Rambus Inc. Aktie

Welcome to the Rambus First Quarter Fiscal Year 2026 Earnings Conference Call. [Operator Instructions] As a reminder, this conference call is being recorded.

I would now like to turn the conference over to John Allen, Interim Chief Financial Officer. You may begin your conference.

Thank you, operator, and welcome to the Rambus first quarter 2026 results conference call. I am John Allen, Interim Chief Financial Officer at Rambus. And on the call with me today is Luc Seraphin, our CEO. The press release for the results that we will be discussing today has been filed with the SEC on Form 8-K. We are webcasting this call along with the slides that we will reference during portions of today's call. A replay of this call can be accessed on our website beginning today at 5:00 p.m. Pacific Time.

Our discussion today will contain forward-looking statements, including our expectations regarding projected financial results, financial prospects, market growth, demand for our solutions, other market factors, including reflections of the geopolitical and macroeconomic environment and the effects of ASC 606 and reported revenue, among other items. These statements are subject to risks and uncertainties that may be discussed during this call and are more fully described in the documents we filed with the SEC including our 8-Ks, 10-Qs and 10-Ks. These forward-looking statements may differ materially from our actual results, and we are under no obligation to update these statements. In an effort to provide greater clarity in the financials, we are using both GAAP and non-GAAP financial presentations in both our press release and on this call. A reconciliation of these non-GAAP financials to the most directly comparable GAAP measures have been included in our press release, in our slide presentation and on our website at rambus.com on the Investor Relations page under Financial Releases.

In addition, we will continue to provide operational metrics such as licensing billings to give our investors better insight into our operational performance. The order of our call today will be as follows: Luc will start with an overview of the business. I will discuss our financial results, and then we will end with Q&A. I will now turn the call over to Luc to provide an overview of the quarter.

Luc?

Good afternoon, everyone, and thank you for joining us. We opened 2026 with a strong first quarter meeting our financial targets and broadening our portfolio to address the accelerating demands of AI. The quarter reflects solid momentum as we execute against our road map to support long-term profitable growth for the company. This is an exciting time for Rambus, and we are well positioned to capitalize on the market trends in the data center and AI. For decades, we have developed foundational technologies and solutions across a wide range of memory and interconnects. That heritage positions us well as systems become more diverse, memory dependent and performance-driven. To give more context, there are several market and technology trends playing out across the data center and AI that continue to work in our favor. As AI adoption accelerates and inference use cases expand, workloads are becoming more persistent and context rich. And performance is increasingly defined by how efficiently data can be stored, accessed, moved and secured. To support these workloads, AI infrastructure is becoming more complex and heterogeneous, combining a mix of traditional and AI server platforms to support orchestration, data management and real-time execution at scale.

At the same time, the expansion of inference and particularly agent AI with continuous reasoning and multistep workflows is driving more always on activity and placing even greater demands on memory capacity, bandwidth, latency and power efficiency. Together, these trends are driving new memory and connectivity architectures to support purpose-built solutions across a wider range of use cases and form factors. This increases our opportunities for rich a chip content and broader adoption of our industry-leading IP, reinforcing our position for sustainable long-term growth.

Now let me turn to our quarterly results, starting with our Chip business. Our performance reflects strong execution and ongoing leadership in our core DDR5 LCD chips. We delivered product revenue of $88 million in Q1, in line with our guidance and up 15% year-over-year. Looking ahead, we expect to deliver double-digit product revenue growth in the second quarter. We continue to see increasing customer adoption of new products and remain well positioned to support the ramp of next-generation platforms as they enter the market. We continue to execute on our strategy of delivering comprehensive industry-leading chip solutions to address growing customer and market requirements. As I mentioned in my opening remarks, we recently expanded our product portfolio with the introduction of our chipset for Gen X standard LPDDR5 SOCAMM2 building on the same signal and power integrity expertise we have applied across multiple generations of DDR. This chipset is the first offering in our road map of LPDDR based server module solutions and includes new voltage regulators as well as the SPD hub support reliable, power efficient server class operation.

As part of that road map, we are actively working with industry partners on the definition and development of LPDDR6-based SOCAMM2 solutions, which would offer a natural upgrade path for future generation AI platforms. As AI server architecture is diversified to address varying performance, power efficiency and form factor requirements, some platforms are now leveraging LPDDR-based memory. While SP memory offers attractive power characteristics, it was originally designed for mobile environments with very short signal paths and tight power margins, making reliable deployment in service systems inherently challenging. SOCAMM2 addresses these limitations through a compact CPU proximate module architecture with optimized signal routing and localized fire management to enable LPDDR modules to operate in server environments. The Rambus SOCAMM2 chipset enables power efficient, reliable operation of up to 9.6 gigabits per second in a compact module form factor. As LP base server modules scale to higher speeds and bandwidth in future generations, they will require increasingly sophisticated interface power and control functionality. This progression is similar to what we have seen in DDR-based-server modules and reinforces our opportunity to extend our road map of high-value chip content across memory types in the future.

As I mentioned previously, the ongoing expansion of AI is driving demand for a broader range of memory types and form factor. To meet these needs, we continue to build on our leadership solutions in DDR5 including chipsets for RD and MRM and selectively expand our road map of novel solutions as they begin to play a complementary role in heterogenous systems. With active engagements across customers and ecosystem partners, we are helping shape next-generation server modules, reinforcing the opportunity for richer chip content and sustain growth.

Turning now to Silicon IP. We saw strong customer traction in the first quarter with continued design wins at Tier 1 companies and growing engagement across our portfolio. We remain focused on delivering industry-leading premium IP that enables differentiated solutions for AI in the data center, including accelerators and networking chips across a wide range of architectures. This increasing momentum for custom silicon in AI, especially among hyperscalers as they tailor hardware to their own software stacks and deployment needs, optimizing for performance, power efficiency and total cost at scale. This is driving an accelerating pace of design and expanding demand for value-added IP to support memory bandwidth, advanced connectivity and security.

During the quarter, we saw growing traction for our value-added PCA retirement and Switch IP to support increasingly complex AI systems across scale-up and scale-out environments. We also expanded our memory IP portfolio with the introduction of the industry's fastest HB and 4E controller, setting a new benchmark for AI accelerator memory throughput. In addition, we launched a new network security engine designed for Ultra Ethernet to protect distributed AI clusters. All of these IP offerings are in great demand and further strengthen our position as a critical enabler of next-generation compute and connectivity solutions for AI infrastructure.

In summary, we executed well in the first quarter. We delivered solid results and expanded our offerings to both chips and IP to extend our leadership in our core markets. As we look ahead, Rambus is well positioned to capitalize on the mega trends in data center and AI. Our sustained technology leadership, disciplined execution and increasing traction across our portfolio of leadership products will continue to fuel our results.

With that, we expect strong growth in 2026, and I'm confident in our long-term trajectory. As always, I want to thank our customers, partners and employees for their continued trust and support.

Now I'll turn the call over to John to walk through the financials. John?

Thank you, Luc. I'd like to begin with a summary of our financial results for the first quarter on Slide 3. We delivered first quarter revenue and earnings in line with our guidance, with solid contributions from each of our diversified businesses. We also continued our strong track record of cash generation. This performance reflects the continued strength in our business model. Our strong balance sheet and disciplined capital allocation enabled us to invest in growth initiatives while returning value to shareholders.

Let me now provide you a summary of our non-GAAP income statement on Slide 5. Revenue for the first quarter was $180.2 million, which was in line with our expectations. Royalty revenue was $69.6 million, while licensing billings were $70.8 million. The difference between licensing billings and royalty revenue mainly relates to timing as we do not always recognize revenue in the same quarter as we bill our customers. Product revenue was $88 million, representing 15% year-over-year growth, driven by continued strength in DDR5 products and ramping new project contributions. Contract and other revenue was $22.6 million consisting predominantly of silicon IP. As a reminder, only a portion of our silicon IP revenue is reflected in contract and other revenue, and the remaining portion is reported in royalty revenue as well as in licensing billings.

Total operating costs, including cost of goods sold for the quarter were $104.6 million. Operating expenses of $69.9 million were up sequentially due to seasonal payroll-related taxes in connection with equity vesting. Interest and other income for the quarter was $6.9 million. Using an assumed flat tax rate of 16% for non-GAAP pretax income. Non-GAAP net income for the quarter was $69.3 million.

Now let me turn to the balance sheet details on Slide 6. We ended the quarter with cash, cash equivalents and marketable securities totaling $786 million, up $24 million from Q4 2025, with strong operating cash of $83 million, partially offset by $38 million in taxes paid on equity vesting and $17 million in capital expenditures. We increased our inventory balance by $14 million during the quarter and expect to continue building inventory strategically in the second quarter. Our strong balance sheet gives us the flexibility to increase inventory to support our product revenue growth and manage through potential supply chain constraints. First quarter depreciation expense was $8.5 million. Free cash flow in the quarter was $66.3 million.

Let me now review our non-GAAP outlook for the second quarter on Slide 7. As a reminder, the forward-looking guidance reflects our best estimates at this time, and our actual results could differ materially from what I'm about to review. In addition to the non-GAAP financial outlook under ASC 606. We also provide information on licensing billings, which is an operational metric that reflects amounts invoiced to our licensing customers during the period adjusted for certain differences. We expect revenue in the second quarter to be between $192 million and $198 million. We expect product revenue to be between $95 million and $101 million a sequential increase of 11% at the midpoint of guidance. We expect royalty revenue to be between $72 million and $78 million and licensing billings between $76 million and $82 million. We expect Q2 non-GAAP total operating costs, which includes cost of sales to be between $114 million and $110 million. We expect Q2 capital expenditures to be approximately $14 million. Non-GAAP operating results for the second quarter are expected to be between a profit of $78 million and $88 million. For non-GAAP interest and other income and expense, we expect $7 million of interest income. We expect non-GAAP tax expenses to be between $13.6 million and $15.2 million in Q2. We expect Q2 share count to be 110 million diluted shares outstanding. Overall, we anticipate Q2 non-GAAP earnings per share to range between $0.65 and $0.73.

Let me finish with a summary on Slide 8. In closing, we delivered solid results in line with our objectives, driving ongoing profitability and cash generation. Our diversified portfolio remains a core strength with each of the businesses contributing meaningfully to our performance. Our patent licensing business continues to deliver consistent, predictable performance supported by the long-term agreements we have in place. Our silicon IP business is well positioned, driven by critical interconnect and security technologies addressing the accelerating demand for AI solutions. Our product business grew 15% year-over-year and is poised for sequential growth in the second quarter. We remain focused on delivering long-term shareholder value with year-over-year revenue growth in 2026.

Before I open the call up to Q&A, I would like to thank our employees for their continued teamwork and execution. With that, I'll turn the call back to our operator to begin Q&A. Could we have our first question?

[Operator Instructions] Your first question comes from the line of Kevin Garrigan with Jefferies.

Can you just help us think about your product revenue into the June quarter. So last quarter, you discussed the low double-digit revenue impact from a onetime OSAT issue. And I think we may have been expecting a larger sequential increase for June just kind of given the strong -- how strong demand has been. So can you just walk us through the drivers for the June quarter product revenue, and why the recovery might be a little bit more measured?

Thank you, Kevin. Yes, sure. So the first thing I would say is that the issue that we had talked about in the prior call is behind us. Everything is being resolved, and it's a question now for us to restabilize the supply chain, which we are doing, and we see a normalization of that supply chain. So it is behind us. And the revenue for Q2 is guided at 11% over Q1. So that's the right trajectory. And we continue to expect to grow sequentially after that in an environment where our footprint continues to be very strong. I mentioned in earlier call that it was an older generation of of DDR5. If the market is transitioning from Gen 2 to Gen 3, which is a good catalyst for us. So I would say we met or we guide to double digit in the second quarter. We met what we said we would meet on the operational strain in Q1, and we will continue to grow sequentially quarters after that. We don't see any issue with the demand, and we don't see any more issues with the quality issue that we had in Q1. So we feel quite confident for the rest of the year as the market moves from Gen 2 to Gen 3.

Okay. Great. And then just as a follow-up on your LPDDR5 SOCAMM2 server module chipset, when would you expect to start seeing revenue from this chipset? And what kind of milestones should we watch to gauge traction.

I would see this as having a very good strategic impact at this point in time. The financial impact in the short run this year is going to be very minimal, just because the volumes are very small for this type of solutions. As a reminder, it only addresses a very small portion of the AI workloads. The volumes are small, the content is small as well. But strategically, I wouldn't put it in the model for 2026, but it's strategically very, very important because there is a trend to look at LPDDR in the server environment in the long run. LPDDR still has issues to address the server requirements, but it also has traction. It has benefits. So we see this as a stepping stone for us. It builds on the fact that over the last few years, we have developed our product line as chipset. So we have the whole chipset for the SOCAMM2. We have our own teams for power management development, and these are the two new chips that we are proposing for this solution. So we see this as a stepping stone. It allows us to engage with us with other AI players in the industry. And we are working on next generation as well. But I don't think that the financial impact is going to be significant this year, just given the volumes.

Your next question comes from the line of Tristan Gerra with Baird.

A quarter ago, you highlighted shortages and sounded a little bit maybe not cautious, but muted on the growth opportunity, and you provided a fairly muted data center unit forecast. How are shortages for component potentially impacting your revenue this year, what are you seeing that's different now than a quarter ago? And given the outlook for DRAM to remain very tight next year, how should we look at your product revenue growth? And specifically, your RCD growth with -- excluding the new product layers that will be adding on to that from a year-over-year growth standpoint. So in other words, would you expect the same type of growth next year, year-over-year versus this year? And I understand you're not guiding for next year, but just wanted to get a bit more color on what you see in the market that potentially could constrain on your growth. And clearly, that's an issue for a lot of other companies as well.

Yes. Thank you, Tristan. First of all, let me say a few words about the demand. We do see demand continue to grow for standard servers, which is good for us with agent in particular. We expect the server market to grow faster this year than last year. We model it at low double-digit growth because despite the excitement around AI, there's also a large portion of the server market, but that is not AI-related. But we do see demand growing on the service side, which is really a good catalyst for us. But as we said last quarter, we're watching the situation with supply, especially on the back end. Certainly, since last quarter, the situation has not improved. We're working with our suppliers. But the lead times are long. And there is tension on the back end. So we take this into account when we forecast our business. This is one factor. The other factor that affects or that comes into play when we forecast is the timing of launch of new platforms in the market. As you know, it's been the case in the past for us, the launch of our new products depends on the launch of new platforms in the market, and that's the dependency that we have. So we don't see the situation as materially different than what we saw in Q1. But from a supply standpoint, things have not improved. And really expect the supply situation to be tight going into 2027 as well when we talk to industry players.

Okay. That's useful. And then as my follow-up question. Any additional color on the MRDM opportunity? I know you've talked in the past about some very neutral shipments late this year. Specifically within Fencing. Any additional color as to where it could be in terms of revenue in '27. I think you've talked in the past about your expectation that you probably fully realize that $600 million SAM for MRDIMM by '28. So what should we be looking at for next year kind of in between? And what's really driving that? What's going to be driving the demand? Is it going to be mostly influencing? And any additional color you may have beyond what you've said in the past on customer interest for this technology, and where it's going to ramp?

Thank you, Tristan. First, we continue to make progress in the launch of these products and the interaction with our customers on this MRDIMM. We're excited by the opportunity. For the reasons we've always talked about larger capacity, larger bandwidth in the same ecosystem, so the adoption is easier. The main, I would say, factor affecting the ramp of our MRDIMM is going to be the timing of the launch of the platforms from Intel and AMD in particular, where they do have this capability attached in the next-generation platform. So we continue to see the ramp starting in 2027 in earnest and a SAM at this point in time, which we still value at about $600 million as I keep saying, the SAM once the products are in the market and we get feedback and the market gives us feedback, we're going to have a much better view of that SAM. But at this point in time, this is the right number to keep in mind.

Your next question comes from the line of Aaron Rakers with Wells Fargo.

I guess kind of just building off that last question first. When you kind of think about the $600 million incremental opportunity around MRDIMM, I can appreciate that there's a lot of unknown variables at this point. But I'm just curious as you rolled up that expectation what assumption are you making in terms of attach rate on AMD Venice and Diamond Rapids at this point? And how might that evolve? I mean, I would assume that you're being rather conservative on that attach rate at this point? And then also on that, how do you see CSL starting to play out?

At this point in time, we modeled a lower attach rate. As I said, until my experience until the product is in the market, it's hard to make those models more significant. There are a lot of variables coming into play. As we just said, the most important one is the timing of rollout of these platforms in the market. There's also the whole situation with DRAM pricing and the prices of modules and how our customers' customers are going to make the decisions between the combination of modules they want to have in the current memory cycle environment. So we model, I would say, a conservative percentage for MRDIMM at this point in time. But ramp will start when the platforms ramp in the market, and that's when we're going to have a better view.

And any thoughts on CSL?

Oh, sorry, I missed the second part of your question, sorry, Aaron. CSL, we do have very good traction on our IP business. We are not planning to launch a semiconductor product at this point in time. We do have this on our shelves, if you wish, as we designed one a couple of years ago. But we do see the -- with agentic AI, we do see demand for standard DIMMs and MRDIMMs as being the main benefactors of that. And that's where we will continue to focus our attention.

Yes. And then final -- one final quick one. When we -- when you guys talk about the opportunity to grow sequentially in product revenue into the back half of the calendar year, I'm curious if you were asked about seasonality in the second half versus first half, if there's anything that changes your views maybe relative to the last couple of years on -- I think you've seen some decent growth second half versus first half.

Yes. Thanks, Arun. That's a good observation. We actually do see second half shaping out slightly different than the first half, better growth in the second half. A lot of times it had to do with the launch of new platforms. They typically beat the market. If they are on time in the second half of the year, and that's where you have more products there. But even if you look at the first half of this year at the midpoint of our guidance in Q2, and you look at the first half of last year, we're still growing close to 18%. So the first half, despite our issue in Q1 is still much higher than the first half of last year. And we believe the second half is going to show growth. We do see some seasonality. And typically, our second half is stronger than our first half.

Your next question comes from the line of Gary Mobley with Loop Capital.

If I take the sum of your license billing in your contract and other revenue in the first half of this year for the results in the guide and compare that to the same period last year. It looks like you're generating some abnormally strong growth. Is that due to any sort of variance in the patent licensing? Or should I take this to mean that your silicon IP business might actually be running north of $150 million annually right now.

So thanks, Gary. We can see some quarter-to-quarter variations in these two categories just for the nature of the business. I would say that underlying this. We see very good traction on our silicon IP business. Actually, AI has an impact on our silicon IP business, which is also very positive as people who develop custom solutions for AI by looking for new interfaces and new security solutions like the ones I mentioned in the prepared remarks. So we do have very good traction on the silicon IP business, and we continue to expect this business to grow 10% to 15% a year based on that. Our other business, our patent licensing business, it can also be changing from quarter-to-quarter. We do renew agreements on a regular basis. And sometimes, these agreements are structured in different ways depending on the customers and what they want to do. So we have some strong quarters, some quarters that are not too good. But on average, this business continues to be stable at $200 million, $220 million. So I would say I would not pay too much attention on the quarterly split on these revenues. But the fundamentals are really, really good. What I would add to this is, if you look at our patent licensing business, our silicon IP business or our product business, they all benefit from what's happening in the memory subsystem area. They all benefit from AI and the move from the move from to AI inference. So -- and that gives strength to our results. And when we have a challenge like we had last quarter on the product line, then we have these two other product lines also that that allow us to meet our numbers.

Okay. Just my follow-up one to ask about CPU roles in AI optimized servers. I think there's been a lot more noise recently indicating a higher ratio of CPUs to GPUs in AI-optimized servers driven by genetic workloads. And you sort of hinted to that. To put this into a question, I'm curious if we moved to a point in time where we might see a 1:1 ratio CPU to GPU. Does this alter your view on the growth rate of your SAM for your product revenue or the size of it?

