base. The architectures of these chips are designed to be optimized to support a customer's specific workloads with the intention to deliver lower hardware cost, power savings and overall better price to performance than a general-purpose GPU.

As AI models become more highly specialized, custom silicon architectures will also be an increasingly important means to enable differentiation and performance between models. And as compute infrastructures grow and scale, the benefits to deploying a custom ASIC platform successfully are magnified. Because custom ASICs also require highly tailored bespoke systems to be designed around the silicon, customers often require greater support from solutions providers, presenting us with better opportunities for value-added engagement on engineering and design.

We believe this fast-growing segment of the market better lends itself to our competitive strengths in customization and managing complexity and that there are fewer players that have our track record in supporting these kinds of platforms at scale. We are exceptionally optimistic about the future of our CCS business. We have confidence in our outlook, supported by visibility to upgrade cycles, strong customer demand forecasts and a robust pipeline of potential new opportunities. And we feel that we are in a prime position to capture the incredible market opportunities in front of us.

With that, I would now like to hand the call over to Todd, who will take you through the latest in our ATS segment.

Thank you, Jason. It is great to be with all of you this morning. Since we spoke last year, we've been focused on strategically remodeling the ATS portfolio for higher sustained profitability and higher mid- to long-term growth. Specifically, our previously discussed reshaping activities in A&D are offsetting otherwise solid base demand across the segment, leading us to expect revenues in 2025 to be approximately flat year-over-year. We have already seen the significant benefits of these actions on our profitability. After exiting 2024 with a segment margin of 4.6%, we have already improved to 5.5% in the third quarter of 2025 and expect to achieve 70 basis points of full year margin expansion.

Looking ahead to 2026, we anticipate revenue to be approximately flat to mid-single-digit percentage growth. We are seeing strong growth in our Industrial and HealthTech businesses. However, this demand will be partially offset by further selective reshaping across some of our markets. Over the medium to long term, our objective is to grow our portfolio at or above the growth rates of our underlying markets on a consistent basis, while balancing growth with sustainably higher profitability.

Our target financial framework for this segment is supported by our engineering-led strategy and our focus on deepening our long-standing relationships with the leading customers in our markets. Over the past number of years, our ATS business has made investments to deepen our engineering base by developing market-focused teams with specialized expertise in their respective industries technologies. Today, our team constitutes a global network of highly talented engineers along with labs and advanced manufacturing sites to support our customers across regions and markets.

Engaging customers early in the product life cycle strengthens our relationships by allowing us to offer a more holistic vertically integrated solution. This approach more fully leverages our core competency as an organization, helping our customers navigate complexity and solve hard problems, while having the added benefit of reinforcing our value as a highly capable partner and driving higher margins for the portfolio.

Today, about 1/3 of our more than 100 customers engage with our teams on engineering services, relying on us for support in testing, design as well as in accelerating their time to market on new product development. We believe this presents an excellent opportunity to deepen our engagements within our existing customer base.

And lastly, as discussed earlier, we are also continuously assessing and actively managing our customer portfolio. Our commercial strategy is focused on deepening our long-standing relationships with the leading Tier 1 OEMs in our markets. In pursuit of growth, we are intensely focused on maintaining the quality of our customer base and having a strong margin profile for our portfolio.

Now I'd like to briefly walk through each of our businesses, starting with Industrial and Smart Energy. In our industrial and smart energy portfolio, we anticipate growth in 2026, driven by demand recovery in our macro-sensitive end markets. Longer term, we are engaged on exciting new opportunities in robotics and automation as well as in on-vehicle technologies such as telematics and battery energy storage for heavy industries. We are also pursuing programs in the growing data center power infrastructure market, including power distribution, conversion and control equipment, leveraging some of our hyperscaler relationships. While it is still early days, we are encouraged by the traction we are seeing.

Next, let's move to Aerospace and Defense, which as a U.S. military veteran is near and dear to me. Our 2026 outlook sees base demand remaining healthy, supported by the ramping of new program wins, although we expect that growth will be offset by tougher comps from the first half of this year, driven by our reshaping activities. Longer term, we project healthy demand from U.S. and European defense spending, which we expect will become a greater share of the portfolio. And in our commercial aerospace business, we expect to see growth aided by program ramps with new and existing customers.

Moving on now to semiconductor capital equipment. In semiconductor capital equipment, we saw strong growth in the first half of 2025, although we are seeing a moderation of demand in the second half, consistent with the broader sector. We expect this to continue into at least the first half of next year as our customer conversations indicate foundries are holding off on adding more capacity until there is greater clarity on tariffs and trade restrictions.

To obtain greater efficiencies in our network, we are taking this opportunity to consolidate demand across some of our sites. At the same time, we are continuing to ramp new high-complexity programs in lithography and advanced semiconductor packaging. Long term, we believe the significant push for the near shoring of wafer fab capabilities in the U.S., Europe and China, driven by geopolitical factors is expected to support healthy demand for new capacity. We have exceptional capabilities and proof points in the semiconductor capital equipment market and anticipate this demand to serve as a tailwind for our business starting in the second half of 2026.

