Veritone Aktie

Welcome to the Veritone Inc. First Quarter 2026 Financial Results Conference Call. [Operator Instructions] Please note this event is being recorded.

I would now like to turn the conference over to Cate Goldsmith, Investor Relations. Please go ahead.

Thank you, and good morning. Before the market opened today, Veritone issued a press release announcing results for the first quarter 2026 ended March 31, 2026. The press release and other supplemental information are available on the Investor Relations section of Veritone's website. Joining us for today's call are Veritone's President and Chief Executive Officer, Ryan Steelberg; and Chief Financial Officer, Mike Zemetra, who will provide prepared remarks and then open the call for a live question-and-answer session.

Please note that certain information discussed on the call today, including certain answers to your questions, will include forward-looking statements. This includes, without limitation, statements about our business strategy and future financial and operating performance. These forward-looking statements are subject to risks uncertainties and assumptions that may cause the actual results to differ materially from those stated. Certain of these risks and assumptions are discussed in Veritone's SEC filings, including its annual report on Form 10-K. These forward-looking statements are based on assumptions as of today, May 12, 2026, and Veritone undertakes no obligation to revise or update them.

During this call, the actual forecasted financial measures we will be discussing include non-GAAP measures. Reconciliations of these measures to the corresponding GAAP measures are included in the press release we issued today.

Finally, I would like to remind everyone that the call today is being recorded and will be made available for replay via a link on the Investor Relations section of Veritone's website at www.veritone.com.

Now I would like to turn the call over to our President and Chief Executive Officer, Ryan Steelberg.

Thank you, Cate, and thank you, everyone, for joining us today. As we look at the AI landscape today, one thing is becoming increasingly clear, the AI and data economies are now converging at scale. Over the last several weeks alone, the world's largest technology companies have continued to significantly increase their AI infrastructure investment plans, reinforcing the accelerating demand for AI compute, orchestration and high-quality training data. At the same time, enterprises, governments and content owners are increasingly recognizing the strategic value of the proprietary data assets, particularly unstructured data, including audio and video. This is exactly where Veritone is positioned.

Through aiWARE, Veritone Data Refinery and our market-leading applications, we sit at the intersection of both sides of the AI data economy. We help organizations transform unstructured data into AI-ready, semantic and monetizable assets while simultaneously supporting the growing demand from large hyperscalers and model developers for differentiated training data. Further validating these efforts and momentum, we are thrilled to announce that in Q1, both Google and NVIDIA have also now signed with Veritone for VDR data services. Our near-term pipeline for VDR has now expanded to nearly $70 million, helping to reinforce our full year 2026 guide of $130 million to $145 million.

In addition to VDR, we continue to strengthen our market position through strategic partnerships and customer expansion across our core growth areas. We announced a multiyear strategic agreement with Oracle to scale aiWARE, VDR and our award-winning applications on Oracle Cloud infrastructure, further strengthening our infrastructure scalability and enterprise AI capabilities across both commercial and public sector markets.

We also recently announced a new collaboration with the Washington Post to help unlock and monetize its news archive through our licensing and AI data solutions platform. In our Hire division, now rebranded as Broadbean by Veritone, we continue to expand our enterprise and government footprint, including the recently announced appointment with the U.K. Department for Work and Pensions to support workforce recruitment, modernization initiatives. And in the public sector, we continue to see growing demand for our AI-powered investigative and evidence management solutions across federal, state, local and international markets.

Before I discuss the business highlights in more detail, I want to provide an update on our focus and path to profitability. On our last call, we stated that we expected to achieve operating profitability as early as Q4 2026 driven primarily by the scaling of VDR and the continued growth in public sector. Today, we are taking another important step forward by proactively lowering our breakeven floor by approximately 30%.