So we are excited with where the market is evolving with agentic AI and Inference. If you look at the types of architectures, software architectures, hardware architectures that inference requires then you clearly see that the ratio between CPUs and GPUs and is changing in favor of CPUs. So overall, that's a very good thing for us. It's just coming from the nature of what inferences or what agentic AI is. So that's a good thing for us. Is it going to be a one-on-one very difficult to say at this point in time? Everyone is trying to optimize now the memory subsystems. Everyone is trying to use HBM, where it's really good, huge LPDDR, where it's really good. and use DDR and MRDIMMs where it's really good. And I would say that DDR and MRDIMMs will continue to be the workforce of these inference AI solutions. But the fact that all of these systems starts to coexist, HBM, DDR, LPDDR, is really good. They all try to resolve a different part of the AI workload and this plays to our strengths because this is what we've been doing forever at Rambus. But I would say that the move to AI inference and the move to Agent I will change the ratio in favor of CPUs. And that's good for us.

Your next question comes from the line of Sebastian Naji with William Blair.

Maybe my first question, I wanted to ask about the new SOCAMM products that you announced last week. Could you maybe just comment on what Rambus' dollar content looks like for each SOCAMM module? Just across the different voltage regulators and the SPD Hub, any unit economics you can give us?

Given the current competitive environment, let's stay away from giving pricing on these things. But I would say that the content on a scan from the standpoint of Rambus, we have voltage regulators and an SPD Hub. So the content is minimum. This is what I was saying earlier on 1 of the questions. I do believe that this is strategically important for us because in the long run, LPDDR may play a larger role, especially in next-generation LPDDR solutions in the data center. But from a content standpoint, it stays minimal and the volume stays minimal. I will leave it there.

3

Okay, okay. That's fair. And maybe just turning back to the RDIMMs. Could we get an update on the progress you're seeing with companion chips how much revenue came from those companion chips in Q1? And then maybe just relatedly, how important is it for your silicon customers that they have all of these DIM components bundled together coming from 1 provider versus having to put these together from different providers.

Yes. Thank you. The -- John, go ahead.

Sure. The newer product, Sebastien, they're contributing low double-digit percent of our total product revenue during the first quarter. We would expect it to be roughly the same in the second quarter as we see some growth in the overall revenue contribution from that part of our business.

Yes. And what I would add to this is that we -- this is steady growth quarter-over-quarter. You saw this in 2025, every quarter. we had a slightly higher percentage. We continue to do that. And we expect to continue to do that for the second half of the year with this. And we expect maybe to exit the year at mid-double digit of product revenues on -- coming from our new chips. So to your other question, it is becoming more and more important for customers to have the whole ship set from one supplier, especially as the performance requirements increase. And the reason has to do with interactivity, making sure that all of these chips on a module well together at very, very high speed and very, very harsh environment is becoming more and more difficult to achieve. And that's why our customers request us to have the whole solution and to help them go through these generational changes.

Your next question comes from the line of Kevin Cassidy with Rosenblatt Securities.

The -- during the quarter, as you're building inventory, Were there any orders that you had to leave on the table that you weren't able to book because you didn't have the inventory there may be some upside surprise.

No. We've not been in that situation, but there are a few market dynamics that we have to anticipate. One is, as I said earlier, we do see supply tightening, especially on the back end. So we want to make sure that we have -- if that situation continues, we have enough supply supply our customers. The second thing that is happening is that the SaaS transitions between generation. And you remember, we were talking about Generation 1 moving to Generation 2, we indicated in the last call, that Generation 3 is ramping very, very fast. So we want to make sure that on this new generation of products, we also have enough inventory because the ramps on the customer side can be quite steep, and we just don't want to lease them.

Okay. I understand it. And maybe even when you're using your balance sheet to build more inventory, it's when Intel reported, they said they even were able to ship some previously written down inventory. It seems like the demand for CPU is so strong and also DRAM that maybe older generations are -- will get a little bit of a revival. Is that anything possible? Or -- it sounds like you're saying everything shifting to Gen 3 very quickly.

But from a demand standpoint, it's certainly the bulk of the demand for DDR products is shifting to Gen 3. But what you're describing in terms of using inventory of all products to serve demand is something that we continuously do and look at. That's part of our inventory management processes.

Your next question comes from the line of Mehdi Hosseini with SGI.

This is Bastian filling in for Mehdi. My first question is on LPDDR SOCAMM2 chipset. Would you mind clarifying the content of the chipset. It seems that the solution consists of 1 SPD and 3 voltage regulators. Do you expect to add any PMIC content there? And what does the pricing look like of the SPD and voltage regulator relative to the DDR DIMM, and I have a follow-up.

Sure. So yes, on the SOCAMM solution, we have 1 SPD hub and 2 types of voltage regulators, 3 voltage regulators in total, the 2 types. 112 amp regulator and 2, 3 amp regulators. So that's the content. So as I said, the content is minimal. You're talking about PMIC. There's no power management IC per se. That function is done by the voltage regulators that in this generation of product. But the way -- that's why we say it's very strategic for us. The way we look at this is that when LPDDR 6 is available, that LP memory already will offer even more speed and even more power capabilities, then it will require possibly more complex chips for power management, and we will work on those. And one can imagine as well that as the market evolves in the longer run, the market will probably need as well as the equivalent of our CDs in the long run. And this fall exactly in our strategy, and that's why I'm talking about the stepping stone. We want to make sure that we are early in these new technologies. They do not cannibalize the old technology. They are complementary to them. And in the long run, they have the potential to grow quite nicely, and they build on strengths that we have, which has to do with Signal Integrity and power integrity. Now in the short run, for the SOCAMM2 and LPDDR5x, as I said, the volumes and the content -- the dollar content is going to be very low, but that's a very interesting and strategic stepping stone for us in that area.

That's really helpful. And I guess my second question is on DDR5. How should we think about the timing of the ramp of Gen 4 and Gen 5 as they go to higher volume manufacturing?

So Gen 4 is going to start to ramp this year, but Gen 4 is kind of a niche generation, if you wish. It doesn't have the same traction as Gen 1, Gen 2, Gen 3, or Gen 5. I think everyone is now waiting for Gen 5. We're going to start shipping products that correspond to Gen 5 towards the end of the year. But just like for the Gen 5 is completely dependent on the timing of the ramps of the next-generation platform of Intel and AMD. This is where they're going to be adopted. And that's why we do see initial volumes this year but the bulk of the volume just like for MRD is going to start in 2027.

Your next question comes from the line of Mark Lipacis with Evercore ISI.

Question on the DIM attach rate. Is it different for CPUs used to perform orchestration and agent versus CPUs used in standard servers versus CPUs that might be put next to the GPUs and the HPs and the custom ASICs, should we think about the attach rates differently for these?

It's a very good question. Very difficult question also, Mark. I would say that the way we look at it is if you look at insurance and agent the functions that have to be performed by these standard CPUs are are closer to standard CPUs. I think the highest attach rate that you would find is really close to the GPUs, HBN platforms. That's where you have the heaviest load, if you wish, for the CPUs. So that's how, at this point in time, I would compare it. I would say if you take a DGX box with GPUs and then the CPU there are the CPUs that use the most memory in terms of capacity and bandwidth. I would say that when you go to inference, then it's probably a little less, but it's difficult for us at this in time to model that.

Sorry, I guess my phone dropped. I don't know if my question came through. But Luc, I was wondering is -- should we think about the DIMM attach rate differently for CPUs that would be used in orchestration for agent versus CPUs used in standard servers versus CPUs that are used for inferencing that get put next to the GPUs and the ASIC and the XPUs. Is there a different density there for the DIMMs?

So it's a very good question, Mark, but a very difficult question to answer. I would say the way we look at it at this point in time is that the highest of memory capacity and bandwidth really resides close to the GPUs and this GPU HBM clusters, if you wish. That's where you have the most need for very high capacity and very high bandwidth, which, on average, could be higher than what we found in in France and other solutions. But we have not modeled that at this point in time. It's hard to model. But we do see in aggregate the fact that inference is being added to training as a very good -- there are good traction for the use of standard beams or MRD in general. The attach rate is difficult to model at this point in time.

Got you. Okay. That's fair enough. And then the tightness in the back end that you're noticing is this -- do you know or can you explain what the cause of that is? Is that because of the idea that a lot of the back end happens in Southeast Asia, and they procure a lot of energy from the Mid East. Is that it, or is it capacity? Is it more like just the whole industry is in a great recovery time and the capacities utilization rates are really ticking up. Do you have a sense of the cause of the tightness in the back end?

There's a couple of reasons. One is the demand, especially in the data center, has become very high recently. So there's increased demand there. And the second reason is that a lot of semiconductor suppliers have moved their back-end supply chains away from China to other countries in Asia, and that has put a strain on the total capacity of these back-end suppliers. So it's the combination of the we've not seen an effect yet not yet of the war there are discussions about some basic elements like gas that are going to be affected. But we don't see this yet. The main reason at this point in time is increased demand, especially in the data center, combined with semiconductor companies moving their supply chains outside of China.

Okay. That's really helpful. And the last question, if I may. As you think about your market share in this year, you of the view that you are a share gainer or you keep share flattish or down? Like what is your view on your ability to gain share?

Yes. So we continue to gain share, '24 to '25. We were -- we exited '25, we had met 40% share. There's no indication that we are not going to continue on that trajectory. This year, the market is really at a high level, transitioning from Gen 2 to Gen 3. And our footprint in Gen 3 is really, really good as well. So there's no sign of any erosion of the share if we add the other components, then we'll grow faster than market because we add content as well to what do we ship to the market. So again, we're very pleased with where we were in 2025. As you know, Mark, we tend to talk share on a yearly basis. They can fluctuate from quarter-to-quarter, but we don't see any sign of erosion of our share going into 2026.

At this time, there are no further questions. This concludes the question-and-answer session. I would now like to turn the conference back over to the company.

Thank you, everyone, who has joined us today for your continued interest and time. We look forward to speaking with you again soon. Have a good day.

Thank you. This now concludes today's conference.

All right. Good afternoon, everybody. Thank you for being here. I am Marco Lagos, I'm Morgan Stanley's head of U.S. semiconductor investment banking. And I am thrilled to be here with Luc Seraphin, CEO of Rambus, this afternoon.

Just to read a quick disclosure here. For important disclosures, please see the Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative.

So with that, Luc, thank you for being here. Great to have you. Why don't we start sort of high level? And I'm going to take it easy on you given the crazy 48 hours of travel you've had, which is symbolic of all the great client and customer interaction you're having.

So let's start at the top. So framing the business just in general, what is Rambus today? But the question really is, I think a lot of investors think about Rambus primarily as an IP company. That is -- IP licensing company. How would you describe Rambus today? And what has changed the most meaningfully in the last, call it, 3 to 5 years?

Yes. For some people, Rambus is known as an IP company, and that's understandable. We still have a very strong and very enduring IP portfolio, which is foundational to our business. But we are not only an IP company. Today, we are really a system-relevant semiconductor company that offers complex chips, chipsets, but also system solutions to the data center memory subsystems, basically. So we've shifted the company over the last 2 to 3 years from being essentially an IP company to becoming a semiconductor company that offers complete system solutions for data centers and AI.

Wonderful. So just kind of touching on that though, talk a little bit about the economic model, product revenue, royalties and IP. How do those things all sort of interact? And why is that mix strategically attractive?

So we have an interesting business model. Intentionally, by the way. On one hand, we have our IP business that brings cash flows, that brings longevity because we have long-term agreements. And that brings us reach in terms of the number of customers that we touch and the number of platforms that we touch with that IP business. And we combine this with a product business, a semiconductor product business that brings growth to the company, that brings customer intimacy, that business. And that brings also relevance into the memory subsystems for AI.

And the combination of the two is quite unique in the semiconductor industry, but it's a very balanced business model that allows us to invest in innovation. That high cash flows that we get from our IP business allows us to reinvest into products in a market that keeps accelerating in terms of the type of platforms that we have to roll out to our customers.

So it's not just the sort of the magnitude of the cash flow, it's the predictability of it that lets you sort of plan ahead as well?

Yes. The magnitude of predictability, those two aspects allow us to go through cycles while continuing to invest in new products and new solutions.

You've now mentioned AI and data center a couple of times. I'm assuming that's the market that you guys are most focused on today? Or is that -- or is it different than that?

This is certainly the focus of the company today. We are going through a memory supercycle in AI. And it's not only a question of having more bandwidth. It's a question of having the right level of power. It's a question of reliability of the systems. It's a question of interoperability of the systems. So it requires a lot of the investment, deep understanding of those systems and intimate relationship with customers. And that's where we position the company today. We certainly address other markets with our IP business, but the focus is really data centers and AI infrastructure.

Wonderful. So with that in mind, right -- and this is directly related. Why don't you tell the audience a little bit about the Rambus DNA? So what is the heritage of the company and why is it right place, right company, right time?

So our DNA is really memory subsystems. And it's not about simply providing chips for the memory subsystems in AI infrastructure, it's understanding how memory interacts with the system. And we've been doing this for the last 30 years. It's -- we understand memory subsystems really, really well. And as I said earlier, it's about how do we see those systems evolve in terms of bandwidth, in terms of power management, in terms of reliability and in terms of interoperability between the different chips.

It's not something that is easy to reproduce. We typically develop solutions ahead of the market, several years ahead of the market, several years before they actually ramp in the market. We work closely with our customers. We work closely with the standard organizations. So we have a pretty good view of where our road map is going. But that system understanding of how memories operate in a data center is really, really our DNA. And what we've realized over the last few years is that memory subsystems have become a critical point of success in AI infrastructure. It's actually a bottleneck, and resolving that bottleneck is what we know how to do.

Excellent. So as it relates to -- obviously, memory is going through a supercycle. I think we've seen some -- just amazing performance from all the traditional memory players in general. How are you positioned for sort of the key memory architectures for AI and data center? So just to name 3 -- and you can talk about more if you'd like or just keep it to these -- but DRAM, high-bandwidth memory, SRAM and the next-generation CXL-type stuff.

So we actually play in all of these areas one way or another. On the product side, obviously, we focus on DDR module solutions or chips for those modules. And we see growth there not only in terms of continuing to grow our market share, but we grow the content on those modules. And those modules are going to become more complex going forward. So we do have several vectors of growth on the DDR side of the business.

We also interact with other customers with our IP business with our HBM IP, our PCIe IP. So anyone in the semiconductor industry who wants to develop a solution for AI infrastructure has a possibility of using our IP. That's really interesting for us, that IP business in HBM and PCIe, more security because we deal with semiconductor companies that build chips that are going to be in the market in 2 to 3 years from now, but it gives us a very good insight of the trends that are happening in those markets.

That's excellent. And so as we think about sort of customer interaction, right? Traditionally, you could look at -- and I'm able to talk about this more, but it's obviously SK Hynix, Samsung, Micron, the storage players, that's been traditionally where you guys have had the most kind of interaction success. How are you thinking about sort of the pull side of things with the hyperscalers and working with them and interacting with them?

That's a very dynamic market. It's a market that is accelerating. So historically, you're correct. Our main interactions were with the 3 memory vendors and with Intel and AMD because that interface chip that we develop sits between the memory modules and the processors. So they have to be validated by both the memory vendors and by the processor vendors.

What has happened over the last 2 years is that the hyperscalers have increased demands on the performance of those systems. They are actually pushing the market to move faster. So we have interactions with them. They are also proposing novel solutions that we have to look at.

So our interaction with the ecosystem has changed. We're not only dealing with the memory vendors and the x86 processor vendors, we're also dealing with anyone who develops processes based on ARM cores, for example. And we interact with our customers' customers, namely the hyperscalers. So this is a very deep ecosystem interaction that we are in today.

And again, you get an early look at all that stuff just given the development?

Yes, we look at it early. We also take a leading role in JEDEC, which sets the standards for this. Because all of these products, whether they are ARM-based, x86-based, whether they're using different types of modules, they all have to work together. So it's important for the industry to define those standards, and we do play a critical role in this standard setting to make sure that we participate in the definition of the road map.

And that gives, I would say, durability in our revenue outlook because we know what we need to develop ahead of time. We actually participate to the definition of that. And that gives our team a very clear road map into what they need to do in the next few years.

Got it. So just the last question on sort of the memory part of it, right? You're talking about durability of the revenue over time, part of it being driven by the stickiness of your relationship over longer periods of time and understanding things well ahead of when they're coming to market. I would say, in memory, the things -- in memory architecture in general, the things that matter are reliability and performance probably over cost, right? And if that's -- if you agree or disagree with that, first of all? Second of all, if you agree with that, how does that dynamic benefit Rambus?

So reliability in a memory subsystem for a -- in a data center is not an option. I mean, not being reliable is not an option. If you have an issue with that chip that sits between the memory and the processor, then you have a system down or the workload doesn't work. So the cost to our customers and customers' customers is unacceptable.

So reliability is very, very important. Therefore, our customers pay attention to the trust that we have developed over the years. They pay attention to the validation process, the quality that we put into those products more than they care about cost. So that gives us a longer-term, sticky relationship with the ecosystem. And that gives us also a way of maintaining price at a reasonable level. Because cost is important. But I would say that's less important than the quality, reliability of the system. And the ability to develop this generation of products generation after generation at a faster pace, that's really what's important for those customers.

Okay. So shifting away from -- a little bit and wading into AI data centers. And you talked about system, so system complexity. That's obviously a big part of the structural growth story for the company. So one thing we've seen is AI is driving incredible innovation as far as the architectures for memory, right? What has surprised you the most about how quickly those requirements are evolving?

Actually, what has surprised me is how fast it has evolved over the last few years. The -- I would say over the last 2 years, we had to double the pace of rollout of our products. We had to expand our product portfolio. And we also have to look at all the innovations that people are coming up with in terms of memory subsystems.

And that plays at the center of what we do. So we actually like that dynamic that is happening now. It's putting a strain on our development teams. They have to work faster, they have to work better. But it also gives us a lot of opportunities in terms of the type of products that we can develop for the future. And that gives us also some immunity against the cycles because these requirements are always going to be here. The requirements for bandwidth, the requirements for low power, the requirement for reliability, the system understanding are always going to be relevant going forward. And that plays at the center of what we do. But the pace of change has been amazing over the last couple of years.

Right. So I'm going to pivot just a second away from sort of the main path here and just talk about your decision-making, how you manage the company. Obviously, a lot of opportunities you could be [ speeding ] and thinking about. You're talking about the change and the pace of the change and how that benefits you. How do you think about where to allocate time, dollars, personnel? How do you make that decision?

If you look at the history of Rambus, one of the first things we did a few years ago is to actually stop the activities that were not related to the memory subsystems in data center or AI infrastructure. And that was, in retrospect, a very good move because of the acceleration that is happening in the AI infrastructure. So the first decision is to decide what not to do, and we did that.

Then when we look at what we want to do. We want to invest in products that have the potential to grow in the market and where we have the potential to take a leadership position. And as a first approximation, what we're doing is we are developing products for every module solution that either goes into an AI infrastructure or data center or in the client system. And what you've seen over the last past years, couple of years is product announcements in that direction. We announced power management chip for the data center. We announced power management chip for the client space. And we also announced companion -- other companion chips for these two. And all of these chips are defined by JEDEC. Again, that gives us the confidence that they are going to be adopted by the market at large. So that's how we make decisions.

In terms of prioritizing those decisions, time to market is really, really important. The validation process for these products is critical. And it's a long process. So you want to be first in market with the new products. So what we look at is what are the products that are so critical in terms of time to market, and we give those products the priority. So we decide what not to do. We decide to focus on standard products that are approved by JEDEC or defined by JEDEC, and we prioritize the product where we can be first to market.

Wonderful. So getting the train back on track here with the system conversation. Where do you see the biggest increase in content per system over the next several years?

Content per system, when the market moved from DDR4 to DDR5, we had an expansion of the content because some of the functions that were sitting on the motherboard in the DDR4 generation of products had to move to the module itself. So instead of having a buffer chip only in the DDR4 generation, we have a buffer chip and what we call companion chips. So there's a content expansion just in a standard module.

Going forward, the market is going to adopt what we call MRDIMM. And MRDIMMs are module solutions that double the capacity and double the bandwidth of systems. That also multiply the content by a factor of 3 to 4. Because at those speed and those capacity, you need to add a lot of more chips than you currently have on a standard DIMM. So that's a very nice, I would say, growth vector for us.

And as I said earlier, we're also addressing the client market. Because in the client market, some of the challenges that we found today or we have found in the data center, we're going to find them in the client systems as well. As speeds go up and memory capacity goes up, some of the functions that we had to perform in the data center has to be performed in the client systems as well. So that's how we see the expansion. More chips on the current DIMMs, MRDIMMs, that is going to increase the content quite substantially. And introduction of similar types of chips in the client system.