Finally, in our HealthTech business, overall demand remains robust, and we continue to make a concentrated effort towards driving higher portfolio exposure to this market. Last year, we discussed our investments in advanced manufacturing, automation and testing capabilities to support new wins in diabetes care, which are expected to ramp in 2026. Now as we approach the beginning of those ramps, we are anticipating more than $100 million of growth in our HealthTech business in the coming year.

In closing, our focus remains on driving high-quality, sustainable growth. We are successfully executing strategic commercial decisions to reshape our portfolio, which is already yielding significant improvements in profitability. Our portfolio is supported by healthy underlying near-term demand, along with solid long-term fundamentals. We remain confident that our thoughtful approach in each of our markets will position us to drive sustainable and profitable growth for the ATS segment.

With that, I would now like to turn the floor back over to Mandeep, who will discuss our financial outlook and capital allocation priorities.

Thank you, Todd. The outlook for the financial performance of the business in 2026 continues to be very strong. We anticipate revenue of $16.0 billion, representing 31% growth compared to our 2025 outlook. At the segment level, CCS revenue is expected to grow by approximately 40%, driven by strong market tailwinds and program ramps in both our enterprise and communications end markets.

Our outlook assumes continued strength in networking, supported by 800G demand growth and the ramps of our earliest 1.6T programs in the second half of the year. In AI/ML compute, we anticipate very strong growth as we reach full volume production of our next-generation custom ASIC program for hyperscaler applications.

In ATS, as noted, revenue is projected to be flat to up in the mid-single-digit percentage range, as healthy base demand and new program ramps are partially offset by our reshaping activities to drive higher profitability.

We expect non-GAAP operating margin to expand by 40 basis points to 7.8%, driven by favorable mix and productivity improvements. Our non-GAAP adjusted EPS is projected to be $8.20, which would represent a 39% increase year-over-year. We are targeting non-GAAP free cash flow of $500 million. This model represents our preliminary high confidence outlook for the coming year, and we will continue to update you on our forecast as the year progresses.

Importantly, our confidence extends beyond 2026. First, AI-related demand for data center technologies in our CCS business remains very healthy, and we are seeing many signals that suggest these secular dynamics have a multiyear runway ahead. Second, we have solid visibility to the ramping of significant new programs with start dates out to 2027.

Our view for 2027 assumes multiple ramps with hyperscaler customers with new programs supporting the 1.6T upgrade cycle, including scale-up solutions and a next-generation custom ASIC compute platform. We also anticipate the commencement of mass production of our rack-scale custom AI system program with the digital native customer. As a result, we expect these strong growth dynamics to persist. And in support of this, we are aligning our capacity with our customers, assuming that this trajectory continues into 2027.

As we continue to manage our financial priorities through this period of high growth, we intend to maintain a steadfast focus on maximizing shareholder value. We aim to achieve this by compounding our adjusted earnings per share in a sustainable manner over the long-term. This requires us to remain thoughtful and measured in our approach to pursuing earnings growth by assessing current and potential new business through the lenses of margin sustainability, alignment with our long-term strategy, our competitive advantages and return on invested capital. These guideposts help us to maintain discipline in managing our growth and evaluating our commercial opportunities.

Our consistent execution and disciplined approach to financial management has delivered improvements in each successive year across each of our key metrics. Based on our 2026 outlook, we expect revenues to more than double relative to 2022 and to lead to a more than fourfold growth in adjusted EPS over the same period, driven by the sustained expansion of our non-GAAP operating margin. We believe there is still room for additional operating leverage in our business beyond 2026. We anticipate maintaining our solid trajectory and compounding our adjusted earnings per share, which we believe will continue to translate into strong return for shareholders.

Taking a closer look at free cash flow. We have managed to consistently generate free cash flow on a quarterly basis going back many years, enabled by our strong working capital management and operational discipline. We also continue to grow our free cash flow, while simultaneously funding the rapid expansion of our business. Next year, capital expenditures are expected to rise to between 2.0% and 2.5% of revenue, funded by operational cash flow as we invest in our network to support the growth we anticipate over the coming years. We will maintain a disciplined approach to CapEx and working capital management as we ramp these investments.

While the primary aim of our investments is towards driving compounding of our adjusted EPS over the long-term, adjusted ROIC remains an important measure that we use to assess the quality of our investments. This has been reflected in our strong earnings growth directly translating into meaningfully higher returns on capital, which now sits at 35% year-to-date in 2025, having nearly doubled since 2022. This rigorous focus on capital efficiency seeks to ensure that our growth is high quality and that we continue to direct our resources towards its best and highest return use.

Our capital allocation strategy is built on 2 core principles: discipline to ensure we pursue the highest returns and best use of capital and strategic flexibility to maintain optionality to execute on new opportunities as they arise. Today, our highest priority for capital is to reinvest in the business to support long-term growth and the significant organic opportunities we see over the next several years. We also continue to assess M&A opportunities in a disciplined and selective manner, where acquisitions can serve as a complement to our organic growth and help accelerate our strategic road maps.

Our CCS funnel is primarily focused on adding or enhancing our capabilities in areas such as services and design engineering. And in ATS, we are looking to balance our portfolio by adding exposure to or scale in desirable markets that possess strong fundamentals. And finally, we will continue to return capital through share buybacks on an opportunistic basis.