We are simply not waiting for revenue growth to catch up to our cost structure, we are actively improving the operating efficiency of the business and capitalizing on the operating leverage uniquely enabled by the aiWARE platform that we have spent years building. What this demonstrates is a very clear bridge to profitability that is not dependent on aggressive growth assumptions, even under moderate revenue scenarios, Veritone is positioned to achieve operating profitability as early as Q4 2026. As VDR continues to scale and public sector momentum accelerates, the operating leverage and earnings power of the aiWARE platform is becoming increasingly evident.

With that broader backdrop in mind, let me now turn to the progress we are seeing across our core business segments, beginning with Commercial Enterprise where we continue to see growing demand for our AI software, data solutions and content monetization capabilities. Our Commercial Enterprise division maintained its robust 2025 momentum throughout Q1. During this period, we achieved a significant milestone by finalizing the landmark collaboration with the Washington Post to make their news archive universally accessible.

We further advanced the AI supply chain by introducing the Veritone Data Marketplace, providing scalable access to high-quality AI-ready training data sets. Additionally, the commercial team secured an extension with U.S. soccer utilizing our AI-driven products to enhance the monetization of both their archival and current footage.

Staying on the topic of sports. There is no better place for our product applications and agent workflows. Q1 leaning into early Q2 remains a landmark period for our sports vertical. We have successfully live ingested, tag and annotated thousands of hours of live sports data for many of the largest sports right holders in the world, including NCAA March Madness and the prestigious Masters Golf tournament.

By making the data readily available and accessible organizations are able to drive greater experience for their fans and sponsors alike, while maintaining control of their valuable IP, the content produced in these prestigious events. As the value of live sports continues to rise, the foundational understanding required to process content in near real time is a necessity. Our technology allows both rights holders like USTA, Big Ten and NCAA to not just deploy our applications and workflows across the organization but also instantly turn a live broadcast into a searchable monetizable and extensible library. Veritone is ensuring that live sports remains the most valuable inventory in the media ecosystem.

In the Commercial Enterprise, we closed 224 software and license agreements in Q1, including renewals and expansions across sports, media, entertainment and brand licensing. The strategic importance of these wins is not just a number of agreements, it is the expanding archive base, right cleared content relationships and monetization engine they create for VDR, VDM and aiWARE.

We are consistently expanding and refreshing our client portfolio, partnering with prestigious organizations such as CNN, the Smithsonian, Geico and the President Barack Obama Foundation. Our influence across the market continues to strengthen through collaborations with other leading brands, including Titleist, Tubi, Game Show Network and Bauer Media. As we look ahead, we plan to leverage our market-leading AI technologies, extensive expertise and blue-chip client roster to capture the next wave of potential customers to provide material growth for our commercial group.

We are undergoing a strategic reimagining of our vertical applications. Historically, our products and services were architected to address the complexity of the world's largest media entertainment, sports and news organizations. We are now leveraging the same tech stack to create optimized versions of our platform and applications that are accessible to organizations of all types and sizes. This is a significant democratization enabled by our technology.

Over the next few quarters, we are introducing solutions and pricing tiers designed specifically for these expanded verticals and segments. By offering the same best-in-class AI applications and agentic workflows in a more accessible package, we are greatly expanding our addressable market while maintaining our core high margins. A primary driver in this effort is Veritone's cloud-native digital asset management platform or digital media hub.

We are positioning Digital Media Hub as the central audio, video and image repository where mid-market, SMBs, city councils, marketing departments, schools and even individual creators can now leverage the same AI sophistication and dynamic workflows previously reserved only for the largest firms such as CBS News and the NCAA. We look forward to continuing to update you on our progress here in future calls.

Turning to our Hire division, Broadbean, delivered solid Q1 performance despite continued hiring market headwinds. The business met its Q1 revenue plan remained cash flow positive, added 42 SaaS clients globally and expanded in government and enterprise markets. The U.K. Department for Work and Pensions win is a strong proof point with Broadbean being supporting recruitment workflows across a large international public sector environment.

Our Global Media Services unit continued its double-digit year-over-year growth trajectory in both clients and revenue and expanded its business in North America. In Q1, we signed market key new clients, including the Department for Work and Pensions and 3 other government agencies in the U.K. Ministry of Justice, DEFRA and the Home Office. In addition, [indiscernible] in the U.S. and Arano in France. In total, we added 42 new SaaS clients globally.