Wonderful. So if we're talking about what's coming next, right, which newer products or technology should investors pay attention to that aren't really fully reflected in sort of your guide and your financials today?

Well, today, most of our revenue is coming from the licensing business and from the product business. Product business is actually today, the largest business in the company and the fastest growing business. What is not reflected in our numbers yet fully in terms of potential growth is precisely MRDIMMs because it's going to be introduced to the market towards the end of the year, it's going to be in full swing next year. The client systems as well. So all of these new products that we have developed over the last 2 years that are actually in validation process in the ecosystem as we speak are not reflected in the short-term, really in a meaningful way, revenues. But they are going to contribute to the growth in the longer run.

Okay. Then how do you -- as you think about products and newer products again -- and this is back to the management questions, just as a core strength of the company. How do you balance investing ahead of the curve, right, versus waiting for the standards and the platforms to fully form?

Well, that's an iterative process, if you wish. We do talk to customers and partners, the customers or the memory vendors, the partners or the processor vendors. And we talk to the hyperscalers as well. And with them, we have a view of what they wish their road map to look like. We also are part of the JEDEC standard organization. And that's where we also -- with the other members of JEDEC, define products. So that's how we define the products, and that's how we prioritize our products.

Then, as I said earlier, in terms of timing of rollout of the product, it's really about being first to market with the most complex products. If you look at the history of our product introduction for the DDR5 cycle, for example, the first product we introduced is what we call the RCD product, the clock recovery product on the module because it's the most difficult product to design. It's the longest product to validate in the market. So it was important for us to be first with that. We came with companionships later, and we did this on purpose because we didn't want to miss the transition from DDR4 to DDR5. And we didn't miss that transition. When we moved from DDR4 to DDR5, we went from a 20%, 25% market share to 40% plus market share today. So that was the right decision.

We then introduced our companion chips. And in the companion chips, we invested heavily, in particular, in the power management chip because we believe this is something that is critical to the performance of the system after the RCD chip. And it was important for us to build the team so that we have the know-how internally. And we've introduced our first products. And with these first products, we have very strong and positive customer feedback, especially in the high end.

So that's what we do. We pick the things that are difficult to make that are really relevant for our customers. We want to try to be first with those. That's how we develop our road map.

So when I think about sort of -- you obviously -- you talked about choosing what not to do and then thinking about the things you want to tackle and how you're going to tackle them being sort of critical parts of the strategy. You mentioned power management a couple of times, right? That has not traditionally been an area where Rambus is focused as developing a sort of specific product. It's happening now, right? Obviously, competition is robust in that ecosystem with a lot of traditional players out there. How is Rambus well suited to compete and succeed in power? And how much would you say the growth and the future of the direction of the company depend on the ability to succeed in that specific vector?

Yes. Power management is different than power management on a module. A module environment is a very stringent environment in terms of thermal requirements in a server, in terms of space and in terms of how precise this power management, this power delivery has to be. So it's more than being able to do power management at large. It's about being able to do power management for this type of memory subsystems. And the power management chip interacts with the other chips on the module. So the understanding of how the system works is also very, very important.

So we do understand modules. We do understand the interoperability of chips on the module. We understand the ecosystem because these are the same customers with the same validation processes. And we have invested in power management several years ago. We were silent about it. We have built the teams. And the first power management chip that we have released to the market have had -- we were not the first. We came into a market where you had incumbents there. But we've been designing by all of our customers, especially on the high-end side.

So would you say that the way you've tackled power management is purpose-driven, as opposed to reverse engineering with a product that already existed?

Absolutely. It's purpose-driven. This is also an area where we work with JEDEC and the industry. But AI infrastructure will require more bandwidth, more capacity, and this is not going to stop. And on the same type of form factor. So the system requirements are going to become more and more complex. And having the ability to understand power management in that environment, clock recovery in that environment, system interaction in that environment is really, really critical. And I think that as the infrastructure for AI continues to evolve and improve, it's going to play on those strengths of ours.

Yes. So I think it's clear. Connectivity, memory, power. You mentioned clock. Timing is a little bit of a dark art. How do you see that evolving for Rambus? And how do you think you'll be playing that market?

Timing?

Timing, yes.

The RCD chip is actually a timing chip. And the challenge with these chips is you have memory signals at very high speed traveling very close to each other. So there's a lot of cross noise and these type of things. And that's why on our earnings call, we talk about signal integrity. The key here is to be able to deliver clean signals between the memory and the processor in a very noisy environment and at a higher and higher speed. So that's signal integrity. That's actually timing technology.

So in the future, there will be timing requirements. The first adjacent markets that we are addressing, as we said earlier, is the client system. In client systems, we're going to have faster and faster signals traveling on very -- lines that are very close to each other. That's going to create noise. We will have to clean those signals. This is really a -- not as much as engineering work. And that's why it's very difficult to reproduce. This is things that we have learned over decades of work in that specific area.

Terrific. All right. So a little bit of a hypothetical. You're looking out 2 to 3 years in the future. What would you make Rambus -- what would make Rambus look meaningfully different than it does today?

Well, I think we'll continue to have a broader portfolio of products. That's certainly something that is clear. I think we're going to have deeper relationships with our customers and the hyperscalers in terms of defining what's going to be required for next generation of AI. These are two areas that are absolutely clear for us.

There's certainly the capability for us to expand our know-how in signal integrity into the other markets or our know-how in power integrity into other markets. That's where we see the growth vectors above and beyond what's driven by the AI infrastructure. But the AI infrastructure offers tremendous growth potential for us already.

Great. All right. Well, look, I can't hand it off to the audience or without talking a little bit about your financial model. It's attractive. Rambus generates substantial free cash flow, right? Why don't we talk about some of your metrics and your performance and free cash flow. Anything that you're particularly keen for the audience and the investors to understand?

So yes, we generate strong cash flows just because of the structure of our business, the combination of the IP business and the product business. And our #1 priority is always to reinvest into products that are going to continue to fuel profitable growth into the business. And this has been the case over the last couple of years.

As I said, the road map has accelerated. The pace of that road map has been doubling over the last few years. And the breadth of our product offering has been expanding as well. So our #1 priority is to reinvest into our business so that we can continue to grow and to keep our leadership position there.

Then we look at M&A. Of course, that's always a vector, whether it is to expand or accelerate our growth, but also in terms of bringing the right resources internally. And as I said, we are specialized in power integrity, signal integrity, and it's important to have the right resources to continue to fuel that engine.

And finally, we have a -- we are returning cash flow to our investors on a regular basis. Say, 40% to 50% of our cash flow is being returned to our shareholders on average. We pick the right time to do this. But we are able to balance these 3 vectors.

Wonderful. A few minutes left. I wanted to see if anybody in the audience have any questions? We've got one right here up front.

I have one question. So compared to the CXL, I think Rambus is more preferred for the MRDIMM to expand the memory, right? So -- but right now, there's -- some hyperscalers, they have traction -- some traction to use CXL to disaggregate the HBM and the story -- sorry, the memory as computing.

So my question is firstly, like I remember in your previous quarter calls, you said that you -- Rambus will be -- invest more in the MRDIMM compared to CXL. So this thing will change it or not? Will you think about to develop any kind of CXL controller in the future is the first question.

The second question is like -- since the hyperscaler is thinking about to use CXL to disaggregate the HBM to the GPUs or ASIC, will this impact Rambus HBM IP business?

I think there were two questions there.

Two questions, yes.

First is -- the first question was around CXL and MRDIMM. The initial idea about using CXL for memory expansion is -- was the following. If you take a processor in a data center, the processor has a limited number of memory channels. And the AI workloads require a lot of memory. So very quickly, you saturate those memory channels. And once the memory channels are saturated with memory and you need more memory, then the only way to do that is to have another processor with memory, and that's not economically...

Is that the memory wall that people talk about?

This is the memory wall that people talk about. But the -- so one idea was to use the CXL interface to actually add memory. That -- this is where it started. And at Rambus, we play in CXL in two ways. One, we have a CXL IP controller that as part of our silicon IP business, that we sell to whoever wants to develop a chip in the semiconductor industry that has a CXL interface.

The second way we play in CXL is we have developed our own CXL chip, CXL controller chip, but we've not commercialized that chip. And the reason is that CXL defines an interface, it does not define a chip. It's not like JEDEC. And what we see in the market is people who have a CXL controller chip, these are custom chips. The Customer A has a flavor of a CXL controller, Customer B has another flavor, Customer C has another flavor. And for us, the business model didn't really make sense if we had to develop a chip per customer.

This is -- it's -- the market in aggregate is an interesting market, but it's a very fragmented market. But if we want to talk about memory expansion, then the MRDIMM solution is a more elegant solution. Because you just remove a DIMM from a memory channel and you plug an MRDIMM instead and you double the capacity and double the bandwidth with the same software infrastructure and with solutions that are being developed by JEDEC. So that has been adopted across the industry and across customers. So that's one reason. So we still play in CXL, but we've limited our play in providing IP, silicon IP to anyone who wants to develop one of those custom chips for this aggregation, whether it's for HBM disaggregation or DRAM.

Okay. Perfect. Anybody, any other questions from the audience? Okay. If not, I do have one. I know. Yes, I do have one. Back -- we started with the big picture, let's close with the big picture, right? So if an investor puts their money to work on Rambus and they checked back in on that investment in 5 years, what do you hope that they will say that the company got right?

What I would say -- I'd like them to say that we understood how the memory subsystem evolved. We anticipated that, and we invested early with conviction into these new memory subsystems. I hope they would say that we could balance our investments in a way that prioritize products where we have strength. And I hope they would say that we have become a very strong player in the development of AI infrastructure because memory is critical to AI infrastructure.

Perfect. Look at that, almost perfect timing there. Thank you very much, Luc. Appreciate your time today. Thanks, everybody. Oh, we've got one more question in the back of the audience. Sorry about that.

Sorry to keep everyone here. Can you help us think through DIMM growth? Like there's a lot of moving parts on server growth and DIMMs per channel. And just what you guys view the DIMM growth market as and how you should perform in that? And then just a second question for MRDIMM, just how we should think about the mix of that in the ecosystem in like '27, '28?

Yes. So for the DIMM growth, what we quoted, there are several factors entering the DIMM growth is the server growth. We quoted in the last call, the 8% growth from Gartner, which is on the high end when you look at these analysts.

But we believe the DIMM growth can be higher than that. There are several factors coming into it. One is the number of memory channels per CPU. It moved from 8 to 12, it's going to go to 16. So that increased the DIMM growth. You have to balance that with the capacity on each one of the DIMMs because it can increase capacity just by increasing the capacity on 1 DIMM, not only adding DIMMs. And you have to balance also traditional servers versus AI servers. But all in all, we believe the DIMM growth is going to be higher than this 8% that we quote. It's probably going to be double digit, but that's something we continue to track on a regular basis.

Okay. All right. Well, unfortunately, we are out of time. So thanks very much again. I appreciate it. Thank you.

Welcome to the Rambus Fourth Quarter and Fiscal Year 2025 Earnings Conference Call. [Operator Instructions] As a reminder, this conference call is being recorded. I would now like to turn the conference over to Desmond Lynch, Chief Financial Officer. You may begin your conference.

Thank you, operator, and welcome to the Rambus Fourth Quarter and Fiscal Year 2025 Results Conference Call. I am Desmond Lynch, Chief Financial Officer at Rambus. And on the call with me today is Luc Seraphin, our CEO.

The press release for the results that we will be discussing today has been filed with the SEC on Form 8-K. We are webcasting this call along with the slides that we will reference during portions of today's call. A replay of this call can be accessed on our website beginning today at 5 p.m. Pacific Time.

Our discussion today will contain forward-looking statements, including our expectations regarding projected financial results, financial prospects, market growth, demand for our solutions, other market factors, including reflections of the geopolitical and macroeconomic environment and the effects of ASC 606 and reported revenue amongst other items.

These statements are subject to risks and uncertainties that may be discussed during this call and are more fully described in the documents we file with the SEC, including our 8-Ks, 10-Qs and 10-Ks. These forward-looking statements may differ materially from our actual results, and we are under no obligation to update these statements. In an effort to provide greater clarity in the financials, we are using both GAAP and non-GAAP financial presentations in both our press release and on this call.

A reconciliation of these non-GAAP financials to the most directly comparable GAAP measures has been included in our press release, in our slide presentation, and on our website at rambus.com on the Investor Relations page under Financial Releases.

In addition, we will continue to provide operational metrics such as licensing billings to give our investors better insight into our operational performance. The order of our call today will be as follows: Luc will start with an overview of the business. I will discuss our financial results, and then we will end with Q&A.

I'll now turn the call over to Luc to provide an overview of the quarter. Luc?

Thank you, Des. Good afternoon, everyone, and thank you for joining us. 2025 was an excellent year for Rambus. We closed with a strong Q4 and finished the full year with record revenue and earnings. Our financial success is a testament to both our strategy and execution as we continue to deliver products and technologies that accelerate memory, compute and connectivity advancements in rapidly growing markets.

Our diversified portfolio remains a core strength for the company and each of our businesses contributed meaningfully to our results as we delivered a new annual high in cash from operations. This positions us well to continue to invest strategically in our product road map, expand our market opportunity and drive long-term growth. Before I go into detail on our business results, let me take a moment to discuss the important market and technology trends influencing our strategy and highlight several of our key accomplishments in 2025.

Both AI and traditional server markets remained strong throughout the year, driven by the accelerating need for significantly higher compute and memory performance. As workloads become more complex and diverse, and inference rapidly expands across applications, including agentic and physical AI, the demands placed on memory subsystems continue to intensify. This environment drove further adoption of DDR5 as well as other high-performance memory and interconnect technologies where Rambus signal and power integrity expertise are foundational.

The accelerated pace of innovation continued across the industry with customers increasingly operating on 1-year product cadences to stay ahead of demand for greater performance. This dynamic amplified the need for cutting-edge merchant and custom solutions, where our advanced technology portfolio enables accelerated design cycles for our customers.

Against this backdrop, Rambus had a number of achievements that fueled our performance in 2025 and strengthened our position across key markets as we move into 2026. We furthered our leadership in DDR5 with increased market share in RCDs, reflecting both the depth of our expertise and the continued trust of our customers. Our power management chips made meaningful progress with growing adoption of our DDR5 PMICs contributing to revenue growth.

We extended our reach in high performance and AI PCs through the introduction of our complete client chipset. With this addition, Rambus offers a comprehensive chipset portfolio that supports all JEDEC standard DDR5 and LPDDR5 modules across server and client systems. With that, we offer customers greater assurance of interoperability and reliable performance at scale.

And finally, in addition to these chip milestones, we saw increasing design wins and customer engagement led by our latest generation HBM4, GDDR7, and PCIe 7.0 digital IP, as well as a broad range of security IP to safeguard data transmission and storage.

Turning now to our quarterly business results. Chip capped off the year with a strong Q4 performance, delivering product revenue of $97 million. This brought us to a new annual record of $348 million. which was up 41% year-over-year. This achievement reflects our continued product leadership and ongoing market share gains in DDR5 RCDs. In addition, customer adoption of new products continues to progress with growing revenue contributions and volume shipments underway.

For silicon IP, we are strategically focused on delivering industry-leading solutions that empower the next wave of AI hardware. The increasing pace and diversity of AI chip designs, including custom silicon for hyperscalers is driving design wins for high-speed memory, interconnect and security IP. With market leadership and expertise across multiple generations of HBM, GDDR and PCIe as well as a best-in-class security solutions.

Our IP is a critical enabler of the performance required by AI workloads. We see strong traction across our portfolio of cutting-edge solutions. In particular, this growing demand for our interface and security IP solutions as we see the increased need to move and secure data in scale-up and scale-out scenarios.

Looking ahead, the ongoing expansion of AI and the transformation of the data center continues to reshape memory and interconnect requirements. AI training and inference at scale are driving increased demand for bandwidth, capacity and power-efficient performance. The expansion of agentic AI is catalyzing traditional CPU-based server demand and continues to drive the need for more DIMMs per system, higher speed interfaces and sophisticated power management.

Our product and IP sit at the core of this transition, enabling the massive compute infrastructure required for increasingly complex and diverse AI models. In addition, the rise of purpose-built systems and increasingly heterogeneous compute is accelerating the adoption of new memory architectures, higher data rates and advanced security solutions. All of these trends play directly to Rambus' strength, open opportunities to broaden our leadership across next-generation platforms and reinforce the long-term tailwinds for our businesses.

Rambus is well positioned to capitalize on these trends. And in 2026, we expect to grow faster than market. Now as reflected in our Q1 outlook, we experienced a onetime supply chain issue that will affect product revenue for Q1. The issue is being resolved in collaboration with our supply chain partners, and we expect our product business to return to strong growth in the second quarter, fueled by market share gains and the continued ramp of new products I am confident in our long-term trajectory for 2026 and beyond.

As always, I want to thank our customers, partners and employees for their continued support. With that, I'll turn the call over to Des to walk through the financials. Des?

Thank you, Luc, I'd like to begin with a summary of our financial results for the fourth quarter and for the full year 2025 on Slide 3.

we delivered strong financial results in both the fourth quarter and full year 2025 as we continue to execute on our long-term growth strategy. Full year revenue and earnings per share reached record levels driven by a 41% increase in product revenue to $348 million due to DDR5 market share gains, and new product contributions.

In 2025, we generated a company record $360 million in cash from operations, which was up 56% from 2024, an established track record of generating cash enables us to invest in initiatives that fuel our long-term growth. Let me now provide you a summary of our non-GAAP income statement on Slide 5.

Revenue for the fourth quarter was $190.2 million, which is above our expectations. Royalty revenue was $71.7 million, but licensing billings were $71.5 million. Product revenue was $96.8 million as we delivered another quarter of record product revenue. This represents 32% year-over-year growth driven by continued strength in DDR5 products and ramping new product contributions.

For the full year, we delivered $347.8 million in product revenue, which was a new annual record for the company. Contract and other revenue was $21.8 million, consisting predominantly of silicon IP. As a reminder, only a portion of our silicon IP revenue is reflected in contract and other revenue and the remaining portion is reported in royalty revenue as well as in licensing billings.

Total operating costs, including cost of goods sold for the quarter were $103.2 million. Operating expenses of $64.9 million were in line with our expectations and flat compared to Q3. Interest and other income for the fourth quarter was $6.4 million, using an assumed flat tax rate of 20% for non-GAAP pretax income, non-GAAP net income for the quarter was $74.7 million.

Now let me turn to the balance sheet details on Slide 6. We ended the quarter with cash, cash equivalents and marketable securities totaling $761.8 million, up from Q3, primarily driven by record cash from operations of $99.8 million. Fourth quarter capital expenditures were $8.6 million, with depreciation expense was $8.4 million. Free cash flow in the quarter was $91.2 million, and for the full year, we delivered $320.9 million or 45% free cash flow margin.

Let me now review our non-GAAP outlook for the first quarter on Slide 7. As a reminder, the forward-looking guidance reflects our best estimates at this time, and our actual results could differ materially from what I'm about to review.

In addition to the non-GAAP financial outlook under ASC 606, we also provide information on licensing billings, which is an operational metric that reflects amounts invoiced to our licensing customers during the period adjusted for certain differences. We expect revenue in the first quarter to be between $172 million and $178 million. We expect royalty revenue to be between $61 million and $67 million and licensing billings between $66 million and $72 million.

As Luc mentioned earlier, our Q1 product revenue is impacted by a supply chain issue, which has been resolved. We expect resumption of growth from the second quarter onwards. We expect Q1 non-GAAP total operating costs, which includes COGS to be between $104 million and $100 million. We expect Q1 capital expenditures to be approximately $13 million. Non-GAAP operating results for the first quarter are expected to be between a profit of $68 million and $78 million.

For non-GAAP interest and other income and expense, we expect $6 million of interest income. We expect our pro forma tax rate for 2026 will be 16%, driven by tax legislation changes last year. We expect non-GAAP tax expenses to be between $11.8 million and $13.4 million in Q1. We expect Q1 share count to be 110 million diluted shares outstanding. Overall, we anticipate the Q1 non-GAAP earnings per share range between $0.56 and $0.64.

Let me finish with a summary on Slide 8. In closing, I am pleased with our excellent 2025 financial performance and the continued progress we are making against our strategic goals. We delivered record top line revenue growth, resulting in record profitability and cash generation. Our diversified portfolio continues to be a core strength for the company.