Over the past 3 years, our share price performance has significantly outpaced the broader indices and the majority of our peer group. Our stock price reflects the very strong trajectory of adjusted earnings growth we've delivered over the last few years. We are confident that this strong earnings compounding will continue as demonstrated by the 39% adjusted earnings per share growth implied by our 2026 financial outlook. We believe that our valuation is supported by this strong track record and our anticipation of future growth.

With that, I'll now turn the call back over to Rob for his closing remarks.

Thank you, Mandeep. Before we close out, let me briefly reiterate the 3 key drivers that are the foundation for our confidence in our continued success. We are very optimistic about the future of our business. As I stated earlier, we are navigating a period of rapid but positive changes and the pace of those changes only continues to accelerate.

We believe the next several years present a truly remarkable opportunity for our company. We hope that all of you leave our call today with a richer understanding of our unique combination of capabilities and the strategic approach that enables our success. We provided perspective on our long-term vision, highlighted by the proactive investments we're making today to capture the opportunities we've discussed. We have the utmost confidence in our organization's talent, our commitment to excellence in delivering for our customers and our ability to execute on our strategy. Thank you for your time and continued support.

I will now turn the call back over to the operator to begin our Q&A period.

[Operator Instructions] And your first question comes from the line of Mike Ng with Goldman Sachs.

I guess, first, just on the investments that you're making in R&D, the 50% growth next year and the capacity expansions through 2028. I was just wondering if you could talk a little bit more about some of the key products that are supporting your visibility into these investments and are most of the investments grounded by expansions in customers that are new or existing?

And then just as a quick follow-up, I noticed you talked about the new storage platform win with the hyperscaler in 2026. Is that with your current hyperscale AI/ML compute customer, or is that somebody else? How would you size the opportunity in hyperscale storage?

Mike, this is Jason, and welcome, and we're glad you're covering us.

So as we look at our R&D spend and investments year-over-year, we've been making significant increases now for quite some time, and they are directed at where we are focusing our strategy and our opportunities largely around networking, AI/ML, and I would say, storage, rack level solutions and then everything you need to bring that total rack level, fully orchestrated rack level solution together, inclusive of software, liquid cooling, power, et cetera. So that's where we've been focusing our R&D spend. And we've also been making significant, I'd say, advancements and investments in our engineering network. We've now moved up to over 1,100 engineers, 400 of those are in software. We've moved to 7 design centers of excellence. And we're also looking at increasing that in places like Taiwan as well.

Yes. And with regard to the specific customer that you inquired about, it is an existing customer. We have a long-standing relationship with this company, and we've continued to evolve the relationship from providing single system level solutions up to fully integrated rack solutions, of which this engagement is. And that's part of our normal process to go and evolve with our customers.

With regard to storage, we have a few different opportunities where we're engaged in storage. It's less prominent than we would see in networking or our ability to extend some of our networking capabilities into the AI/ML compute space.

Your next question comes from the line of Karl Ackerman with BNP Paribas.

You noted that Thailand and Texas could see a doubling of capacity from 2024 to 2027, yet CapEx will only be 2% to 2.5% of sales. Could you speak to what assurances your largest customers are giving you to support this capacity growth, whether that is in the form of multiyear volume commitments and/or combined investment in tool CapEx?

Yes. Karl, it's Mandeep here. Thanks for the question.

So we've been very disciplined on our capital expenditures for many, many years. We're tracking towards 1.5% right now for this year. And as the revenue continues to grow, the dollars obviously are growing as well. So we're on track on just under $200 million of CapEx this year.

We're anticipating right now somewhere between $300 million and $400 million of CapEx next year. And these are investments that are tied to customer programs. We don't have a built-in and they will come approach. It's always tied back to customer engagement. And so we have very good visibility on the demand profile going out multiple years. So it gives us confidence to be able to invest in these types of areas.

The only other thing I'll mention, and then I can have Steve ask a little bit or comment a little bit on the customer engagement on our expansion. But I just want to make a note that only about 40 basis points of our CapEx spend is maintenance. And so if we're going to spend 2% next year, then that means 1.6% is all on growth, and that gives us a tremendous amount of discretion on where we put those dollars. And so we think that right now, that's sufficient.

This is Steve again. With regard to visibility to forecast and customer demand, we currently have about 12 to 15 months of real solid forecast inputs and demand inputs from our customers, largely around their 2026 budgeting and spend commit processes. But in many cases, we have visibility beyond that. In some cases, for specific customers, specific programs. There's a certain amount of ASICs, for example, that they may have committed to, and it gives us some assurance as to the longevity and the size of the overall program. So we do get extended visibility through being similar to that.

Understood. And if I could for a follow-up quickly. Your growth in CCS is notable, which appears driven by your networking switch opportunity in HPS. Could you speak to the relative mix you have today on 800-gig switch ports? And I suppose as you think about the trajectory of 1.6 next year, could you speak to the opportunity you have in liquid-cooled-based switches, which appear to be a growing opportunity for you, both in '26 and '27?