First, patent licensing continues to deliver consistent and predictable results. Also, our silicon IP portfolio is well positioned to address the accelerating demand for AI solutions. In addition, our product business continues to drive our growth with strong leadership and market share gains in our core RCD business, which is complemented by our expanding new product contributions. Overall, we are well positioned to drive long-term shareholder value.

Before I open up the call to Q&A, I would like to thank our employees for their continued teamwork and execution. With that, I'll turn the call back to our operator to begin Q&A. Could we have our first question?

[Operator Instructions] The first question comes from Kevin Cassidy with Rosenblatt.

Congratulations on the great results. But of course, the questions will be around the supply chain issue. I understand you've resolved the issue. Will there be catch-up -- meaning in the second quarter, can you make up for that revenue lost in the first quarter, or is that just lost market share at -- a competitor picking up the business?

Let me maybe take a few minutes to explain what the supply issue is, that we understand the dynamics in the market. So in Q4, we -- as we said, we identified the back-end manufacturing issue with 1 of our OSATs. We have identified the root cause of that issue, and we have implemented all the corrective actions in collaboration with our supply chain partners.

And before I go into the detail, note that the issue was affecting an extremely low number of parts, which made the identification of the group was a bit difficult because it was hard to reproduce. But we have identified the root cause. We've put the measures in place. And in reality, what we've done is, we've done 2 things. The first thing we've done is once the root cause was identified and the corrective actions were in place, we did actually pull forward fresh material from inventory that was originally staged for Q1 to meet our Q4 customer demand because our customer demand remained very strong in Q4. So that's the first thing we did. We accelerated fresh material once these measures were in place.

The second thing we did is, despite the very, very low PPMs that we observed and because quality is paramount out of abundance of precaution, we actually quarantined all potentially impacted production material. And now we are retesting this material with enhanced screens in place. So these measures have put additional strain on capacity in a tighter supply environment and that impacts Q1, as we said.

But the issue was identified in Q4, we accelerated material through -- after we put the measures in place. We are rescreening parts that were potentially painted, and that's what's creating that issue in Q1. So that issue is behind us. And the lower Q1 product revenue does not change the trajectory of the business. We expect the business to return to strong growth in Q2. And the product revenue for 2026 remain on track to grow faster than market. And that's how I would qualify the issue, Des, I don't know whether want to add anything to this.

No, I think you've summarized it well, Luc. The issue in Q1 is behind us, and we're expecting strong recovery both in Q2 and also for the full year. And as you said, we do expect the business to grow faster than market for the year. So we're very well positioned from here.

Okay. Great. Thanks for that detailed explanation. Maybe a more difficult question, but what -- can you quantify what the revenue would have been?

Kevin, it's Des. What I would see is the impact would probably have been around low double-digit million impact in what's already a seasonally soft quarter for the business. So that's how I would sort of quantify the sort of Q1 revenue impact from there. As Luc mentioned, we are -- we will build inventory by the end sort of Q1, and we'll be in a position to return to strong growth in Q2 from there. But I would say quantification probably in the low double-digit million impact is what I would say, Kevin.

The following comes from Kevin Garrigan with Jefferies.

Can you just talk about how your RCD market share finished for 2025?

Yes. Thanks, Kevin. So we believe that we ended up the year in the mid-40% share for DDR5. We -- the market between '24 and '25 grew mid-single digit, but the portion of DDR5 became more important. DDR4 continues to decrease in terms of share. So in 2024, we were in the early 40s for DDR5, in 2025, we believe we are in the mid-40s on DDR5 in a market where DDR5 dominate even more. And I think as we said in the prepared remarks, we expect to continue to grow faster than market in 2026, despite the glitch we had in Q1.

Okay. Perfect. I appreciate that color. And then just as a follow-up. So there's a lot going on with the Intel Diamond Rapids platform and even the AMD Venice platform. So just kind of wondering if the time line and opportunity that you're expecting on the MRDIMM front hasn't changed at all?

Thanks, Kevin. No, it hasn't. We are monitoring the rollout of these platforms as every generation has been the same dynamic the rollout of our products mostly depend on the rollout of the platforms from Intel and AMD. So we expect our MRDIMM to ramp towards the very end of the year at this point in time, but we will modulate that based on how the platforms roll out from both Intel and AMD are happening.

I think that's nothing new. This has happened in every generation in the past. We are readying our products. We are working with the ecosystem to make sure that we are ready. But eventually, that will depend on when those platforms roll out. As far as we're concerned, we are ready.

Okay. Perfect. Congrats on the results.

The next question comes from Aaron Rakers with Wells Fargo.

I've got a couple if I can as well. I guess, first of all, going back to the supply chain issue. Yes, I can appreciate the issues have been rectified. I know, Luc, you've referenced a couple of times growing faster than the market. So I guess the question I have is, how do you define the growth rate of the market? We've seen a lot of data points where server demand looks like it might be as much as mid-teens, maybe even high teens and some of the commentary recently. So I'm curious if you can just kind of contextualize what you think the market growth rate is in 2026, underpinning your expectation of growing faster than that.

Yes. Thanks, Aaron. We see a wide range of numbers for the market growth. Typically, as you know, there are many variables going into this. One of -- 1 of the basis is really the market for servers. The analysts, the marketing analysts have a range, new for market servers, Gartner is at 8%, we hear from other sources that this could be, as you said, double-digit growth.

But we want to stay prudent with the view of the server growth because we believe the demand is here. But I think some people tend to underestimate the impact of potential shortage, especially on the memory side. So we tend to align with Gartner's view with 8% market growth for the servers. So we certainly exceed that.

But you have other things happening. The number of channels increasing, the introduction of new platforms. In our case, we also are introducing our new products, so we're going to be higher than that. But the basis we use is mid- to high single digit growth for the server market. That's our basis.

Okay. That's very helpful. And then kind of sticking with that, when we talk about your companionship opportunities, I think last quarter, you talked about the PMICs being, I want to say, with mid-single digit contribution to your total product revenue. Can you unpack that a little bit? How fast is that growing? What's the expectations for this year?

Aaron, it's Des here. We're really pleased with the progress and traction that our new products continue to make in the market. New products have grown from low single digit contribution in the first half of '25 to upper single digits in Q4, which was in line with our expectations. As we look ahead to Q1, I do expect the strong traction really to continue where I do expect new products will continue to be -- it will grow to about double-digit contribution of total product revenue.

We have traction across all of our products. But I would say that in terms of revenue contribution, PMICs remains the largest contributor there. Our customers continue to place value in the importance of the interoperability between RCD and PMIC. And as we look ahead into to sort of '26 with the continued rollout of new platforms, I would say that our new products are very well positioned within the market to continue to grow and take market share.

The next question comes from Bastien Faucon with Susquehanna.

I guess 1 question I have is revisiting the average of the DIMMs for CPU expected in 2027. And you mentioned it previously, given the cost of memory and the shortage, has this changed your expectations of having channels are being populated with DIMMs for CPU and I have a follow-up.

Thanks, Bastien, for your question. The DIMMs per CPU, dynamic is a complex one. Typically, what happens is people who want very high bandwidth like in AI types of applications tend to use fewer DIMMs per channel so that they can make the best use of this bandwidth. And people who are in need of more capacity tends to populate more DIMMs on their channels. And then you combine this with the respective growth of standard applications with AI applications, so we continue to see, on average, the number of DIMMs per channel are growing, but it's difficult to -- it's a bit difficult to really put a number on.

I think the memory situation is a broader situation than the number of DIMMs per channel. I think thank God memory is booming these days. This dynamic between HBM and standard DDR, for example. And with the standard DDR, there's a dynamic between the different speeds of these DDR. So I think overall, we believe that the market is going to be constrained. But again, trying to put a number on how the supply constraints on the memory side is going to impact the number of DIMMs per channel is something that is quite difficult to figure out.

All right. That is very helpful. And I have a follow-up. In terms of RCD contribution, what are your expectations of the DDR5 Gen 3 RCD contribution relative to the Gen 1 and 2 in 2026, given the supply chain issue that you've encountered with your RCDs that will be impacting Q1?

Yes. That's a good question. What we saw is, in Q4, Gen 2 was predominant. This is what we were expecting and Gen 3 was starting to ramp. It was growing in Q4 compared to Q3. When we look at 2026, our view is that Gen 3 will continue to grow and will probably be the predominant version of DDR5 throughout the year.

Gen 4 will contribute somehow. But because this is on a different type of core, it will have more limited adoption. The big next step is going to be Gen 5 and Gen 5, as we said earlier, is going to depend on the introduction of the next-generation platform on Intel and AMD. So in summary, we continue to see Gen 2, Gen 3, and the mix between Gen 2 and Gen 3 is changing. Gen 3 is growing, and our expectation at this point in time is that Gen 3 is going to be dominant in 2026.

The next question comes from Gary Mobley with Loop Capital.

I have a multipart follow-up question about the supply chain issue. First, do you see any reputational harm from this with your customer base? Did it impact the companionship business more than the RCD business? And I guess, logically, we should assume a sharp revenue recovery in Q2. It sounds like Q1 revenue would have been about $99 million to $100 million, which is described as seasonally weak.

And therefore, if you're going to recover that revenue and gain share in the year, presumably Q2 revenue would have been up sequentially from that. So can your supply chain recover to that degree that quickly to get back to the $100 million plus per quarter in product revenue.

So thank you, Gary, for your questions. I'll start with your initial questions and let Des comment on the numbers. Your first question is about the reputational risk. No, there's no reputational risk. Actually, when we identified that issue, we had all hands on deck and we work in close collaboration with our suppliers and our customers.

And I think it's really, really important we said over and over again over the last few years, that quality management is really, really important. We had a real life example here, where we identified an issue quickly. We had a very thorough quality process in place with our suppliers, with our customers and we're back on track.

The only issue that is left for Q1 is the fact that we need to replenish our supply chain and makes the best use of our testing capacity as we are also retesting old parts but the reputation has not been damaged. We've been able to identify the problem, fix the problem and put actions in place quite quickly.

Your second question was about whether it affected the RCD or the other chips, it only affected the RCD and actually older versions of the RCD, but the companionships were not affected at all on that. On the numbers, maybe, Das, you want to comment.

Gary, it's Des. In terms of the inventory, I do expect that the inventory will be replenished by the end of sort of Q1, and we'll be able to grow the inventory to a level, which will be able to support our Q2 '26 demand going forward from there. So Again, as Luc talked about, the issue has been contained, we'll continue to replenish our inventory as we go throughout Q1, and that will put us in a good position ending Q1 for meeting customers' demand for Q2 going forward.

Got it. For a follow-up, I want to ask about MRDIMM based on what you're seeing in timing of vendor shipments and Diamond Rapids shipments and sort of queuing the memory ecosystem around those 2 server processor launches. Do you still see revenue contribution, I guess, material revenue contribution from MRDIMMs by the end of the calendar year?

As we said earlier, we are monitoring the rollout of these platforms, and we are continuing activities around MRDIMM. As we said on the earlier call, we see the initial contribution towards the end of the year. That's -- the very initial contribution of this platform is going to be towards the very end of the year. And the main contribution is happening in 2027.

The following comes from Tristan Gerra with Baird.

It looks like you started to be a little bit more bullish on your market share prospect in RCD with companionships ramping. Is that the reason why we're now seeing market share, it looks like is above what your expectation was a year ago in mid-40s and what would be kind of the upside that you think you could get to by end of this year or even next year?

Thanks, Tristan. I'll make the first comment about the market share. When we talk about being in the mid-40s is for DDR5 RCDs. So this market share gains year-over-year are really referring to the RCD chip. And this is the result of the increased design win footprint we were able to secure from generation to generation.

From D4 to D5, we have many more -- a much higher footprint, and at every generation of D5, we increased our footprint in terms of design wins. That translates into our market share for the RCD chip. So when we mentioned that in 2025, our market share was in the mid-40s. That's on the DDR5 RCDs.

And the DDR5 overall generation is still early in its cycle. So there's still room to gain share in the mid-40 now. We always said we could be between 40 and 50, so we're still chasing more share on the DDR5 RCD chip.

The companionships are an addition to this. And they're ramping steadily, slowly, as Des explained into the market as the qualifications take place. But this is going to be additional revenue to the RCD revenue.

Yes. And I was just wondering if the fact that you have companionship, does that help your RCD share? Or is that completely separate? Sense that perhaps you saw some cross-selling opportunities or benefits that will go beyond just the additional TAM of the companionship.

And then also my follow-up question is if there's any update on the potential SOCAMM2 opportunity, whether it's in the current Blackwell platform or the upcoming platform for you to potentially participate?

Yes. I'll answer first on the companionship. The TAM, yes, one way to look at it, as you rightly say, is to add the TAMs. But is there a connection between the two, is an indirect impact as the speeds on the DIMMs continue to increase, it is more and more important for our customers to get their chips from the same supplier for interoperability reasons. These systems are very, very complex.

And if we have all chips in-house, we can do a lot of system testing before shipping those parts to our customers. So that puts us in a favorable position. So there's a positive indirect impact on our ability to grow our PMICs, in particular, but also the other companionship as the speeds on the RCD continue to increase. So that's the answer on that.

On the SOCAMM, we continue to monitor the dynamics there on the SOCAMM, theres definitely an SPD opportunity on the SOCAMM for us. We're talking about next generations and how these next generations can evolve, in particular in the field of power management that could open other opportunities in the future.

But I would say this, as we said in the prepared remarks, we -- our strategy is to have solutions for every JEDEC standard module, whether it's on the client side or whether it's on the data center side. So we will continue to monitor what's happening with SOCAMM or the SOCAMM2, we have an opportunity for the SPD Hub, as the evolution of SOCAMM continues and new chips are being defined, we're going to be part of that definition, and we'll continue to develop chips to support that market.

The next question comes from Sebastien Naji with William Blair.

Could you maybe remind us how much of your product business today is not related to the server market. You mentioned some early success in the client market. And as we think about 2026, does rising memory cost maybe create some friction in this part of the market for Rambus?

The client market remains minimal for us at this point in time for a couple of reasons. One is the adoption of the CKD chip or the equivalent of the clock chip into the client space really is limited to the very, very high end part of that client space. So the contribution is minimal in terms of numbers.

Our goal is still to get 20% share in the long run for that. But these platforms have to -- these platforms have to ramp in the market. Their contribution are still going to be minimal even in 2026 for clients. So the vast majority of the business is in the in the data center space. But this being said, in the long run, the power management and the clock management are going to be very, very important in the client space as well. It's important for us to position ourselves there and to have solutions for all platforms. So that's why we're doing this in terms of the client space. And your second question was?

No, no, no. That was my first question. My second question is on IP side of the business, if I could.. So Rambus has benefited a lot from the explosion in the number of ASICs that are being designed. You have many companies attempting to design their own XPUs. You've also seen an accelerated cadence of new chip releases. As we go into 2026, are you seeing any signs of a slowdown in some of these new chip design starts that could start to impact your IP business?

Well, I start...

Oh, you want to go ahead, Des, go ahead.

Yes. I'll start Luc, maybe you can add on. We were very pleased with how our silicon IP business performed in 2025, it performed in line with expectations. And the portfolio is really well positioned to address the demand for AI solutions from there.

If you look at our portfolio, we have a leading-edge portfolio with critical IP solutions in the high-speed memory interconnect and security IP, which is tailored towards the AI sort of workloads from there. And our expectation is that, that would continue to grow in 2026, in line with our long-term growth expectations from there. So very bullish on the overall sort of portfolio and outlook for the IP business.

The following is a follow-up from Kevin Cassidy with Rosenblatt.

Maybe along those lines of custom, Luc, I think you had mentioned custom hardware. And I wonder if you could give us a little more details on that, how many customers can you support? And what would be the timing of that?

Yes. When we say custom hardware, there are a lot of people who are developing their own chips for their AI infrastructure or the server infrastructure typically accelerators, chips that are dedicated to inference and these kind of things. So every time they do develop those type of chips, they have a potential need for HBM, at high speed, PCIe at high speed or security solutions.

So GDDR sometimes. So as you know, we positioned our portfolio to be at the high end of those standards. So we typically talk to the people who work at the high end of those systems. We can support a large number of customers because we have a limited portfolio in terms of the scope. We focus on PCIe, CXL, HBM, GDDR and security IP. So we have a laser-focused portfolio that addresses potentially a large number of customers who are working on the leading edge of those technologies. That's really what's driving the business for us as opposed to potentially other IP suppliers that have a much broader portfolio.

We narrow our portfolio for the needs of people who develop chips for the data center. And most of these ships are either their own processors, some people develop their own processes as opposed to buying merchant processors, other types of applications are accelerators to improve the performance of their systems.

We have another follow-up from Aaron Rakers with Wells Fargo.

I guess the first one is, Luc, you mentioned like there is risk in terms of memory supply and availability. I'm curious, as you look back at this last quarter or coming out of the last quarter in these first couple of weeks of the first quarter, have you seen any signs of memory constraints impacting your customers' ability to fulfill demand? Or any -- how would you characterize the inventory levels that you're seeing at some of your major customers? Any thoughts on that would be great.

Yes, sure. We are in a small ecosystem, as you know. And when we talk to our customers and partners, we hear those comments. And one of the common theme that we hear is that the demand for servers is solid. There's a refresh cycle, there is a -- that is not over. There's also agentic AI and all the inference applications that drive demand. But what we hear from the same customers is that they're going to be constrained by supply, and we hear this directly from our customers.

And this is why when we look at the market potential for us, we tend to be prudent because we are aware of these comments from our customers in terms of supply. So that's what the basis of what the outcomes are.

We see -- on the supply side, we also see the supply -- on the supply side, we also see lengthening lead times. It's nothing to do with the memory guys, but there's also on the supply side, lead times continue to increase. And that's why we believe in 2026, the demand is solid, but we're going to be more constrained by supply that we're going to be very demand.

Yes. That's helpful. And then Des, real quickly on the gross margin line. I want to make sure I'm clear, like given the supply chain issues, you don't expect any kind of gross margin, any kind of inventory provisions or anything of that nature? And I guess what I'm trying to get at is the product gross margin looks like it's still hovering in that plus 60% range. Is that still the expectation that we stay in that low 60% range here as we look forward?

Yes, that would be the right expectation going forward. If you look at the full year of 2025, our gross margins for around 61.5% which was in line with 2024 performance and consistent with our long-term model of 60% to 65%, I think what you will see is that we have a strong track record of delivering gross margins in line with these targets. And that would be my expectation.

If you really look where we've been operating in the last sort of 3 years, we've been in a tight range of 61% to 63%. And I think that would be a fair way to think about the business in 2026 from our gross margin perspective. We'll continue to be disciplined in our approach to pricing. And as always, we'll continue to drive manufacturing cost savings going forward, which enables us to drive to the gross margins within the range, I mentioned earlier.

At this time, there are no further questions. This concludes the question-and-answer session. I would now like to turn the conference back over to the company.

Thank you, everyone, who is joining us today for your interest and time, and we look forward to speaking with you again soon. Have a great day. Thank you.

Thank you. This now concludes today's conference.

So thank you all for joining us, as you see on the screen and here ahead. Joining me on stage, we have the Chief Executive Officer for Rambus, Luc Seraphin. Thank you so much for being here with us today.

So Rambus has over 35 years' experience in high-performance memory subsystems and is a fabless supplier of the industry's leading ICs and silicon IP solutions that advance data center connectivity, which helps solve the bottleneck between memory and processing. You know this better than I. So for the audience, maybe we can just kind of level set on the Rambus story and bring everybody up to date on how the business is doing thus far and maybe start with the outlook for '26?

Sure. As you said, we started 35 years ago working exclusively on memory interface technology, the technology that is required to connect processors with memory. And obviously, these technologies today are critical to data centers and to AI because that bottleneck between memory and processors is what is making these technologies more and more advanced. We started as a patent licensing company 35 years ago. We still have a Patent Licensing business. That Patent Licensing business is about $210 million a year, 100% margin. It's very stable. It's a cash generator for the company.

But that's not a business that we expect to grow in the long run. We've actually transformed the company into a company that delivers technology for these memory vendors. And we deliver that technology either in the form of silicon IP. We basically develop silicon blocks that we sell to semiconductor companies and these semiconductor companies integrate those blocks into their SoCs, their ASICs, their products. That business, the silicon IP business is about -- was $120 million last year and is growing 10% to 15% a year. And finally, we make ICs. We have a fabless semiconductor model. We started to make ICs, 7 to 8 years ago with 0% share. We started from scratch. That business this year, if we take the midpoint of our guidance for Q4, it is going to be about $340 million. And that business grew 40% over last year. So that's a very high-growth business for us. It's driven by the demand from data centers, whether they are AI-related or non-AI related. But it shows the strength of -- and the importance of memory in those data centers.

Yes. So speaking with memory, you've had some nice growth in the Memory Chip business. Maybe we can start with for that, just the overall size of the market and what are some of the competitive dynamics there and some of the strategic dynamics for you?

Sure. So the chip we develop are actually interface chips that sit on the memory modules and they interface between processors and memory. This is the simplest way to describe that. The market for that interface chip, we estimate at about $800 million a year, and that's the base. When the memory market moved from the DDR4 generation to the DDR5 generation, we found that you have more chips on the module than the simple interface chip. So we're adding SAM or TAM to what we address.

So in addition to this $800 million TAM, we have what we call companion chips, which are all the chips that we have to develop for that module. And these companion chips represent about $600 million, additional to the $800 million. And then these technologies are going to expand outside of the data center into high-end client systems, high-end PCs because we're going to find the same type of challenges from a technical standpoint there. That adds another $200 million of TAM.

And finally, there's a new technology that we may talk about later, which we call MRDIMM, its a technical term, but that adds another $600 million. So you have these layers of TAM expansion based on these technologies that we're going to go after.

Okay. And the transition to DDR5 brought about additional chip opportunities in the form of companion chips, as you're saying. Can you talk about your chip offerings today, the size of that market and the overall timing and expectations for both the revenue as well as market share expectations?

Sure. So when the market moved from DDR4 to DDR5, the industry decided that some of the functions that were performed on the motherboard in a computer had to move to the module, to the memory module because of the complexity. So when you move from DDR4 to DDR5, you're adding companion chips on the module in addition to the interface chip.

And as we said earlier, the interface chip TAM is $800 million. The companion chip TAM is $600 million. These chips, they are a power management chip, which is a chip with the role of managing power on that module, delivering a very stable power at different levels of voltage in a very controlled sequential way. There is a controller chip that we call SPD Hub, and there are two temperature sensors. So there's three types of new chips, four chips in total because we have two temperature sensors in addition to the controller chip. And as I said, it moves the TAM from $800 million to $1.4 billion.

Okay. As the industry transitions from AI training to AI inference. What is the Rambus content opportunity in an AI server maybe compared to a traditional general purpose server? And what does that mean for your chip opportunity? Does that have an impact on the product cadence or content?

Sure. So AI has been generally very good for us. There were questions at the beginning about whether AI servers would cannibalize traditional servers. But I think some of us did not realize that if you take an AI server, you have a traditional server in that same box. You have GPUs and HBM that a lot of people talk about. But next to it, you have a traditional DDR with a traditional x86 processor.

And the reason is to feed the HBM and the GPUs, you need to cache the data, prepare the data. It's all done in a traditional server. So AI has been a catalyst for us for the adoption of high-end traditional servers. So you have those that are growing.

In addition to that, you have the traditional servers where you don't do AI at all, which continues to grow because there's this refresh cycle. And inference is very, very interesting because AI started with machine learning. So it's very computer-intensive, very expensive. And the idea that you do inference on the same equipment is actually a very expensive idea.

So if you want to monetize AI, you have to move to inference. And inference are calling much more on traditional servers because it's simpler, it's on a smaller set of data, and that drives demand for traditional servers. So we see that evolution in a very positive way. AI training was a catalyst for the adoption of the technology and AI inference is a growth driver going forward.

Yes, a lot of tailwinds there. You mentioned this before, so I do want to come back to it, your MRDIMM solutions. So those offer a significant uplift in bandwidth and capacity. Can you talk about the significance of this technology and perhaps provide some additional color on the size of that market and the revenue expectations?

Sure. So we call it MRDIMM. And the idea here is servers, whether they are traditional servers or AI servers, they're hungry for more capacity and more bandwidth. You want to access more data faster. And that's a problem that the industry has always had.

And MRDIMM has been defined by a standard body called JEDEC, like all the products we make. And the idea is to double the capacity -- memory capacity on a single module by putting memory on both sides of the module, and doubling the speed of access to that memory by using special multiplexing techniques. And this has been defined by the industry. And the beauty is you can double the capacity and double the bandwidth, but you can use this on an existing infrastructure. You don't have to change the architecture of the server. So you can remove a standard module and replace it with an MRDIMM and you have double the capacity and double the bandwidth. So that's the significance of it from a technical standpoint.

From an opportunity standpoint, for us, it's really nice because it actually multiplies our content by a factor of 4, between a standard module in DDR5 today to an MRDIMM. And the reason the content is multiplied by 4 is because you have more memory and higher speed, you have a more complex and therefore, more expensive interface chip. You have a more complex power management chip. You still have two temperature sensors and an SPD Hub, but you also have to add 10 chips that did not exist on a standard DIMM that have to do with the multiplexing of the data. So if you take all of these together, the content on an MRDIMM is 4x the content of what we offer today on a standard DIMM.

And in terms of timing, it's going to be linked to the rollout of platforms from mainly Intel and AMD. So next-generation platforms from Intel and AMD, Diamond Rapids and Venice are going to be the first platforms that are going to be MRDIMM-capable, if you wish. And these platforms should ramp in the market by the end of '26, beginning of '27.

Okay. And there continues to be a lot of interest in CXL. And you previously have talked about investments in IP and the products. How do you see those opportunities developing?

So in CXL, we had two ways of approaching CXL. As part of our silicon IP business, where we build pieces of IP that we sell to people who build chips. We have a CXL offering, so people who build chips that have CXL interface can buy that interface from us or that controller from us.

And in that area, we have a large number of customers. We sell them license per use. And if they have -- if they use our CXL controller, they pay us a license to use that controller. We also had a CXL product development where -- and we have a CXL product, but we've not commercialized that product because we realize that CXL defines an interface. It doesn't define a chip. So throughout our silicon IP business, we realized that customers develop CXL chips that are custom to their customers. So it's almost a one-on-one relationship, one chip for one customer. So the market is very, very fragmented. It's an interesting market in terms of its total size, but it's a very fragmented market.

And the main use case for CXL was memory expansion, the idea to add memory to a processor. And we thought MRDIMM was a much more elegant way of doing that, for all sorts of reasons, technical reasons, but also due to the fact that MRDIMM, as we said earlier, is defined by the industry at the product level, and it uses the current infrastructure. So the implementation is much easier, much more elegant. So we continue to watch that area. And in summary, our CXL offering remains today in our -- as part of our silicon IP business, and we don't have a CXL product per se.

Okay. And sticking with the silicon IP business, can you give a brief strategy for the overview -- or an overview of that strategy, I should say, and a bit on what are those key drivers? And what does that long-term growth look like?

Yes. So in the silicon IP business, where we actually develop those pieces of IP that people use when they develop those chips, if you take a large semiconductor companies, they don't have the resources to develop every part of the chip. So they buy these IPs from people like us. We have a very focused laser-focused offering in silicon IP.

So we play in two areas. One, is security. Security is becoming really, really important in data centers, in automotive, in IoT because of all of the data that you have to deal with, you have to make sure it's never corrupted, whether it sits on a chip or whether it's being transmitted between chips. You have to have those security technologies to make sure the data is not corrupted. We are probably the largest silicon IP provider for security as an independent company because we invested in that business in 2011, a long time ago, way before people saw security as being so critical. So that's half of the silicon IP business, security.

The other half has to do -- it's very laser-focused as well with high-speed interfaces, everything that is required in new generations of servers and AI servers. So we've all heard about HBM memory. So we have HBM controllers, GDDR controllers. We have PCIe controllers, which are high-speed links that go into these chips. So that business in total is about $120 million, as we said earlier. It's roughly split half-half between security and high-speed interfaces. It's a business that grows 10% to 15% per year. And the growth drivers is really the rapid evolution of data centers and the need for more speed and more security for your data. So we're very focused, but the demand for these technologies is very high because of the demand for data in the data center.

Yes. Building off that as we kind of pull back a little and just the China exposure. We're hearing that EDA companies are being left out of that market a bit. Would you say that's a similar risk for your IP business?

The risk for us related to China remains very small. We actually have very little exposure to China in terms of customer base. It's less than 5% of our business. We monitor on a regular basis, whether there are any bands that apply to us, which is not the case. So we don't have any of these situations affecting us today. And if it did, the materiality would be very low given the low revenue we have from China.

Okay. I do want to go back to the Patent Licensing that you kind of clicked on for the overview. And that, again, has provided such a solid foundation for Rambus to grow and invest in both chips and the silicon IP business. Can you provide some additional color on those license?

Sure. So as I said earlier, the company started, we have really invented the DDR-type of technologies. That's why it's so widespread. So anyone who builds memory or anyone who builds a chip that has to interface to memory has to be a licensee of our technology.

Typically, the lifetime of a patent is 20 years. So once you have a patent that is relevant, its lifetime is 20 years, which is very nice. Typically, a contract on patent licensing is anything between 3 to 10 years, let's say, 5 years on average. So when we renew our patent license agreements, the patents that were relevant in the previous agreement, most of the time continues to be relevant in the next agreement. And the other thing that we kept at Rambus, we kept a group called Rambus Labs, who continue to work on generating new patents that we think are going to be relevant down the road.

So as I said, from a business standpoint, it's a flat business, $200 million, very high margin. But it also gives us a lens into the future because we have to look at what we think is going to be relevant, 10 years down the road. And that's a very nice thing about Rambus. If you look at the 3 businesses we have, Patent Licensing gives us a very long-term view about what's going to be relevant in terms of foundational technology 10 years down the road. The silicon IP business because we work with most of the semiconductor companies and they develop chips that are going to be in the market 3 to 5 years down the road, that gives us a very good insight about the trends in each one of those markets. And then we have our Chip business, which is growing very fast, where the road map is defined by the industry, which gives us comfort about investing our dollars because we know that the whole industry is behind those road maps.

And they all go together. The three main patent licensees that we had when we started the business, who are the three main memory vendors, have become our first three customers in terms of products. So it's been a nice transition for us. Our main licensees have also become our largest customers on the product side. So we've completely changed the relationship with them.

Yes. Well, and again, understanding those trends allow you to really "skate to where the puck is going to be". Maybe off that, obviously, your financial model is incredibly quality in the sense of it generates strong margins and high cash flow. Can you give a little picture on how you think about the long-term financial model and capital allocation?

Sure. So if you look at the 3 businesses, the Patent Licensing business is a 100% gross margin business, which is quite rare. The silicon IP business is about 95% gross margin. And the Product business is between 61% and 63% gross margin. The mix is going to change going into the future. We're going to continue to grow our product business, which is a lower -- relatively lower gross margins.

But we see some benefits on the leverage on OpEx because once you invest, there's a level that you don't need to increase to keep working on your road map. So we have an operating margin of about 40% to 45%, and we expect to keep that even though the mix is going to change.

In terms of cash, in the last 12 months, we've generated $300 million of cash from operation. So we generate a lot of cash. And our priorities are obviously to continue to invest in our organic growth with the dynamic in the data center, we have to develop more products. We talked about more content. So it means more products, and we have to develop them at a much higher pace. For example, in the DDR5 generation, we have to develop a new interface chip every year. In the DDR4, it was every 2 years. So we had to double that. So organic growth continues to be one of our focus.

We continue to look at acquisitions. We've made a few small acquisitions over the last year, 5 or 6 acquisitions, mostly in the silicon IP side. But that's something that we continue to monitor, as a vector of potential growth. And we finally return cash to our investors. We have a goal of returning 40% to 50% of our free cash flow back to our investors, and we've been on track for doing that over the last past years.

Great. I do want to pause here to see if there's any questions from the audience. We've got one there.

One question on Wafer Scale Engine. Is that an opportunity for you guys? Or do you see that as a threat coming from Cerebras? Wafer Scale Engine, Cerebras?

No, it's probably not a threat because I actually...

You don't know those guys?

No, no. We have a lot of questions about the threats of new processors that are competing with Intel processors or the new processors based on the ARM architecture or the threats coming from different types of memory like HBM.

And what we see is it's actually -- we've never seen any cannibalization there. Because we are fundamentally working on the interface, we're kind of agnostic as to what processors are being used, what core is being used in that processor? And as we said earlier, AI has created this kind of trigger or catalyst for the use of standard memory as well. So we always look at threats. I cannot comment on this particular one. But a lot of threats that we've seen have actually been catalysts for us because we're fundamentally at the interface. And what's built beyond that does not really affect what we do.

Thank you for your question. We have one more back here.

So you've been pioneering the silicon IP designs. Are you facing some competitive threats from fast followers? I'm thinking of eMemory in Taiwan, especially in the domain of security, hardware-based security versus software-based?

So we are mostly a hardware-based security company. Our root of trust is a hardware root of trust. We believe from the very beginning that hardware-based security is much more robust than software-based security.

And at the beginning of that business, by the way, we were competing with very strong software-based security companies that fell off the road. Yes, we have fast followers. But I would say that we with whole humility, we feel confident with our position in the security silicon IP business because of the history we have there. I would say that we do see fast followers. What we do see as well is, we compete a lot with internally-developed solutions. Some customers throughout the last few years have gotten into the habits of developing their own security. But more and more, they rely on external companies because, again, the pace here is really, really fast.

The industry talks a lot about quantum computing, for example. Quantum computing is going to give computers so much more power that some of the security protocols and security solutions that are out there today are going to be not good enough in the future. So for example, in our security offering, we have developed a quantum-safe core that allows our customers to be ready for when quantum computers are going to be out there trying to attack the security of their chips. So it's a fascinating area where we believe we keep having a lead.

Any observation on eMemory specifically or not particularly?

Yes. They're one of the competitors. But as we indicated earlier, we keep having a lead in that market.

We can follow up on that. I do want to be thoughtful of time. But thank you all for joining us, Luc. Thank you so much.

Welcome to the Rambus Third Quarter Fiscal Year 2025 Earnings Conference Call. [Operator Instructions]

As a reminder, this conference call is being recorded. I would now like to turn the conference over to Desmond Lynch, Chief Financial Officer. You may proceed.

Thank you, operator, and welcome to the Rambus Third Quarter 2025 Results Conference Call. I am Desmond Lynch, Chief Financial Officer at Rambus. And on the call with me today is Luc Seraphin, our CEO. The press release for the results that we will be discussing today has been filed with the SEC on Form 8-K. We are webcasting this call along with the slides that we will reference during portions of today's call. A replay of this call can be accessed on our website beginning today at 5:00 p.m. Pacific Time. Our discussion today will contain forward-looking statements, including our expectations regarding projected financial results, financial prospects, market growth demand for our solutions, other market factors, including reflections of the geopolitical and macroeconomic environment and the effects of ASC 606 on reported revenue amongst other items. These statements are subject to risks and uncertainties that may be discussed during this call and are more fully described in the documents we file with the SEC, including our 8-Ks, 10-Qs and 10-Ks. These forward-looking statements may differ materially from our actual results, and we are under no obligation to update these statements. In an effort to provide greater clarity in the financials, we are using both app and non-GAAP financial presentations in both our press release and on this call. A reconciliation of these non-GAAP financials to the most directly comparable GAAP measures has been included in our press release, in our slide presentation and on our website at rambus.com on the Investor Relations page under Financial Releases. In addition, we will continue to provide operational metrics such as licensing billings to give our investors better insight into our operational performance. The order of our call today will be as follows: Luc will start with an overview of the business. I will discuss our financial results, and then we will end with Q&A. I'll now turn the call over to Luc to provide an overview of the quarter. Luc?

Thank you, Des. Good afternoon, everyone, and thank you for joining us. Brands delivered a very strong third quarter with solid sequential growth and revenue above expectations. Product revenue led the way with a double-digit increase and growth that outpaced the market. This was driven by sustained market leadership in DDR5 products coupled with ramping contributions from our suite of new products. We also delivered another quarter of excellent cash from operations. highlighting the strength of our balanced business model, while we continue to execute on our strategic road map. Leveraging our core expertise in signal and power integrity our strategic focus on delivering complete solutions for high-performance memory subsystems positions us well amid strong secular trends in data center and AI markets. Turning to our businesses. I'm extremely pleased with the performance of our chip business. In Q3, we delivered another product revenue record at $93 million and marked our sixth consecutive quarter of growth. As a cornerstone of our success, our DDR5 RCD leadership and ongoing market share gains continue to fuel our top line growth. In addition, customer adoption of new products is progressing well, with initial production shipments now in motion. Looking forward, we expect our continued RCD market share leadership and increasing contributions from new products to drive full year product revenue growth of over 40%. -- our broad product offering, including chips for all GDX standard DDR5 and LPDDR5 modules supports the full spectrum of high-performance computing platforms in servers and client systems. Our full chipset solutions of our customers, not only the ease of one-stop shopping, but also the greater assurance of interoperability, which becomes ever more critical as the complexity of design raises alongside data rates. Through ongoing leadership in our CDs and growing traction across our portfolio of new products, we expect continued momentum and long-term growth. Turning to silicon IP. AI continues to drive design win momentum. The increasing pace and diversity of AI accelerator and networking IC designs is driving demand for high-speed memory interconnect and security IP. -- led by our best-in-class HBM 4, GDDR7, and PCIe 7 solutions are IP is critical to enabling the performance and security required by AI training and inference workloads. -- focused on providing our customers with differentiated features and performance for the most challenging applications, we see momentum across our portfolio of cutting-edge solutions, and we remain on track for our long-term growth targets. As we look ahead, the rapidly rising adoption of AI is driving continued server growth. Training and inference require massive compute infrastructure to support increasingly complex and diverse workloads. Notably, Agentic AI is emerging as a major catalyst for server demand, particularly for traditional CPU-based systems. This is helping to fuel the ongoing hyperscaler and enterprise refresh cycle, amplifying the growth in server unit shipments. In addition, the amount of memory per server continues to grow. AI workloads demand unprecedented levels of compute performance, driving increasing core counts and the need for more memory bandwidth and capacity. This translates to more deals per server, a higher data rates as well as the need for novel high-performance memory solutions and enabling technologies. MRDM is a great example of this as it leverages an innovative architecture to double the capacity in it versus standard RIM scaling the amount of memory per server also creates demand for increasingly sophisticated power management solutions that optimize the efficiency and quality of power delivery. We solve these complex problems for our customers with leading-edge products and are pleased to be on track to intercept compatible future generation systems with our complete industry standard MRDM and RDM chipsets. Going beyond servers, the release of each new client platform continues the trend of server class technologies waterfalling into AI PCs as performance targets continue to raise. This drives demand for faster memory and more module chip content. Leveraging our fundamental signal and power integrity building blocks our client chipsets are progressing well with growing customer traction, and we look forward to meeting this rising market need. The secular growth trend in data center as well as the rising performance requirements across the computing landscape driven by AI are highly favorable to Rambus and align directly with our long-term strategy. Our groundbreaking memory connectivity and power management solutions are foundational to enabling the next generation of AI and HPC platforms by advancing system memory bandwidth and capacity capabilities. Having identified the increasing technical demands of data-intensive applications as opportunities, -- we have developed a road map that builds on our leadership in signal and power integrity to enable robust high-performance memory subsystems. In closing, Q3 was a very strong quarter to solid financial results. Our continued product leadership in DDR5 and increasing momentum in new products are underpinned by the company's strong alignment with positive secular trends in data center and AI. This gives us confidence in our ongoing success and our ability to deliver long-term profitable growth. As always, I want to thank our customers, partners and employees for their continued support. And with that, I'll turn the call over to Dave to walk us through the financials. Des?