Yes. Karl, from a number perspective, why don't we start, we've been seeing tremendous growth in 800G this year to the point where we'll end 2025 with roughly a 50% split between 800G and 400G in terms of the products that we're delivering. As we look into 2026, we're seeing the 800G demand, in particular, accelerating. There are going to be projects where 400G continues to be used. We've been given some examples by our customers where 400G is expected to be used for many years still. But the growth is primarily going to come from 800G.

On the 1.6T program, we won a number of them. And we have one customer right now where we have visibility to that ramping towards the back end of next year. And so one of our customers will be really taking up their 1.6T awards, but then we anticipate further 1.6T ramps as we go into 2027. And so the portfolio is shifting to the higher-end technologies as we would have expected.

Jason can add on that.

Karl, I would say when you look at where we carved out this industry-leading position in networking, it started in 400-gig, and we were able to translate all of those engagements into 800-gig. And those engagements have been expanding incrementally to new opportunities, and we fully plan to translate all of our 800-gig engagements into 1.6T as well, and we're on track to do that. And liquid cooling plays a key role in those solutions, particularly on 800-gig and 1.6T.

Your next question comes from the line of Samik Chatterjee with JPMorgan.

And maybe if I can start with the digital native customer that you're expecting to ramp in 2027. And we've seen multiple sort of announcements from some of your partners around sort of the sizing here. But trying to sort of think about how should we think about the magnitude of the implications for you starting in 2027. Maybe if you can give us something to point us in the right direction of sizing relative to your enterprise business today? And do you have -- what are you thinking in terms of capacity to then sort of cater to that magnitude of the digital native customer ramp? And I have a quick follow-up.

Yes. Samik, it's Mandeep here. Thanks for the question. So we're very excited still about the engagement we have with this digital native customer. We are very actively engaged with them on the design cycle and that's going to continue as we go through next year.

Our plan of record right now is that we would not see mass production begin in 2027. And so when we've given an outlook of $16 billion for next year, that does not include any meaningful level of revenue coming from this digital native opportunities. The gate to that is really in terms of timing is really going to be around silicon availability. And so if silicon is available sooner for mass production, then we may be able to produce sooner.

Right now, our assumption is that we will receive samples in the middle of '26 and then again, go towards mass production in '27. From a capacity perspective, we are working with the customer very closely on where -- how we can support them both in Asia as well as in North America. When we are talking about 2% to 2.5% of CapEx for next year, that's inclusive of the capacity that we're going to need to deliver what we're already seeing in 2027.

Jason can add a little bit more.

Yes, Samik, I would say, with all the growth we're seeing across the portfolio, we're also excited about what I call a healthy competition on who will be our largest customer in the next 2 to 3 years.

So the way to think about it right now, Samik, is going to be that we believe it will be at least a few billion dollars in the first year, multiple billions of dollars, I'll say. One of the areas, of course, that we still need to line up on is the treatment of the silicon isn't included or not included and we're still having those conversations with our customer and our providers.

Okay. Got it. And a quick sort of follow-up on 2027 outlook. I know you're saying the growth momentum continues into 2027. And I didn't hear you explicitly say that -- so just wanted to confirm from everything you're telling us in terms of new program ramps in 2027, the growth acceleration in 2027 should be higher [ rated ] to the growth that you're forecasting for 2026, just with the digital native customer, the 1.6T ramps. Is that a fair statement?

Yes, of course. So I'll give you a framework on how to think about 2027. Obviously, we're not going to be giving numbers at this point. It's just too far out. But we are very confident right now on the demand profile that we're seeing and the awards that we've been receiving over the last 12 months or so, in many cases, don't have programs that even ramp until 2027.

That being said, it's probably 12 months too early to talk to you about what 2027 really is going to look like. The way I would just think about it right now is our CCS business grew by about 40% in 2024. It's on track to grow about 40% right now in 2025. And our outlook or guidance for next year is essentially implying about 40% growth again in CCS in 2026.

And so at this point, I think it's fair to continue to extrapolate that as you bring it forward. We do have many opportunities that could be -- that could accelerate that and could go above. So to your point, it could be through the digital native win that we're ramping as well as other really strong programs that we've won with some of our largest hyperscalers. But right now, we think at least 40% into 2027 is what we're -- we have visibility to.

And then just on the ATS side, ATS this year is going to be approximately flat or we said going into 2026, it's going to be low single digits. The growth should resume at a higher level as you go into 2027. And I think the way to think about that right now is high single digit.

Your next question comes from the line of Ruben Roy with Stifel.

Sorry about that I was on mute. Mandeep, maybe to follow-up on that just last topic. You talked about potential additional operating leverage beyond '26. I'm just wondering, how you're measuring the potential for operating leverage relative to this really strong revenue growth that you guys are seeing, especially as we went through some of the new design activity, kind of increasing design activity with your customers, rack levels, designs, et cetera. Just do you have some thoughts on longer-term operating leverage?

Yes. Thanks for the question. So we continue to see the benefits of both operating leverage as well as positive mix in our numbers, on track for about 7.4% operating margin at the company level for 2025, and we're guiding that, that can expand now going into 2026. We do continue to believe that there's opportunities for even more margin expansion. But again, I'm not giving formal numbers for '27 at this point.