Thank you, Luc. I'd like to begin with a summary of our financial results for the third quarter on Slide 3. We are pleased with our strong Q3 financial results as we continue to execute on our strategic initiatives. As Luc mentioned earlier, -- we continued our market leadership position in DDR5 products and have started to see increasing contributions from our suite of new products. Our diversified portfolio continues to deliver strong results, which led to outstanding cash generation in the quarter of $88 million, which further strengthened our balance sheet. Our consistent ability to generate cash allows us to strategically invest in our product road map to drive our long-term growth. Let me now provide you a summary of our non-GAAP income statement on Slide 5. We Revenue for the third quarter was $178.5 million, which was above our expectations. Royalty revenue was $65.1 million the widening billings were $66.1 million. The difference between licensing billings and royalty revenue mainly relates to timing as we don't always recognize revenue in the same quarter as we bill our customers. Product revenue was $93.3 million as we delivered another quarter of record product revenue. This represents a 15% sequential increase and a 41% year-over-year growth driven by continued strength in DDR5 products and ramping new product contributions. Contract and other revenue was $20.1 million, consisting predominantly of silicon IP. As a reminder, only a portion of our silicon IP revenue is reflected in contract and other revenue and the remaining portion is reported in royalty revenue as well as in licensing billings. Total operating costs, including cost of goods sold for the quarter were $99.3 million Operating expenses were $64.6 million as we continue to invest in our growth opportunities in a disciplined manner. Interest and other income for the third quarter was $6 million. Using an assumed flat tax rate of 20% for non-GAAP pretax income, non-GAAP net income for the quarter was $68.2 million. Now let me turn to the balance sheet details on Slide 6. We ended the quarter with cash Cash equivalents and marketable securities totaling $673.3 million, up from Q2 primarily driven by strong cash from operations of $88.4 million. Third quarter capital expenditures were $8.4 million where depreciation expense was $8 million. We delivered $80 million of free cash flow in the quarter. Let me now review our non-GAAP outlook for the fourth quarter on Slide 7. I -- as a reminder, the forwarding guidance reflects our current best estimates at this time, and our actual results could differ materially from what I'm about to review. The economic environment remains a dynamic environment, and we continue to actively monitor this situation. In addition to the non-GAAP financial outlook under ASC 606, we also provide information on licensing billings, which is an operational metric that reflects amounts invoiced to our licensing customers during the period adjusted for certain differences. We expect revenue in the fourth quarter to be between $184 million and $190 million. We expect royalty revenue to be between $59 million and $65 million and licensing billings between $60 million and $66 million. We expect Q4 non-GAAP total operating costs, which includes COGS to be between $103 million and $99 million. We expect Q4 capital expenditures to be approximately $10 million. Non-GAAP operating results for the fourth quarter is expected to be between a profit of $81 million and $91 million. For non-GAAP interest and other income and expense, we expect $6 million of interest income. We expect the pro forma tax rate to be 20%, with non-GAAP tax expenses to be between $17.4 million and $19.4 million in Q4. We expect Q4 share count to be 109.5 million diluted shares outstanding. Overall, we anticipate the Q4 non-GAAP earnings per share range between $0.64 and $0.71. Let me finish with a summary on Slide 8. In closing, our team delivered strong third quarter financial results, setting another record for product revenue and continued strong cash generation. Our robust balance sheet continues to allow us to invest in market expansion opportunities. Our product portfolio, including silicon IP and chip solutions is strategically aligned to capitalize on the growing opportunities in data center and AI. Before I open up the call to Q&A, I would like to thank our employees for their continued teamwork and execution. With that, I'll turn the call back to our operator to begin Q&A. Could we have our first question?

The first question comes from Tristan Gerra with Baird.

You recently quantified the MRDIMM TAM opportunity. Is it fair to assume you can replicate the market share with MRDIMM that you have currently in DDR5? And also, when do you think you can fully realize the TAM that you quantified for MRDIMM? Is that something that we could envision for '28?

Tristan, thank you for your question. We're very pleased with the progress we're making with the MRD development. We do believe that with time in the long run, we can reach similar market share as we have with the DDR market share we currently have on DDR5. The timing of that really depends on the rollout of platforms from our main partners on the CPU side, Intel and AMD. But to the extent that they roll out their platform, I think it's fair to say that we're going to ramp in large volumes towards the very end of '26 and probably '27. So '28 is probably a good time to look at this type of market share. But the other thing I would add regarding MRDIMM is that it's a much more complex system. And because of the system requirements, we will need a tight coupling of the chips on that MRDIMM. So there's an opportunity for us to have more content as the interoperability of all those chips on that MRI is going to become very critical.

Great. And then as a quick follow-up regarding the recently announced Ethernet scale-up networking architecture at OCP, does that provide opportunities for Rambus on the licensing side?

Thank you, Tristan. Our SIP portfolio is very focused on high-speed memory and high-speed interconnect and security. Certainly, with our networking customers and our memory customers, we are on the leading edge of technology, whether it is on GDDR and HBM on the memory side or PCIe 7 on the networking side. What we've seen recently is an acceleration of demand for the latest technology. The transition from PCIe 5 to PCIe 7 is moving very, very fast, and that's certainly an opportunity for us.

The next question comes from Aaron Rakers with Wells Fargo.

I've got a couple as well. I guess, first, kind of sticking on the technology evolution of what we're seeing in some of these processors. There's a lot of news recently around the move towards [ SOCAMs ] and [ SOCAM2 ], in particular, is getting JEDEC standardization. Can you help us maybe think about Rambus' opportunity set in [indiscernible] modules and when maybe you would expect to see that? And any kind of framing of kind of the dollar content opportunity on those modules?

Thank you, Aaron. The first thing I would say is that we are excited to see the emergence of these new architectures that actually play on our strength and focus in signal integrity and power integrity. As you know, the first attempt at [indiscernible] didn't work that well. And as a reminder, the attempt was actually to take benefit of the low power and high bandwidth of LPDDR, but to put this on modules. And when you put this on modules, you actually break the signal integrity and the reliability of the system. So we are pleased to see those efforts going into JEDEC because I think the industry is eventually going to resolve those issues. And as a reminder, as we said in our remarks, -- we currently have solutions for all JEDEC module systems, both for LPDDR and DDR and both for clients and servers. So the fact that it's going through JEDEC is actually a good news for us. There will be opportunity for us, certainly opportunity for the SPD Hub chip. There's going to be some development on voltage regulators. But as we said, power management is also something we're focusing on. So we see this as an opportunity. We don't expect the volumes to be very high. These things actually go into system-on-chip solutions, very tight systems where the volumes are not necessarily very, very high. It's early to say what the content is going to be. But that's certainly an area we're going to play in given that the company focuses on both the signal integrity and power integrity. And the fact that it's moving to JEDEC is good news for us.

Yes. Very good. And then maybe as a follow-up to that answer is on the PMIC side, your product chipset business did really well this quarter, growing over 40%. And it looks like that appears to be sustainable as we look forward. How do we think about the opportunity of PMIC? How much does that represent of your product chipset business today? And maybe unpack of how quickly you're seeing that ramp looking forward?

Thank you, Aaron. The way we look at these products is we have a whole suite of new products, including the PMICs and the PMIC is actually a suite of products. If you remember the product announcements we've made over the last few years, we have the first generation of PMIC family announced in Q2 of last year. the clock driver announced in Q3. Then we have the second generation of PMIC announced in Q4 as well as Gen 2 RCDs and MRDIMMs. And this year, in Q2, we announced the family of PMICs for the client space. So what you see is we do have a whole suite of products that are not -- that are companion chips. We're pleased with the progress this year. In Q2, these chips represented low single-digit contribution to our product revenue. As we indicated, Q3 was on track with mid-single digit. And in Q4, it's going to be mid- to high single digits. So in aggregate, we're pleased with the momentum. But it's not going to be a step function. We have different stages of qualification and preproduction on different modules on different platforms, current platforms and future platforms. We have products that are in early qualification. We have products that are in preproduction and we have products that are in full production now. But there's a strong momentum. And as these products percolate through the ecosystem, we do believe that we're going to continue to see growth. Now specifically to PMIC, what we have observed is that we have lots of success with a very high-end PMICs. There's a lot of excitement there. They are the most complex PMICs to make, but they're also the ones that are showing the best performance compared to our competition. So that's exciting for us. These very high-end PMICs are going to be linked to next-generation platforms with AMD and Intel, but we certainly have the early generation of PMICs also rolling out in the market. So difficult to separate PMIC from the rest. As I said, we have many products at many stages of development with our customers. But what we see is very strong momentum to grow that revenue quarter-over-quarter given the progress we're making.

The following comes from Gary Mobley with Loop Capital.

Let me extend my congratulations to the solid results. And I want to start asking about any sort of supply chain considerations first on your side. Do you see any extension of the order lead times that your customers are seeing as they place an order with you given any sort of constraints that TSMC may have? And then away from you, are you seeing any sort of impact on the market any sort of constraints on high-capacity server DIMMs or the DRAM to support that market, considering most of the memory IDMs are prioritizing HBM at this point?

Gary, it's Des here. Thanks for your question. Like others within the industry, we are carefully monitoring the supply situation. With regards to Rambus, I was pleased that we were able to grow our inventory in the third quarter. We grew inventory by about $6 million, which will support our growth in Q4. In addition, I would highlight that we've not seen any notable buildup of customer inventory in the sort of third quarter. Really looking at our own supply chain and manufacturing, in terms of front-end manufacturing, it is important to note that we are not on leading-edge technology nodes. And on the back end, we continue to have strong long-term relationships with our manufacturing partners. And we do see some pockets of tightness, and we continue to work with our partners to improve the lead times there. And looking at Q4, I would expect to see a slight increase in our own internal inventory to support customers' Q1 2026 demand. I would say, overall, we have a robust supply chain, which has enabled our strong product revenue growth, and we'll continue to work with our manufacturing partners to support our growth objectives going forward.

Got it. That's helpful. In the RCD market specifically, I would assume that you're running about or above 40% market share. Do you see a natural cap there given that this is more or less a sort of a 3-supplier market, maybe 2 additional suppliers in sort of the nascent stages of their development. Do you see that as a natural cap? Or do we see maybe 45%, 50% market share on the horizon?

Thanks, Gary. What we said last year in 2024 is that on the DDR5 generation, we were in the early 40% market share. We actually disclose market shares once a year because of some fluctuations we have every quarter. But if you look at the current outlook for this year, it looks like we're going to continue to grow share. The market for servers or DIMMs has increased mid- to high single digit. And as Des indicated in his prepared remarks, we grew 40% year-over-year. So we have certainly gained share this year on this market. And we still believe we can continue to gain share. We always have the objective of 40% to 50%. So there's room to gain share. We also early in the DDR5 cycles, it's been 3 years in, and we expect the DDR5 cycle to last about 7 years. So we do expect to continue to have the possibility of winning shares. The other thing is I think that when the products become more complex and the interoperability becomes more complex as well, because we have a complete chipset that's going to help us continue to gain share. Certainly, there's going to be a cap, but we don't see the cap in the near future at this point in time.

The next question comes from Mehdi Hosseini with Susquehanna.

This is for the team. I think it would be very helpful if you could remind us how to think about different TAM and give us an update. In the past, we have talked about the buffer chip companion, CXL [indiscernible] IP perhaps with the diversification in the DRAM with the inclusion of MRD, there are some changes there. And in that context, it will be great if you could give us what the TAM will look like, let's say, 2 or 3 years from now? And I have a follow-up.

Yes. Thank you, Mehdi. So we'd like to separate the I would say, the product from the silicon IP. On the product side, we estimate the time for the RCD market to be around $800 million. then you add to this $600 million of companionships, half of it being power management chips and the other half being the other companion chips. And then -- and you can think about the market growing mid- to high single digits in aggregate. There's additional, I would say, tailwinds in to this with the increase of number of channels and the increase of number of DIMMs per channel. But this will translate into -- not into a step function, but some tailwinds to that TAM. Then in addition to that, we see a TAM of about $600 million for the MRDIMM itself, which adds to this. But the MRDIMM as we discussed earlier, is not going to hit the market before very late in 2026, '27, depending on the rollout of the platform from AMD and Intel primarily. Now if you turn to the Silicon IP business, it's hard to have a TAM number for the Silicon IP business. What I would say is that as part of our portfolio, we are at the center of what matters for AI. Our portfolio is focused on PCIe 7 and the future generations on HBM 4 and future generations and on GDDR and future generation. So there's a pool for design staff on all of these IPs. But it's hard given the type of business model on the licensing side, it's hard to estimate a stand for this. But what I would say is that we are on track to meet our growth targets in that business of a double-digit growth.

Okay. Great. Just a quick follow-up here. Should I assume that MRDM margin is comparable to product? Or would it be more like an IT type of the margin?

Mehdi, it's Des here. In terms of the MRDIMM, this is obviously a chip product that we will be selling here. What I would say is I would keep it within the same sort of margins of our product business. Our Long-term goal of that business is 68% to 65%, and I would keep the MRDIMM margins within that. We continue to produce strong margin results on the chip side, and we're really pleased with the portfolio that we have.

Sure. Great. And my second question has to do with just looking beyond December and seasonality. I'm under impression that when it comes to servers and companionship, maybe there could be better than seasonal trend into early part of the '26. And I wanted to see how you're looking at those trends. And I'm not asking for a guide, I'm not asking for a guide. I am not asking for a specific revenue guide but just trend -- [indiscernible] that seasonal trend that I see in the server and AI will also apply to Rambus.

Thank you, Mehdi. We do see the market for servers to continue to grow between mid- to high single digits going into next year. There's some tailwinds, as we said, because of the growth of inference, for example, or Agentic AI, that's going to create tailwinds for standard CPU types of solutions. But we do see a growth between mid- and high single digits for the server market next year. Typically, in Q4, we have our customers being prudent with the inventory before the year-end and that happens every year, but that's included in our guide for for Q4 that we just gave. And things are going to be back on track in Q1 of 2026. We keep saying that one of the reasons we don't guide beyond 1 quarter is that things are changing very, very fast, and visibility is not the best. But we do see all the favorable tailwinds for our business going into 2026.

The next question comes from Kevin Cassidy with Rosenblatt Securities.

Congratulations on the great results. Just looking at the market, the DRAM market and maybe Gary touched on it with the lead time stretching out and prices going up, is there any concern at all of servers de-specking as the price of DRAMs go higher? Or is the need for DRAM and AI allocations so strong that there won't be a DEXTENZA-specking.

Well, that's a good question for the memory vendors. I would say that historically, we've been kind of agnostic to DRAM pricing. We -- I think what the industry is going to have to go through is to deal with the growth of demand for data centers in general, and to have some arbitrage between the different types of memory. But I don't think that the DRAM pricing is going to have any impact on the demand for our products at Des. Can you hear me?

Yes. Kevin, I would just add in the fact the inventory levels within the channel continue to remain sort of lean. When I look at inventory in Q3 versus Q2. And this is of our chips that our customers are holding. We saw no notable inventory build. And I would really put that down to 2 factors. One, it's been the multiple generations of DDR5 in the market and really the legacy overhang of overordering of DDR4 inventory from a couple of years ago. So I would say the inventory position just now is lean in terms of our sort of chips.

Right. Okay. Great. And maybe just along that, you [indiscernible] here 2 years into this DDR5 cycle and maybe it's 3 generations of DDR5 modules. What's the bell curve like of your shipments? And what is that doing to ASPs as you go forward?

Kevin, it's Des. We've been really delighted with how we've been able to execute on the DDR5 cycle. We're in the middle of a fast paced DDR5 transition with multiple generations in the market today. I would say that in Q3, the predominant of our shipments was the second generation of DDR5 with growing and [indiscernible] production volumes of the third generation coming into the market. And as I look ahead into sort of Q4, I would still expect the predominance to be the second generation which really growing contribution of the sub generation into the market. In terms of pricing, what we've talked about in the past is when we move from 1 generation to the next generation. We do see a bump up in sort of pricing, which is obviously beneficial for us there. And we'll continue to sort of see that benefit going into the numbers. We saw the benefit in the gross margin outlook in the third quarter on the [indiscernible] chip side, which increased about 300 basis points, which was really a combination of the product as well as continued manufacturing savings coming into the models. So overall, we're really pleased with how we're executing on the DDR5 generation. And irrespective of what fourth generation is ramping into the market through our early investment in continued leadership. We have confidence in our overall market share and leadership position.

Thank you. The following comes from Nam Kim with [ Arete ] Research.

I want to ask about outlook for CXL. There are a lot of perspective on how this market develops especially with the CXCL 3.1 expected next year. And your competitor like [ Montag ] becoming increasingly aggressive on the controller side. At the same time, greater adoption of MRD in the future could [indiscernible] memory capacitor constrained. So can you share your view on how you see [indiscernible] market evolving and what the Rambus strategy is in terms of controllers or other engagement in this space?

Nam, we have 2 players or 2 possible players in CXL. One is on the silicon IP business. We do have CXL controllers of different generations, and this has been part of this focused portfolio talking about where we do have traction, a lot of people developing chips need a CXL interface, and they have the possibility of buying that interface from us. So this has been 1 of the driver vectors of our growth in Silicon IP business. But what we have observed is that every 1 of our customers tends to develop a bespoke solution for one, sometimes only 1 or 2 customers. So the chips that use the CXL market is very fragmented. That's how we look at it. And although we did have and we do have a CXL product development, we believe at this point in time that it does not make economic sense to actually roll out that product in the market because what we noticed is that we would have to develop a specific chip for a specific customer, who themselves will have a specific customer as well. So we'd rather play on the SIP side with CXL. so what I would say is that CXL is very exciting in terms of being an interface that is accepted by everyone. But for us, it's not that exciting in terms of products. And we do believe that the usage model that is the most promising is actually memory expansion. And to your question, a very good question. The MRD [indiscernible] that because it uses the current infrastructure of standard servers. And just by using this MRD type of architectures, we can double the capacity and the bandwidth using that same infrastructure. So that's the option we've taken at this point in time. As the market develops, as we've done in the past, we can [ tilt ], but at this point in time, this is where we are.

The following comes from Kevin Garrigan with Jefferies.

Let me echo my congrats on the results. On the MRDM opportunity, you talked about customers starting qualifications. I mean is there anything more that you need to do or can do to kind of help yourselves capture share there? Or is it pretty much all in the customers' hands at this point?

It's in customers' hands, our hands and the hands of the people who deploy the platforms like Intel and AMD because they have to be ready with their platforms as well. But I would say, on our hand, what plays in our hand is really the fact that we have a complete chipset for MRDIMM. And that's critically important because when you double the capacity and you double the bandwidth that interoperability is critical to the MRDIMM actually working. And I think that customers are going to be looking at their suppliers like us to really help them not only on the development of the chips but also on the testing of the whole platform, given how compact is going to be and how fast it's going to have to run at. So this is what I think is going to play in our hands. The fact that we have invested for a long time in signal integrity and power integrity allows us to have a complete chipset and having a complete chipset is going to help us with interoperability testing with our customers.

Yes. Got it. Got it. Okay. That makes sense. And then just as a quick follow-up, going off of a previous question in your silicon IP business. You guys are doing well in HPM, but can you just talk about how traction has been with PCIe 7 and secured IP in that business?

I'll start and let Des jump in. Typically, we don't split these things. But at a high level, security is about 50% of our business and between controllers, memory controllers, the PCI controllers, that's the other 50%. I would say security is widespread in terms of its application, it's really into lots of applications with lots of customers in very different markets. PCIe and HBM, we tend to work with a large number of customers who are much smaller and we tend to work on the bleeding edge solutions for these. So we mentioned HBM 4 and PCIe 7. So we typically work with large customers who we need to develop the the latest and fastest solution mostly for the data center and the AI market. So it's a different dynamic there. Higher -- typically, we have a higher ASP, longer-time development with the bleeding edge solution for memory and PCI. It's a much broader and faster cycle in the [indiscernible] side. That's the way to look at it.

The final question is a follow-up from Aaron Rakers with Wells Fargo.

Thanks for the follow-up question. Just kind of thinking back again to the architecture evolution in this demand that you're seeing. When you guys look at your RCD business today, how do you assess kind of the number of channels today that you're shipping into on a per socket or per CPU basis, and how that has evolved and whether not moving from 8 to 12? And do you see 12 going to 16 channels as we look out into '26.

Thank you, Aaron. Certainly, AI workloads need more memory than standard types of applications and more bandwidth. So the very fact that the industry is converging to 12 channels is good. But remember, the -- it's only lately that Intel moved to 12 channels. So it's going to have a, I would say, a modest impact, but positive impact. We do see these memory, these CPU vendors announcing the 16-channel solution, and that's going to be necessary. There's talk also no plans of going beyond maybe to 20, but the issue is you cannot just add channels after channels. It creates constraints on the packaging design, the chip designs. So I see there's going to be a limitation there. But that's certainly a tailwind for us. That's going to help, as we said earlier, continue to grow our product business.

And on that channel discussion, how does that work with MRDIMMs?