When you look at our ATS business, the business has done very well on doing some selective pruning in order to really focus on the highest value engagement. And so we're really happy with the margin expansion that we've seen in ATS already. And we think that there's opportunities to continue to expand and get it above 6%, hopefully in the near to medium term.

On the CCS side, which is operating in the low 8s right now, what's working to our favor is the fact that we will continue to be seeing growth in networking, which are primarily our HPS products. And our HPS products are accretive to the company and accretive to CCS. And so as we see growth in that area, we will continue to see some margin upside.

That being said, we do continue to evaluate how we can support our customers on multiple areas such as doing complex rack integration work. And so sometimes that will be margin dilutive. And so we're always managing mix, but we think that there's a lot of opportunities for expansion.

Thanks for that detail, Mandeep. And if I could ask a question -- a follow-up question for Jason. A lot of discussion around scale up networking just in recent weeks, with a new standard announced, et cetera. You talked about a multibillion-dollar new market opportunity for Celestica, specific to scale up. And I'm wondering if you can maybe hash out a little bit around that opportunity relative to scale out. Do you have some sort of advantage as you discuss scale up with your customers, given how well you're doing on the scale-out switch side? And maybe just a little bit around the competitive environment, how you see that playing out over the next several years as you think about your scale-up opportunity.

Yes. Ruben, yes, I would say we're well positioned for the scale of opportunity, and that comes from incumbency, and I would say, capability. When you look at, as you mentioned, a lot of where we carved out this industry-leading position in 400G, it started largely and I would say, scale out. And now it's starting that capability, and that value proposition is very much applicable to scale up. We've talked about a large digital native where we provided a fully orchestrated rack and solution. I mean that's a great example of a great -- a significant scale-up opportunity. So I would say that we have a large and growing funnel of opportunities, and we're going to be very mindful about where we have our most strategic engagements as we continue to grow and look at taking share.

Your next question comes from the line of Tim Long from Barclays.

Yes, Two, if I could as well. First one on kind of HPS. I think this digital native is a good AI/ML win for digital native. So curious about the pipeline that you're talking about for other kind of compute-related opportunities. Could you just talk about that funnel and how we should think about new opportunities being HPS or not, number one?

And then number two, just back to that -- the networking piece. As you look at new customers outside your large [indiscernible], should we assume those are mostly SONiC related? Or do you see opportunities for other Neo clouds or others to maybe develop their own switching stack where -- and what are the competitive differentiations for you with SONiC versus other proprietary NOS?

Tim, Jason here. So on AI/ML compute, I mean you commented on it, I think digital native win that we've talked about would be a great example of where we've deployed our entire value proposition into a fully orchestrated solution, driving an AI/ML solution. We talked last year a bit about POCs that we're doing with silicon providers, and we talked about the AMD MI355 example, and that POC has garnered a lot of attention, I would say, in the industry.

And so we have a large and growing funnel of opportunities in AI/ML. But we're going to be very, I'd say, very focused on where we have the strongest strategic alignment and where we believe the program will be successful in the AI architecture and ecosystem. So growing funnel of opportunities, but we're going to be careful about where we engage and where we believe the adoption rates will be higher.

Yes. With regard to these opportunities, I think one of the things to consider is just the strong position that we have in networking and its applicability to these AI compute kind of opportunities. So there's a lot of things that transfer over the network connectivity, the signal performance, power, density, design, all those things are also very applicable and relevant in the AI space. So we continue to leverage our networking strength to win in new opportunities in the compute space.

With regards to software, most of our hyperscale customers drive their own NOS, but they rely on us to provide the key layers in the stack and have full testing and qualification capabilities of their software on our system. There's also more comprehensive choices that are emerging now, and our customers are evaluating those. We still have a full -- fully capable and broad software engineering team, and we're working to support many of our customers with these new software technologies. And we continue to support them at the firmware level in most of the networking and compute systems that we do today.

Your next question comes from the line of David Vogt with UBS.

So maybe for Rob or Mandeep, I want to unpack the CCS business for a second. Obviously, switching has been sort of the driver of the business the last 2 to 3 years. And you kind of talked about over the next several years, switching growth or maybe data center CapEx growth being kind of in the mid-20s. Are you sort of inferring that ultimately, the bigger driver over the next 3 years plus will be sort of the compute opportunity along with ancillary opportunities like optical as we think about '27 and '28? Just trying to get a sense for how you're thinking about the composition of product within broadly defined CCS going forward. And then I have a follow-up.

David, I'll start, and I'll ask Jason to chime in. We have a very strong position with the hyperscalers on networking across the board, given that position, we're looking to grow our share of wallet into other areas. And one of them, in particular, we do AI/ML compute. And with others, we're in several advanced conversations to expand our solutions to them, especially on the HPS front where we're not just build it, but we actually have some engineering and design content in supporting them.

Jason elaborated on a couple of those opportunities, and I'll turn it over to him for additional color.

David, so thinking about the CCS business overall, I mean, we've got a lot of strength in our hyperscaler digital native portfolio in networking, AI/ML, specifically on the custom side, there's incremental opportunities we've talked about at scale up as well as merchant AI/ML solutions. So there's a lot of growth, a lot of potential, a lot of funnel of opportunity there.

When you look at the value proposition that got us where we are, there's a lot of opportunity to take that and pivot into the very large enterprise space. And we're going to do that in a very disciplined way. I've talked about that in the past. We have a portfolio solutions business today where we have branded product. We have SONiC, we have Celestica SONiC offering that's enabling that. We have a growing services capability that's rounding out the capability that will allow us to play more effectively in enterprise as well as hyperscaler.

So we're effectively doubling down on our enterprise efforts. I recently just brought in Ganesha Rasiah. He's our Senior Vice President and General Manager of our Enterprise line of business, and he will be leading the charge as we chart our course on where and how we're going to double down in enterprise. And it's going to be underpinned by all of this value, the scale, this capability that we've established in our hyperscaler space and applying it to specific markets within enterprise to be successful.

Great. No, that's helpful. And maybe just maybe one more for Jason then. On the enterprise portfolio since you're talking about expanding capabilities and bringing in new talent looking for new opportunities, you did reference, I think, in the deck an opportunity for a new hyperscaler application in storage for '26. Can you kind of expand upon that, kind of what that actually is and maybe share what the customer is looking for and what you're bringing to that solution going forward?

Yes. Maybe I'll start, David, and then I'll ask Steve to weigh in as well.

I mean we do think storage is -- it's going to be an opportunity with AI. There's more and more data that is out there. And we're seeing it specifically, we've got some traction in the hyperscaler space on a specific program where there's a specific use case, I would say, that's being adopted, but we also think there's more opportunity for storage and enterprise as well. We've had a solid high-end storage business in enterprise for a long time. We're well positioned with the market leaders there. And I think storage is an opportunity as AI continues to deploy.

Yes. And this is Steve. I will just add to that, that we have had some success, as noted here with hyperscalers and providing custom storage solutions to them. We're being very selective about where we engage and finding areas where we think we can differentiate and bring value to our customers. And so it's a narrower scope today for us, but there are opportunities, and we intend to continue to build on the success that we've had there.

Your next question comes from the line of Thanos Moschopoulos with BMO Capital Markets.

Can you provide some color with respect to the growth that you're expecting in CCS outside of hyperscalers and digital natives, so like OEMs and other types of customers, what would your outlook be for that in '26 and beyond?

Yes, in terms of the incremental growth opportunities, we've talked about scale up is a very large market where we're getting a lot of traction. I would say, while our AI/ML business has been largely underpinned by the custom level solutions, there is a growing set of opportunities. We're getting a lot of traction in merchant-based AI/ML systems that represents a number of opportunities for us. And then I'd say fully orchestrated rack level solutions continues to be an opportunity as well as the services that we were wrapping around our solutions as well.

Specifically in terms of, I guess, OEM type customers and maybe enterprise campus type opportunities or other beyond hyperscalers, digital natives. Is that forecast to grow meaningfully in '26, or is the growth in '26 primarily driven by your core hyperscalers?

The growth is underpinned by our hyperscalers. They are leaning heavily into both switching as well as compute. But the rest of the portfolio is still growing as well. We have a large optical program that goes beyond the hyperscalers. We're seeing very nice growth in that area, and that product be used directly into data centers. We've announced that we are building a 1.6T switch with an OEM on their behalf. And so that's going to see some growth. But we do -- there is a very high level of growth coming from hyperscalers versus the others.

Your next question comes from the line of Steven Fox with Fox Advisors.

Just one question on the HPS business. I know you don't give margin -- specific margins on the business. But I was wondering, given all the programs you see in the future and how you may be vertically integrating more, sometimes, I guess, additive, sometimes dilutive to margins. How do you see the direction of HPS -- just the HPS business going and why? And of course, that would be excluding any kind of changes in your consignment activities like with the new program.

Yes. So we're very excited about what's happening in the HPS portfolio. We're on track this year to spend probably about $120 million on R&D. Next year, we're going to be increasing that by at least 50%. It could be as high as $200 million. And that's just reflective of the engagement that we're tied to.

Today, this year is probably going to be about a $5 billion portfolio. That $5 billion is the vast majority of it is switching. And in the switching scenario, it actually does include the silicon, as you know, so it's turnkey. And yet, we still make very good margins in this area, margins that are accretive to ATS right now, which is 8.3%. I know the question sometimes comes what's the exact number, but what we just say is it's accretive.

As we look at the portfolio going forward, we continue to see very strong growth on the networking side, but now we're starting to see compute come in as well. And so as compute comes in, especially as you think about this large digital native win, we've got to think through still on how sort of things can be provided. But today, our compute programs [indiscernible] to us. But overall, we are getting paid for the value that we're bringing forward on the engineering side.

Your next question comes from the line of Paul Treiber with RBC Capital Markets.

Paul has actually lowered his hand. So we will move on -- and your next question comes from the line of Robert Young with Canaccord Genuity.