So the MRDIMMs is going to intercept given the next generation of platforms on AMD and Intel. These next-generation platforms and AMD and Intel [indiscernible] announced 16 channel. But MRD is a very dense solution. So the number of DIMMs per channel is going to be the question. But these new platforms for Gen 5 are going to be around 16 per -- 16 channels per CPU, and that's the generation that Intercept MRDIMMs.

I will now pass it back over to Luc for closing remarks.

Thank you to everyone who has joined us today for your continued interest and time. We look forward to speaking with you again soon. Have a very good day. Thank you.

Thank you. This now concludes today's conference.

Welcome to the Rambus Second Quarter Fiscal Year 2025 Earnings Conference Call. [Operator Instructions] As a reminder, this conference call is being recorded. I would now like to turn the conference over to Desmond Lynch, Chief Financial Officer. You may begin your conference.

Thank you, operator, and welcome to the Rambus Second Quarter 2025 Results Conference Call. I'm Desmon Lynch, Chief Financial Officer at Rambus. And on the call with me today is Luc Seraphin, our CEO. The press release for the results that we will be discussing today has been filed with the SEC on Form 8-K.

We are webcasting this call along with the slides that we will reference during portions of today's call. A replay of this call can be accessed on our website beginning today at 5:00 p.m. Pacific Time. Our discussion today will contain forward-looking statements, including our expectations regarding projected financial results, financial prospects, market growth demand for our solutions, other market factors, including reflections of the geopolitical and macroeconomic environment and the effects of ASC 606 on reported revenue amongst other items.

These statements are subject to risks and uncertainties that may be discussed during this call and are more fully described in the documents we file with the SEC, including our 8-Ks, 10-Qs and 10-Ks. These forward-looking statements may differ materially from our actual results, and we are under no obligation to update these statements. In an effort to provide greater clarity in the financials, we are using both GAAP and non-GAAP financial presentations in both our press release and on this call.

A reconciliation of these non-GAAP financials to the most directly comparable GAAP measures have been included in our press release, in our slide presentation and on our website at rambus.com on the Investor Relations page under Financial Releases. In addition, we will continue to provide operational metrics such as licensing billings to give our investors better insight into our operational performance.

The order of our call today will be as follows: Luc will start with an overview of the business. I will discuss our financial results, and then we will end with Q&A. I'll now turn the call over to Luc to provide an overview of the quarter. Luc?

Thank you, Des. Good afternoon, everyone, and thank you for joining our second quarter conference call. Rambus delivered a very strong second quarter, exceeding expectations for both revenue and earnings, while continuing momentum in our growth initiatives. This achievement was driven by our memory interface chip business outpacing the market with 43% year-over-year growth, another quarter of record product revenue. The strong performance highlights our sustained leadership in core DDR5 products as we continue to execute on our strategic road map of Signal and Power Integrity solutions and to drive the adoption of our new products. .

We also generated record cash from operations of $94 million, showcasing the efficiency of our execution and the robustness of our business model. Our balanced portfolio and diverse revenue streams across chips, silicon IP and patent licensing position us exceptionally well in the market. They also provide stability in a dynamic macro environment and enable our continued product investment to drive long-term growth.

Our chip business continues to be a key growth engine for the company, with Q2 marking our fifth consecutive quarter of product revenue growth. As I mentioned in my opening remarks, we delivered a historic quarter of record product revenue. Our strength in DDR5 continues to be a cornerstone of our success, with increased sales of our core products driving above-market growth. Looking forward to Q3, we expect our ongoing LCD market share leadership, combined with early contributions from new products to drive double-digit sequential product revenue growth.

We have growing traction for the record number of new products introduced throughout last year with chips progressing through the respective stages of customer qualification and adoption. Additionally, we remain actively involved in the definition of future generation products with the industry. As we look further into the future, we are also very pleased that our industry standard MRDIMM chipset is advancing on schedule and we're excited about its role in meeting the growing memory performance requirements of next-generation server workloads.

Going beyond servers, we recently launched our client memory module chipset for AI PCs. With that introduction, we are proud to now offer chipsets for all GTx-standard DDR5 and LPDDR5 modules. Our client chip solutions waterfall our proven server class technology into new applications and extend our reach into next-generation high-performance PCs, opening up a growing market opportunity in the coming years. Our expanding product offerings support the next wave of high-performance computing platforms in servers and client systems.

Through ongoing leadership in our cities and growing traction across our portfolio of new products, we expect continued momentum and long-term growth. Turning to silicon IP. We delivered solid results in Q2, and we remain on track for long-term growth. AI and data center applications continue to drive strong demand for our high-speed memory and interconnect IP as well as our security IP.

Our IP solutions are foundational to enabling the performance and security required by next-generation accelerated computing ICs. We're seeing strong demand and design win momentum across our portfolio, led by a best-in-class HBM4 and PCIe 7 solutions. Now as we look ahead for the company, the data center will continue to undergo profound transformation, driven by exponential growth of AI workloads and the increasing complexity of high-speed performance computing.

Across the ecosystem, the shift towards scalable heterogeneous, compute architectures is accelerating demand for novel high-performance memory solutions and enabling technologies. These trends align directly with Rambus' long-term strategy. We are strategically focused on advancing system memory bandwidth and capacity through groundbreaking memory, connectivity and power management solutions.

These capabilities are foundational to enabling the next generation of AI and HPC platforms. We have built a road map that addresses the increasing technical demands of data-intensive applications. Our leadership in signal and power integrity core to enabling robust high-performance memory subsystems places us at the heart of this transformation. With our strong balance sheet and ongoing focused investment, Rambus is poised to capitalize on these secular growth trends.

In closing, Q2 was a standout quarter for Rambus. We achieved excellent financial results, delivered record product revenue and continue to execute on our road map. We are excited to enter the second half of the year with strong momentum, and we expect another quarter of record product revenue with double-digit growth in Q3. Our leadership in DDR5, increasing customer traction for new products and strong business model position us well for continued success and long-term profitable growth.

As always, I want to thank our customers, partners and employees for their continued support. And with that, I'll turn the call over to Des to walk through the financials. Des?

Thank you, Luc, I'd like to begin with a summary of our financial results for the second quarter on Slide 3. We delivered a strong quarter, exceeding our expectations for both revenue and earnings. Our chip business continued to drive our growth as we delivered record results marking our fifth consecutive quarter of product revenue growth.

In addition, our diversified portfolio generated record quarterly cash from operations of $94 million. Our ability to consistently generate cash is a key aspect of our strategy and enables us to continually invest in initiatives that fuel our long-term growth. Let me now provide you a summary of our non-GAAP income statement on Slide 5.

Revenue for the second quarter was $172.2 million, which was above our expectations. Royalty revenue was $68.6 million while licensing billings were $66.4 million. The difference between licensing billings and royalty revenue mainly relates to timing as we do not always recognize revenue in the same quarter as we bill our customers. Product revenue was $81.3 million as we delivered another quarter of record product revenue. This represents a 7% sequential increase and a 43% year-over-year growth, driven by continued strength in DDR5 products.

Contract and Other revenue was $22.3 million, consisting predominantly of silicon IP. As a reminder, only a portion of our silicon IP revenue is reflected in Contract and Other revenue and the remaining portion is reported in royalty revenue as well as in licensing billings. Total operating costs, including cost of goods sold for the quarter were $93.2 million.

Operating expenses of $60.4 million were in line with our expectations. Interest and Other income for the second quarter was $4.8 million Using an assumed flat tax rate of 20% for non-GAAP pretax income, non-GAAP net income for the quarter was $67.1 million. Now let me turn to the balance sheet details on Slide 6.

We ended the quarter with cash, cash equivalents and marketable securities totaling $594.8 million up from Q1, primarily driven by record cash from operations of $94.4 million. Second quarter capital expenditures were $10.4 million, while depreciation expense was $7.4 million. We delivered $84 million of free cash flow in the quarter. We consistently deliver value to our stockholders as we continued our stock repurchase program in the quarter.

Let me now review our non-GAAP outlook for the third quarter on Slide 7. As a reminder, the forward-looking guidance reflects our current best estimates at this time, and our actual results could differ materially from what I'm about to review. The economic environment remains a dynamic environment, and we continue to actively monitor the situation. In addition to the non-GAAP financial outlook under ASC 606, we also provide information on licensing billings, which is an operational metric that reflects amounts invoiced to our licensing customers during the period adjusted for certain differences.

We expect revenue in the third quarter to be between $172 million and $178 million. We expect royalty revenue to be between $57 million and $63 million and licensing billings between $58 million and $64 million. We expect Q3 non-GAAP total operating costs, which includes COGS to be between $98 million and $94 million. We expect Q3 capital expenditures to be approximately $12 million.

Non-GAAP operating results for the third quarter is expected to be between a profit of $74 million and $84 million. For non-GAAP interest and other income and expense, we expect $5 million of interest income. We expect a pro forma tax rate to be 20% with non-GAAP tax expenses to be between $15.8 million and $17.8 million in Q3. We expect Q3 share count to be 108.5 million diluted shares outstanding.

Overall, we anticipate the Q3 non-GAAP earnings per share range between $0.58 and $0.66. Let me finish with a summary on Slide 8. In closing, I am pleased with our strong financial results and ongoing execution. Our diversified portfolio and disciplined business model continues to drive profitable growth with strong cash generation. Our robust balance sheet allows us to invest in market expansion opportunities in the data center and AI while consistently delivering value to our stockholders. Before I open the call up to Q&A, I would like to thank our employees for their continued teamwork and execution.

With that, I'll turn the call back to our operator to begin Q&A. Could we have our first question.

[Operator Instructions] Our first question comes from the line of Aaron Rakers with Wells Fargo.

I'll just ask my question and my follow-up together here. I guess, first on the product revenue line, strong growth, up 43.5% year-over-year. I'm curious, Luc, how do we think about the contribution from the RCDs, your positioning of -- I think your target's been 40% market share in D5. And where we're at as far as seeing the ramp of the PMIC opportunity? And then as the follow-up real quickly, can you just remind us again, as we think about Granite Rapids from Intel, from a CPU perspective, and we look at the road map going forward, is the expectation that we see continual memory channel expansion with next-generation platforms, i.e., moving from 12 to 16 and so on going forward.

Thank you, Aaron. To your first question, yes, we're very pleased with the growth of our product business with this 43% year-over-year growth in the second quarter. RCD remains very strong for us. And our belief is that we continue to gain share with the expansion of DDR5 in the market. we were slightly above 40% share at the end of 2024, and we expect to continue to gain share this year. and we do start to see the contribution for new chips, power management chips, but all the chips that we're introducing to the market.

It's still modest. It represents low single-digit contribution to the product revenue in Q3, but it's going to grow to mid- to upper single-digit contribution in Q3, sorry, is low single digit in Q2. And we do see momentum there. So modestly we do see momentum across the board. And as we said, we have different stages of qualification and adoption of these different products in the market.

we feel very comfortable with the momentum there. With respect to the different platforms, our partners continue to roll out platforms. We do sell chip ahead of the platform deployment. So to the platform you mentioned, we're starting to see volume shipments of products on the RCD side. We do believe that in addition to the Intel platform, AMD and the ARM-based platforms are also going to roll out products that will create demand for our DDR5 RCD chips.

And the fact that these platforms are transitioning from 12 channels to 16 channels is also going to create further demand for DDR5 in the quarters and years to come. So that's the good news for us.

Our next question comes from the line of Gary Mobley with Loop Capital.

I had some questions about the PC market. I know it's not what everybody is focused on. But if I'm not mistaken, your newly introduced PMIC products are geared towards Panther Lake and with that launch imminent, can you share with us whether or not you've got any visibility into the PMIC sales into the PC market ramping this year or next? And are you generating yet any [ client clock ] driver revenue from the PC market?

Thank you. Gary, yes, as we said in earlier calls, we do see the requirements that we initially or historically saw in the data center flowing into high-end PCs and the need for the equivalent of the LCD or the equivalent of the power management chip flowing into the high-end PC market. .

So we introduced the clock driver last year, and we are starting to see modest traction. Modest fraction, not because the product is not successful. It's just the market is limited at this point in time. it really targets the very high-end speed layer of the market. And over time, it's going to slow down or the segments of the market. We were with the reception of our PMIC products into the data center. And that's why we announced PMIC products for the client market, a Gen 2 PMIC for deals in the client market as well as a [indiscernible] solution for the PC market.

So we're planting the seeds in a market that we think is going to be very fertile going forward. But that's going to address the high-end PC markets for us and then slow down. So we do expect the contributions from this client market to start to be visible in 2026. Well, this year, we're going to just see the initial shipments of qualification and preproduction orders.

Just a follow-up, I wanted to quickly ask about inventory. It appears as though you're dollars of inventory are getting lean and days of inventory, especially lean. And so the reason I bring this up is, are your lead times extending? And if they are, is there a motivation for your memory customers to start to maybe sort of ensure or hedge against that in the form of higher inventory?

Gary, it's Des here. That's a good question. Our inventory levels in Q2 come down to about 120 days, which was mainly driven by more finished goods inventory at the end of the quarter. It is important to note that our inventory holding at June 30, it's just a snapshot in time and really based upon our current view of demand. We will have sufficient inventory to support our customers' demand through at the end of the year. We do have long-term relationships with our supply chain partners, and they are fully supportive of our growth plans going forward into 2026 and beyond.

If we look ahead, given [indiscernible] expanding product portfolio and strong cash generation, we are comfortable with holding more strategic inventory on our balance sheet. And this is something we will definitely endeavor to do here over the next couple of quarters. As it relates to lead times, I would say that they remain within sort of normal levels and consistent with sort of [ prior quarters from there ], Gary. Thanks, .

Our next question comes from the line of Kevin Cassidy with Rosenblatt Securities.

Congratulations on the great results. The AI ASIC market is exploding in the -- it's called the XPUs. Can you say how that ASIC market might be changing the demand for your silicon IP Thanks, Kevin.

Yes, sure. What we see with the AI market exploding in the Americas, so the XPU solutions, ASIC solutions, is that the need for very high-speed connectivity and the need for very high-speed memory interfaces increases and accelerates and that translates for us into an acceleration of our development for solutions such as [ HBM4, HBM4E ] as well as PCIE7. So we are engaged with customers. This market tended to be quite -- it's a bit like the RCD market. Everything is accelerating.

But we do have several engagements on these leading-edge technologies on the HBM4 and PCIE7 in particular, as well as for the security solutions. The need to actually secure data when it sits into those chips or secure data when it moves around between those chips is becoming critically important. So that's also giving traction to the sales of our silicon IP in the security area.

Okay. Great. And maybe a more mundane discussion is that there have been announcements for DDR4 end of life. Does that change anything for Rambus? Or I guess a couple of years ago, we had an inventory issue. So I guess that's out of the way now. But what does it mean going forward?

It doesn't change much for us. DDR5 sales remain very limited. And this has been our message for several quarters now, and we don't see that picture changing. We do see slowly inventories going down in the market. We hear about the last time buy orders we expect DDR4 demand to remain low, even decreasing. And maybe it's going to be on a case-by-case basis when people work through this last time buy orders.

Our next question comes from the line of Mehdi Hosseini with SIG.

Yes. wanted to better understand the mix of the product revenue, especially given the increased contribution from the companionship. How should I think about the DDR5 RCD chip of -- RCD popular chip versus a companionship. How is that mix evolving? .

The way to look at it is we introduced a lot of products and there are different stages of introduction and qualification with our customers.

But in Q2, these new products represented low single-digit contribution in percentage terms of our product revenue. And when we look at Q3 that contribution in terms of percentage is probably going to be mid- to upper single-digit percentage of our product revenue. So as I said earlier, we planted the seeds. We see traction, and we're very happy with the traction with our customers. The contribution today is modest, but we do see very strong momentum in terms of adoption of these products.

Got you. Would that increase contribution continue into year-end?

Yes, it will continue in a year-end. We -- again, we're still in the phase of introduction preproduction of these products. So when we look at the view of our product revenue for Q4, we're comfortable with where the Street sees us, and we see a slightly higher contribution from our new products. But the real thing is going to be 2026 when the platforms are in full swing into the market.

Okay. All right. If I may squeeze my second question, I want to better understand the same kind of a diversification in your silicon IP. There was a significant improvement on a Q-over-Q basis of almost $6 million. Is that driven by HBM4? If not, what is driving that sequential increase in Silicon IP? And if HBM4 was not a factor when should we expect customers to come back and buy more IP for that specific application HBM4.

Mehdi, it's Des here. We are really pleased with the performance of our silicon IP business, which delivered strong results in the first half of the year. we're really on track to meet our sort of annual growth expectations for the full year from here. What I would say is, when you look at the different revenue categories of contract and Other in licensing billings, these can move around on each sort of quarter which is really dependent upon the IP that we are selling to customers. So what you did see in Q2 is an increase in our contract and other sort of line, which represents more customizable IP being sold.

And we saw the corresponding sort of decline on the licensing billings line, which is off the shelf IP. But what we really see here is a really strong momentum in the business, which has really been led by the memory controller solutions of HBM4, PCIe7 and also nice traction on the leading-edge security IP solutions. But overall, for the full year, we do expect the business to grow in line with our overall sort of expectations from here.

Our next question comes from the line of Natalia Winkler with Evercore.

my first one is about margin opportunity. Look, I was wondering if you can help us with an update on how you guys see that market and maybe sort of the ultimate proportion of the CPU market that might be using that technology.

Yes. So MRDIMM [indiscernible] enter the market towards the end of 2026. Depending on the availability of platforms, this is not an exoneration platform, but the one after. But it's important to engage with customers very early on. So at this point in time, we're very pleased with the progress we're making with our customers in terms of design, winning and engagements on the qualification side.

But that will contribute to the revenue towards the second half of 2026 and beyond. You remember the MRDIMM content is much larger than the content of the standard RDIMM for DDR5 because the RCD is more complex, the power management chip is more complex, but you also have 10 DB chips that were not present on the standard [ DIMM ]. So we're very excited with the progress, but that's going to have an impact in 2026, second half and beyond.

The market is difficult to assess at this point in time, but we expect in full swing, it could represent about $600 million market for MRDIMM that you can compare to the market for RDIMM today, which is about $800 million. So -- that's a significant growth potential in terms of SAM. But that's something that's going to happen in '26, the second half and beyond.

That's very helpful. And then my second question is around ARM CPUs. If you could help us understand if there's a little bit of a trade-off from a standpoint of units of the CPUs and the channel count, if you guys view the ARM CPU market different from x86?

We're kind of agnostic as to the CPU that is being used. Certainly, different, I would say, platform providers offer a different number of channels. we kind of take that into account when we estimate the market size. But for us, the very fact that people are developing chips based on ARM that are in competition with the x86 platforms is a good thing. It creates tension in the market. Competition in the market that is good for the rollout of higher speed RCDs and companion chip solutions. So we're kind of agnostic, but we see this in a positive way. .

Our next question comes from the line of Tristan Gerra with Baird.

Is fair to assume that the customized IP that you sold in the quarter that is more related to custom at. And also, when you talk about the contribution going from low single digit to mid to upper single digit this quarter from new products, I'm assuming companionship is really the vast majority of that increase in -- and is that more on the Granite Rapid platform?

Yes. To your second question, it's a combination. We introduced 8 new products last year, mostly companion chips. The chips that we introduced this year our companion chips for the client space, mostly in the power management area. And different customers at different stages. When we mentioned this low single-digit going to mid- to upper single digit, these are all these new chips that we introduced mostly companion chips. Your first question was, can you repeat your first question, please.

Customized IP. Yes, yes. It was regarding the customized IP and whether this was related to custom ASIC.

Yes, mostly it's custom ASICs. It's people developing their own chips to address the demands of the AI market. There's a lot of interest now for AI insurance, in particular, which drives the need for AI chips for high-speed interfaces. So yes, it's mostly for ASICs, ranging from start-up companies that want to enter that market all the way up to more established companies that already have a footprint into that market.

Okay. And then just as a quick follow-up, what is typically the type -- the time line between when you collect this customized IP versus the timing when the custom ASIC is ramping? And the reason I'm asking is because there is a number of hyperscalers that are at different stages of ramping customer ASICs over the next couple of years. And I think you've mentioned that, that increase in customized IP was happening in the quarter, but not necessarily sustainable or lumpy, but shouldn't we see an increase of medium term from customized IP revenue over the next -- in the medium term into next year?

Yes, that's a good question. Typically, our IP business is a licensing business. So our customers pay us when we deliver the IP for a license for one use or several use depending on the contracts. So we typically see the revenue. It depends 12 to 24 months before the products ramp into the market. So our current sales address chips that are going to be in the market in a couple of years from now.