The 40 basis points of margin expansion -- operating margin expansion in the 2026 guidance, just against all of the big jump in scale and some of the shift to higher-end networking technology and the software mix. It just seems a little bit conservative. And so relative to some of your networking peers, margins are lower. And so I was wondering, is pricing a strategic advantage for you? Or are there any headwinds to note? I think you already mentioned the fact that the full rack solution isn't ramping until 2027. But are there any other headwinds there to note to better put that 40 basis points expansion in the context?

Yes. So I'll start and then I'll let Jason or Steve talk about the commercial environment and our ability to capture share with price.

But essentially, what's happening right now is that when you look at the 7.8% next year, and again, you're putting 40% growth on the CCS business, maintaining the margins that CCS has and maintaining the margin that ATS has will yield that 40 basis points improvement. We are working towards expanding margins in both businesses, and we do believe that, that is an opportunity. It's early on in the year, and so this isn't that different than the approach that we've taken in previous years, which is we will guide margins in terms of where we are today, knowing that we are working on various levers to expand that. But I would say more to come as we go through 2026.

Rob, and I would just comment on where we're seeing the values where we're driving the differentiation from our competitors. It's largely a technology leadership, customization for optimization and then our advanced manufacturing processes and execution. I mean we've pivoted now that we're into platform solutions, we pivoted from a technology partner to a technology leader. We believe we were the first with a fully functioning 800-gig switch. We believe the same on 1.6T. Those are examples of technology leadership that our customers are relying on.

Secondly, the ability to optimize -- to customize these solutions for our customers' specific architectures for optimization in those workloads in those large language models, that continues to be a strong area of differentiation for us. And then the last piece would be the advanced manufacturing processes and capabilities, which I believe is often underestimated and undervalued. It's very difficult to take these very complex designs and put them through the new product development process and then ramp at scale into production, it's not easy to do. And those continue to be areas that our customers value.

This is Steve. I would just build on what Jason had said there. When we deliver this differentiated value, and we do it reliably and consistently over several different platforms or iterations of -- new iterations of same platforms, there's a lot of strengthening of our incumbency and our customers start to recognize the value of our solutions and we're less compelled to compete on price. And so that's a key part of sustaining and maintaining the margin trajectory that we have. And it's also a function of the opportunities that we choose to pursue. So we have a tremendous amount of opportunities in front of us. We're moving away from the more transactional engagements and focusing on those operations -- those opportunities where we can really differentiate, as Jason said.

And Rob, I would just add to that. There's -- in the -- every now and then, we'll see a competitor will -- I mean the competition is stiff and there's a lot of competitors coming in and we'll lead with price. And every now and then we'll see someone will chase a program on price only that 3 to 6 months later haven't come back due to challenges with execution and delivery.

That sounds great. Second question for me would be just on the shorter refresh cycle you noted in networking and maybe the quicker move to 1.6 to 3.2. Does that make it harder for new entrants? I would assume that in existing data center deployments, it's very hard to dislodge Celestica. But maybe if you could talk about that as it relates to greenfield and new build, and I'll pass.

Yes. Maybe, Rob, I'll start, and then I'll hand it over to Steve. So as you -- first of all, technology, the generations are getting quicker and it's getting faster. So if you're behind, they're going to have a harder time keeping up. So we saw a lot of folks struggle in 400 as we went into 800. As you go in from 800 to 1.6, it's getting faster and it's getting harder. So if you weren't optimized around 800, you're going to really struggle to get into 1.6T and the same applies to 3.2, et cetera, et cetera. So we're well up the curve. We're a technology leader in the space. We've been making significant investments. We've been building talent for many years to get to where we are, and we don't plan on slowing down.

Yes. This is Steve. Just to build on what Jason has said there, our recent experience with 1.6T is that we've had demonstrated very strong performance here in delivering solutions from the initial receipt of silicon to complete functional power on the systems. We've done it in days. And I think Broadcom knows would acknowledge that typically with some of our OEM and ODM competitors, they measure that achievement in terms of weeks. And so I think that what we've talked about, the carryover from one iteration to the next is just proven to be true for us as we support our customers.

Your next question comes from the line of Paul Treiber with RBC Capital Markets.

Just a question on the long-term visibility that you're getting from customers at this point. Are you seeing it reflected in the program wins? Are there either explicit volume commitments? Or are there other commitments or the nature of the contracts that allow you to have that longer-term visibility that maybe you didn't have several years ago?

This is Steve. Yes, I think it's a good question. I think that we'd like to have as much visibility as we can to the future of these programs. But we do have some comfort in that we continue to see awards come to us for the duration of the program and the follow-on next generation of those programs tend to be awarded to us as well. So -- so we do have longer-term visibility of the programs we currently have and what's coming next down the funnel. So overall, very good.

Yes. Just maybe as an example on the compute program that we have right now, which is going to be very healthy in 2026, and it's ramping very nicely right now. That is already in -- we've already won the follow-on programs for that program to the point where the silicon hasn't even been finalized yet because it's going to be on the next-generation silicon. And so we see those ramping into 2027.

And then we've talked about the digital native win as well, which is a program award that will be ramping in '27. And then our R&D efforts continue to be working on the next generation of products as well, which we know will get adopted eventually by the market. We're already working on 3.2T. And while we don't expect it to be mass production until maybe 2028, we would anticipate that when that migration happens from 1.6 to 3.2, that we're going to be in a full position to win that share.