And that's why we do see demand for these leading-edge technologies. People are using -- looking at HDM4, HDM4E, PCIE7 for the next generation of products, and we're going to be on that, I would say, a leading edge as we move forward. Then it depends on how successful these customers are.

They are customers that have been developing chips for many years, and we'll continue on that path with us. And startup companies, they are more and more start-up companies paying licenses to us as they move forward. Whether the chips are going to be successful or not, it's a different question. But again, it's important for us to have the revenue recognized at the time we sell the license when they actually decide to use these leading-edge technologies into their products.

We have a follow-up question from the line of Mehdi Hosseini with SIG.

Yes, Yes. One is looking to next year 2026 and 2027. And want to better understand how you're thinking about the opportunities associated with the client market, PC market versus CXL. It seems like CXL3.0 is more like a late '26 if it doesn't push out again. Would the incremental opportunity from PC market be enough, be large enough to offset if there is more push out in CXL adoption?

So thank you, maybe. The way we look at it is that you're correct. CXL may push out even further, but we do see MRDIMM really being the solution that is going to be adopted in -- for use case that has to do with memory expansion in particular. So on the data center side, we have high expectations for the deployment of MRDIMM. As I said, with revenue in the second half of '26 and '27, but I think that would address a lot of the use cases that CXL was supposed to address in terms of chip business.

Now clients is different. Clients, there's not really a CXL market for clients at this point in time or small -- for chips per se. But the client business for us, we really see this as an extension of our companion chip market for the data center. As we said earlier, the technical requirements that we're going to find in high-end client systems are very similar to the ones that we currently find in data centers.

So this is going to be a driver for SAM expansion for clock driver chips and power management ships into the client business. So that's a different area of growth for us, different than the MRDIMM in the data center space.

Thank you for your question. At this time, there are no further questions. This will conclude the question-and-answer session. I would now like to turn the conference back over to the company.

Thank you to everyone who has joined us today for your continued interest and time. We look forward to speaking with you again soon. Have a great day.

Thank you. This now concludes today's conference.

Thank you for joining us today. My name is Duksan Jang. I'm part of the U.S. semiconductors and semi-cap equipment team here at Bank of America. I'm very delighted to host the Rambus team today, Luc Seraphin, Chief Executive Officer; and Desmond Lynch, Chief Financial Officer. Thank you so much for coming.

Thank you, Duksan.

Did you have any disclosures that we might have to make? Or are we...

Yes. We just encourage everyone pleased to read our documents on file with the SEC. They've covered a lot more in the company than we will talk about today, Duksan.

Awesome. I think we can start high level. When we talk about the state of the union, what are you seeing in the demand environment today, especially perhaps versus the beginning of the year since we've had so much ups and downs this year?

Yes. So we're very pleased with how the year started for us. We do see some very nice tailwinds for the server market, both in the traditional server market and the AI servers that drives demand for more bandwidth and capacity in those servers, which, as a consequence, drive demand for our products. So we had a very nice first quarter on the product side with 52% growth compared to the same quarter last year. But we do also see demand for our silicon IP business. As people develop custom chips for AI, they need high-level security IP. They need high-speed interconnect controllers as well as high-speed memory controllers. So the overall AI environment and the additional traditional server environment has been quite good for us at this point in time.

Awesome. I'll get back to the Silicon IP business. But starting with the product side, as you mentioned, a very good quarter in the first quarter. How should we think about the overall market size, just stepping back and if you can talk about the competitive dynamic?

Sure. So the -- we traditionally started by building these what we call RCD chips or buffer chips that are little controller chips that sit on memory modules and work on the interface between the processors and the memories. The market for this chip, we estimate is about $750 million in size. But the nice thing with the DDR5 generation of products for modules is that in additional to the RCD chip on the module, DDR5 demands that we have, what we call, companion chips. Chips that did not exist on the module in the DDR4, the prior generation. So these companion chips add an additional $600 million SAM to the $750 million SAM.

And after that, what we see is some of the requirements that we see today in a server -- on the server environment, are going to be demanded as well in the -- on the client space. High-performance client systems are going to require chips that are similar to the RCD chip. And we believe that we'll add a couple of million dollars more of SAM to that. So we do see a SAM expansion that is coming from the fact that there's more content on the server memory modules. And there's also an adoption of similar technologies on the client side.

What would you say are the biggest drivers for this market? People talk about memory channel, the number of channels, the bandwidth, the capacity. What would be the biggest driver for Rambus?

So I think the first general comment we would make is that whether it's an AI server or a traditional server, there is a very high demand for more bandwidth and more capacity. And the reason is that server technology moves faster than memory technology. So there, you have more and more cores on every CPU, every core needs its dedicated memory. So that drives demand in general whether it's an AI server or a traditional server for more bandwidth and more memory. What it translates into for us is that in the DDR4 generation of product a few years ago, we had to develop an RCD chip every other year. Now today, in the DDR5 generation of products, we have to develop a new chip every year. So the cadence has been multiplied by 2. And as I said earlier, we also have to develop those companion chips.

The drivers are really the growth of AI servers, the growth of traditional servers, the number of channels per CPU. In the DDR4 generation, there were about 8 memory channels per CPU. In the DDR5 generation, it's a mix of 8 and 12, converging to 12. And we believe that at the end of the DDR5 generation of products, people will probably converge to 16 channels. It means that you have 16 memory channels on each CPU. And then the other driver is how many modules can you populate per channel. Some applications require 1 module per channel, other applications require 2 module per channel. So that's how the market grows. There is a growth of a number of channels and also the number of modules you can put on every channel. That's how the market is growing.

Just going back to my earlier question on competition, I know you guys are the leaders in the RCD chips. I think you said about 40% exiting last year. Your goal is 40% to 50%. What do you think needs to happen for you to reach the high end of the target or even push beyond that?

Yes. That's good question. In the DDR4 generation, we started with 0%, and we worked our way up to about 25% share. In the DDR5 generation, we enjoyed a little north of 40% last year because we invested very early in every sub-generation of product. And that's really, really important in that ecosystem because the qualification processes in that ecosystem are very complex and take a lot of time. So if you are the first one to introduce a new sub-generation of product into that ecosystem, you actually march out the resources of every ecosystem member and they work with you in getting that product out.

So we've been very good at investing very early in every sub generation of product. And that's what took us from the 25% share that we enjoyed in the DDR4 generation to the 40%-plus share in the DDR5 generation.

Now the ecosystem for reasons of security of supply, we'd always want to have multiple suppliers, typically 3 suppliers. So we have 2 competitors. One is Montage, a Chinese company, and one is Renesas, who bought that business from IDT. And I think the ecosystem will always require to have this type of arrangement because these little chips just sit between processors and memory. And if any one of these vendors fails for whatever reason, then you block the whole supply chain of that ecosystem. So I think we can grow -- continue to grow our share from 40%. Our goal is to get to about 50%. But then there's going to be some sort of saturation naturally in terms of share. So we have to count on the market growth, but more importantly, the content growth as we introduce all of these companion chips on the same module.

Understood. Talking about content and I know during the first quarter earnings call, you mentioned you're generally CPU agnostic, whether it's x86 or ARM. But how should we think about, just given ARM CPUs tend to be generally higher number of course, does that benefit you? Or if CPUs like NVIDIA Grace, they use LPDDR, how does that work into your content?

So there are 2 different questions. Whether it's an ARM core or an X86 core, we truly are agnostic. And if people -- what people are looking for is to add more and more cores for reasons that have to do with computational power. But the more core they add, the more memory they have to add. So all of this is good for us, whether it's ARM on x86, I think we actually welcome that competition. We welcome the competition between the ARM-based processors and the x86, and we welcome the competition within each one of these camps because they will drive demand for more buffer chips.

With respect to LPDDR, this is a niche market today. LPDDR is typically used in client applications. It brings some benefits in particular, in terms of power. That's why it's called low-power DDR, but it also comes with challenges that have to do with reliability, with the physical requirements that you have there. Our company, Rambus, has been in that business for 35 years. Every leg of our business has to do with memory technologies. So we do have a patent portfolio that covers LPDDR and DDR. We do have our silicon IP business that has cores in LPDDR and DDR. And when it comes to products, the vast majority of products today are DDR. If there was a compelling reason for growing in LPDDR solution on the product side, we would be ready to do that.

Understood. And then staying on top of this AI topic, we're obviously seeing a lot of demand moving away from training and more towards inference. Does that also have an impact on your product cadence or content?

It will be another tailwind for us. Typically, inference systems are simpler than training systems. A lot of things that are currently being used on GPUs and HBM can actually be run on more standard processors on the inference side. So that will drive demand for us. The nice thing about this market is that whatever processor you use because they have to use DRAM on the other side, those DRAM interfaces are standard interfaces. So whether that DRAM interface is on the standard processor ARM based or x86 based or whether it's on a custom chip that people develop for AI inference, for example, you will have the DDR interface. And on the other side of the DDR interface, you will have a module with that standard product. So all of these are good tailwinds for us, and we're looking forward to enjoying the rise of AI inference.

I do want to just go back to the earlier LPDDR question. Just because when we talk to ARM, when we talk to NVIDIA, they obviously have very aggressive outlooks for their great CPU. So if you were to develop a product on the LPDDR for the service side, how long would that generally take for you to obviously develop and then ramp?

The first thing I would say is that the current LPDDR solution has soldered solutions, they're not on modules. So you don't have the equivalent today of a buffer chip, right? So it's a bit like HBM. Today, HBM doesn't require a buffer chip. So we watch that, but to the extent that the market goes into solutions where LPDDR can be reliably integrated on a module as opposed to being soldered, then the development of the chip would be similar to the development of the buffer chip. So these developments last couple of years. Then the qualification in the market takes time as well. But that's -- for us, that's very similar technology, whether it's LPDDR or DDR. That's a very similar environment. These chips that we have to develop for modules.

And module environment is a very specific environment in terms of thermal requirements, noise requirements, so that's an environment we know well. And the ecosystem is an ecosystem we know well. The vendors of LPDDR memories are the same vendors, it is the DDR memory. The end users are going to be the same end users. So the whole ecosystem is very similar. As a consequence, it would take a similar time. But this push, as you say, it's a very interesting concept. But that's an ecosystem that we'll have to converge on a standard solution because every chip has to talk to every chip and every one of those chips has to talk to every memory module.

So the industry will have to converge onto a standard solution just as we do today with buffer chips, typically through JEDEC, and we are an active member in JEDEC and part of those discussions.

Got it. And then on to everyone's favorite topic tariffs. So you said patent licensing is not affected. But on the Silicon IP and product side, it's tougher to gauge the indirect impact. How should we think about the overall impact today, just given, obviously, every day, we're hearing so much more. But compared to, say, at the end of April when you reported, I think a lot of the nuances have more stabilized. So how should we think about it today?

So if you look at our business, our patent licensing business, as you correctly say, is completely immune. These are legal agreements that are long-term agreements with our customers, and there's no exchange of technology there. So that business is about $210 million, 100% margin. So that gives us a very solid base in terms of protection against tariffs. But the silicon IP business is also not affected by tariffs. We actually provide IP to our companies. Actually our exposure to China, even with our IP business is very small as a company. It's a low single-digit percentage of our business. So even if there were questions about tariffs with silicon IP, and they're not, then that would be having minimal impact on us. Then the question is about our product business.

Our product business last year was about $250 million. We review our situation with respect to tariff almost on a weekly basis. And at this point in time, we are not affected. One of the reasons is that our front-end supply chain is in Taiwan. Our back-end supply chain is in Taiwan and Korea and not in China. And we're selling our products to the memory vendors who typically buy them in Asia.

So at this point in time -- and these products are exempt at this point in time. Things can change, but we are under these exemptions. So at this point in time, there is no impact. There might be indirect impacts that we're watching. One is, if other companies shift their supply chains away from China to other areas in Asia, will this create a supply crunch that indirectly affect us with our suppliers? And the second thing is the overall uncertainty in the market that we -- are these tariffs going to destroy I would say, demand. But these are indirect effects for us that we're watching. In terms of direct effects, there's no direct effects at this point in time.

Understood. Just going back to the China exposure. Obviously, we're hearing EDA companies being left out of that market. Would you say that's also a similar risk for you on the IP business?

There's always this risk, but we -- that's not something that is new to us. As much as we review tariffs on a regular basis, we also review restrictions with respect to IP on a regular basis. We've been doing this for years, well before tariffs were in place. And at this point in time, we've had very, very little impact. And as I indicated earlier, our exposure to the China market is very small. It's low single digit. So even if we had a 100% impact, that would have a low single-digit impact on our business. But today, there is no impact.

Understood. Moving on to the companion chip opportunities. You launched 8 new chips last year. I believe you said you expect about low single-digit contribution in the first half. How should we think about it as we go into the second half? And obviously, next year, we should see some more of a ramp. If you can either quantify or either qualify or have some description for us.

Yes. The -- as we indicated earlier, when the market moved from the DDR4 generation of memory modules to the DDR5 generation of memory modules, the industry through JEDEC, by the way, everyone has to agree, the industry decided that some functions that were sitting on the motherboard in the DDR4 generation of products had to be implemented on the memory module instead in the DDR5 generation of products. So when you move from DDR4 to DDR5, on the module instead of having 1 RCD chip in the DDR4 generation, on the DDR5 generation, you have 1 RCD chip, 1 power management chip, 2 temperature sensors and 1 controller chip, which we call SPD Hub.

When that transition happened in the market, our strategy was to make sure that we secure the RCD chip market share first because that's the most complex chip to make. And that explains why we could move from 25% share on DDR4 to more than 40% in DDR5 because we wanted to focus on that. That transition was extremely strategically important for us because that's the most complex chip. And then we started to develop our companion chips.

The next most complex chip on that module is the power management chip. And in the first generation of DDR5, we were not playing. There were a lot of players. Actually, a few have survived, a lot of have not survived. And one of the reasons is that doing a power management chip is one thing, doing a power management chip in a module environment where it's very noisy, it's very tight in terms of real estate, it's thermally challenged is a different thing. So we invested in our power management chip team and in-house development about more than 2 years ago. We've introduced our power management chips last year in April. And we have also introduced the other companion chips.

Now like everything in that market, you have to intercept the platform from Intel and AMD. That's how the market works. So these platforms that use our generations of power management chip and companion chips are going to start ramping if they're not late, in the second half of this year. So the way to look at it is you were right today, it's a low single-digit portion of our revenue as we ship preproduction qualification quantities. When this platform ramps towards the second half of this year, we're going to see our share growing, and we're going to see the bulk of that growth in 2026.

We've been public about saying that for these companion chips, our objective is to reach about 20% share at this point in time because the competition landscape is a bit different. But obviously, you will try to do more than that.

On your MRDIMM chipset, Obviously, qualifications are ongoing. It probably depends a lot more on the customer side when they ramp their products. But what would you say is a realistic ramp timing for Rambus? When would this be more material for you?

Yes. So for people who don't know, the MRDIMM chipset, it's a very interesting concept. It's the idea that on a memory module you actually double the amount of memory and you multiplex the access of the memory onto the memory bus. So what it allows you to do or the industry to do is with exactly the same infrastructure, the same CPU architecture, you can picture the idea of removing a standard DDR5 module and plugging in an MRDIMM instead, and you, all of a sudden, double the capacity and double the bandwidth. So it has a lot of traction because as I said earlier, people are always looking for more bandwidth and more capacity. And it had to be -- again, the industry had to converge on the exact definition of this MRDIMM. That's why when we announced it, we say, is the first JEDEC compliant because that can give you security that the industry is going to use it.

So as we explained for the companion chip, this MRDIMM is linked to a platform launch. And this is a platform launch that will happen in 2026. So we have developed the products, we have sampled the products to our customers. They're going through all of their lengthy qualification process. But the product will ramp with the ramp of the follow-on generation of CPUs, which, at this point in time, is scheduled for the second half of 2026. So we're going to see the initial ramp of those products in the second half of 2026.

Got it. And then last one on products. If we talk about the client opportunity, and you've alluded to this earlier as well, but the clock drivers, how should we think about the opportunity there? And it's ramp timing?

Yes. So why do we go there first? Some of the challenges in the data center have to do with the environment. You have to transmit signals faster and faster between the processor and memory in a very noisy environment. And it's very tough to do, especially when you have to double the speed at such a fast pace. So that's why we developed RCD chips on the CPU side. And the RCD chip is all about what we call signal integrity. It's about transmitting very smooth signals in a very noisy environment without losing data.

Those requirements did not exist or don't exist today on the client space. But as client systems become more and more performant in terms of speed, what we see is that on the high end side of next-generation platforms on the client side, we're going to face similar challenges in terms of signal integrity. And we're going to have to have chips that actually reconstruct those signals as we do on the CPU side for data centers. And that's what the client -- clock driver is.

So it's going to address a very small portion of very high-end PCs, if you wish. So the market is going to be modest. We expect the market to be about $200 million for that. And the ramp is starting now and it's going to grow quarter-over-quarter through 2026. But strategically, what's going to happen is, as time passes, there are more and more client systems that are going to require that signal integrity function. And the client systems are also going to require some elaborate power management functions. And what we see is that we're going to see the technologies we developed for the data center waterfall into the high-end client systems. And with time, more and more of these client systems are going to use these technologies. So CKD is the first one of that building blocks that we are building for the future.

Got it. Moving on to silicon IP. Obviously, the HBM market is the one that's driving. How should we think about your content when the HBM3 stack moves from 8 high to 12 high? And then on to HBM4, is there an uplift there?

So our silicon IP business for people who are not too familiar with this, this is a very different business model. So we actually develop memory controller in the case of HBM, memory controllers, and we sell a license of these memory controllers to typically semiconductor companies and these semiconductor companies integrate this into their chips, whatever chips, these maybe an ASIC, it may be a CPU, DPU GPU, custom ASIC. So what this means is that we have to develop those controllers probably between 18 months and a couple of years, we have to engage with those customers 18 months to 2 years before those chips are actually in the market.

So in terms of HBM3 and HBM4, we were -- we've been engaged with customers for a couple of years now in HBM3. We announced HBM4 last year, and we were engaged with customers on HBM4 last year already. We actually, I think, indicated when we commented on our Q4 results that one of the reasons we had good silicon IP results in Q4 was actually driven by the demand for HBM4 controllers. And our strategy on HBM has always been on these controllers to be a little higher in speed and performance than what the market requires. So we have very early engagements into -- with our lead customers, ahead of what the market needs because we have to be like 2 years ahead.

The size of the stack does not really drive our development, but the speed really drives the development. We always have -- we always try to have -- to be at a slightly higher speed than what the market requires. But the demand for AI training, in particular, where you have GPUs using HBM memory, drives the demand for HBM silicon IP controllers. And what we have to understand is that in a GPU HBM environment, there's no equivalent of a buffer chip. There's not a chip that sits between the GPU and the HBM memory. As you said, there's a stack of memory, but inside the GPU, there's an HBM controller that we sell silicon IP that drives the connection to these HBM memory.

So it's been a good driver of our growth of the silicon IP business. As you know, our silicon IP business is about $120 million a year. We say it's growing 10% to 15% per year. Part of this 10% to 15% has been actually driven by the demand for HBM over the last couple of years.

Got it. I know we're running out of time, but an important question for Des, as we think about the margin trajectory, Q1 was a little bit weaker on the product side. You have a lot of different factors going on. I mean you have the price negotiations, the cost downs, price erosions. So how should we think about the second half outlook and into 2026, you also have the companion chips ramping?

Yes, it's a good question. I would say on the product gross margin side, we have a long-term target of 60% to 65%. If you look over the last sort of 3 years' annual performance, we've been operating at 61% to sort of 63% from there. So certainly within our sort of targeted range. So we're very pleased with how we've been able to operate and this is a healthy margin for the chip business. What we said is we've done a really nice job as a company. We've been disciplined on the price side as well as been able to continue to make manufacturing cost savings to maintain that sort of margin level.

As it relates to the new product contribution, that will be contained within the overall 60% to 65% sort of gross margin target. Obviously, any given quarter depending upon mix and where the products are within that cycle, the margin can move around a little bit. But I think in the long term, we have a good track record of delivering on the product gross margin side, and that's something we'll continue here sort of going forward from here.

Awesome. I think we run out of time. So thank you so much for coming. Thank you for the audience as well.

Thank you.

Thank you.