Yes. And Paul, I would just add to that. Steve mentioned forecast visibility between 12 and 15 months and in some cases, beyond. I mean there are certain programs that have very specific capability requirements where we're talking even beyond that. So as we look at the power requirements, the capacity that will be required beyond what I'd call an extended forecast outlook, we're in deep conversations with capacity planning, power planning well beyond, I'd say, the '26, '27 time frame that we're accounting for as we make our investments and our expansion plans.

And a follow-up question. The -- to what degree are you shaping -- proactively shaping the portfolio, either disengaging on less strategic programs? And then on the strategic programs. Are there any metrics you can share in terms of like win rates or success on rebidding the next generation of those contracts?

Paul, thanks for the question. This is Todd within ATS. Yes, I would say we are just conducting a comprehensive review really on an ongoing basis of our portfolio doubling down, as I said in my comments, on the larger Tier 1 customers and then using this opportunity really to take out or exit reshape, if you will, margin-dilutive customers. That's why you're seeing the improvement in margin in ATS this year.

And then we've had a number of smaller customers where candidly, the climb is not worth the view in terms of just the effort to support their businesses. They're nonstrategic. In some cases, they're tied to our smart energy portfolio, which given the one big beautiful bill and the loss of tax incentives and the change in dynamics around clean energy are impacting their end demand.

So we're using this opportunity then to disengage and exit from those smaller customers, nonstrategic customers, margin-dilutive customers really to strengthen the ATS portfolio and to improve our overall margin profile as well as our growth going forward.

Paul. From a CCS perspective, we are, from a hyperscaler digital native perspective in a great spot strategically. We feel very good about that. And then enterprise, largely the same. There is a smaller customer where we are no longer strategically aligned, but it's not material to the overall business. And I'd say overall, from a CCS perspective, we feel great about where the portfolio is.

Your next question comes from the line of Todd Coupland with CIBC.

Great. I had a question on the switching business. I'm wondering with your largest customers, are you single sourced or are there dual source suppliers in any of those?

This is Steve. With the majority of our largest customers, we tend to be the preferred supplier when it comes to new technology. And so we're often exclusive for some period of time through the development of the product through NPI and then to ramp.

As Jason mentioned earlier, we do have excursions from time to time where our customers will look at dual sourcing or multi-sourcing maybe for business continuity purposes or maybe to chase a lower price for some period of time. But we tend to see a lot of those come back to us. While we maintain the preferred position on new products, we see some of the next-generation products come back to us as well for an exclusive period again through the development and through NPI and RAN. So that's been a pattern that we've seen repeat with most of our hyperscale customers over the various product transitions.

And I just wanted to circle back to the 1.6. You were quoting some market share stats earlier in the presentation. Can you just remind us what that win rate implies for ports, I guess, through '26 and '27 on the 1.6?

Yes. What I would say, Todd, is, as I mentioned earlier, where we've had our engagements in 800-gig, we're on track to have those engagements in 1.6T, and there's incremental opportunity beyond that in the scale-ups market in particular. And as I noted, we have a healthy funnel. We're excited about it, and we believe it's going to be a big growth driver for us.

Yes. I mean, Todd, we're winning our disproportionate amount of share as the technologies become more advanced. And so some of the materials in the slide we were highlighting when you look at the Ethernet switch market share, we're above 50% this year. And last year, it was 40%. As there is further deployment of 800G switches and as 1.6 starts to get delivered, we would anticipate that, that will continue to be positive for us. But we are -- we do continue to win the funnel of programs, which is what we're [indiscernible].

This is Steve. I can't give you 4 counts, but I can tell you, we have 10 programs currently underway and 1.6T. And we've had a significant share win with a number of customers on 1.6T.

Your final question comes from the line of Jesse Pytlak with Cormark Securities.

Just on your optical programs, can you speak to the breadth of customers that you're engaged with on these? And are these programs commonly becoming bundled with switching programs at all?

Yes, Jesse, so we have a few primary optical customers where we have deep engagements and we're making POCs and investments in that space. And there is a strong correlation between optical and networking. And we think when you look at things like CPO technology as an example, we think we'll start to see some deployments in 1.6T, and we really think we'll start to see more CPO ramp in 3.2T as an example.

Yes. This is Steve. I would just add, as Jason mentioned, the co-package optic outlook. We still -- we do see that it's going to be a dual existence for some period of time. So pluggables won't go away, but there will be a hybrid deployment of different strategies around co-packaged optics. And many of the optical capabilities that we're developing today will be very applicable when it comes to embedded or co-packaged optics in the future, which designs.

And there are no further questions at this time. I will turn the call back over to Rob Mionis, CEO, for closing remarks.

Thank you, and thank you all for your continued support. We're pleased with the results to date and our continued momentum into Q4 and into 2026 and beyond. We're also looking forward to seeing you later on this afternoon at our events luncheon. Thank you again, and have a wonderful day.

And that does conclude today's call. Thank you all for attending. You may now disconnect.