On the Partnership's ATS front, we had significant milestones with all 3 global human capital platform leaders, Workday, Oracle and SAP. Our Workday partnership focus keeps delivering. We signed 24 new common clients in Q1, in line with our goal to sign 100-plus new clients this year and accelerate our new logo wins from this client ecosystem to over 50% year-over-year.

Our onboarding team is receiving Workday training and certifications in Q2 to improve our ability to onboard and activate new clients at scale. More importantly, we released the alpha version of our new Workday job management integration, which will enter beta in Q2 with an expected general release in Q3. We also completed the preparation of our SAP partnership agreement, which was signed in early Q2. At the same time, our Veritone group collaboration with Oracle is strengthening, and we released Oracle integration updates that improve clients, interface and campaign tracking.

Other highlights include the successful beta and expected Q2 release of our job acceleration feature on the programmatic advertising platform, which creates a unique self-service feature that allows clients to fund individual job campaigns and track candidates to specific high-priority roles. As we look ahead, we are unveiling our next-generation job management modules to more clients and continuing our groundbreaking agentic AI Broadbean framework with a Q4 alpha release target date. This isn't just an upgrade. It's a productivity revolution for our 30,000-plus monthly Broadbean users.

Moving on to Public Sector. I'm proud to report that we are off to a strong start in 2026 characterized by material growth and activation across the entire government landscape. We are seeing a powerful convergence of demand in both federal and [ SLED ] markets, culminating in a 69% year-over-year quarterly growth rate.

Our Veritone applications and iDEMS suite are doing more than just improving workflows, they are revolutionizing the very nature of productivity and efficiency for our public safety customers and end users. We are enabling mission-critical outcomes and materially driving up case closure rates that simply were not practical or even possible before Veritone.

Our integration with Thorn Detect is a strong example of how we continue to enhance iDEMS with trusted AI capabilities that support investigator safety and help agencies accelerate the identification and handling of harmful material. This ability to continuously evolve our applications as AI matures has dramatically expanded our total addressable market across SLED, higher education, Fed SIV and international agencies.

Security and data sovereignty remain critical focus areas for Veritone. aiWARE and iDEMS are engineered for the world's most sensitive environments, whether deployed in top-tier government cloud or entirely network isolated air gap environments, we meet the strictest sovereignty requirements. This is sure that any agency regards of its security posture can utilize our tools to support its most vital missions.

Furthermore, our recently announced partnership with Oracle allows us to scale this to an even higher level of performance in global security, providing us with a distinct competitive moat. Unlike the closed ecosystems of many of our competitors, aiWARE is built completely as an open platform. Our unique ability to ingest data at massive scale while integrating seamlessly with virtually any application or data set without vendor lock-in continues to differentiate Veritone in the market. This positions Veritone as a foundational AI infrastructure partner and the infrastructure of choice for federal AI modernization and the Department of War's AI-first strategy. This open architecture is also a key advantage for our SLED customers and strengthens our ability to collaborate with partners delivering complementary technologies and capabilities.

These foundational elements are translating into significant high-value wins. This quarter, we achieved deep integration within the Air Force OSI to support their [indiscernible], [indiscernible] and counter-espionage requirements and we expect the Air Force's use of aiWARE and our applications to greatly expand and accelerate in 2026 and beyond.

Additionally, another Department of War agency, the Defense Logistics Agency, went live with its own private instance of aiWARE and iDEMS, providing the basis for growth within the JPS Trust modernization program. Despite some operational government operational delays, I'm excited that we are back on track and moving aggressively forward with several of our marquee land-and-expand federal accounts. Our public sector pipeline currently sits at record levels.

We are seeing significant traction with the DOE's Project Genesis, the Department of Homeland Security and multiple form state and federal agencies. Our Q1 wins, which include a major U.S. university, top 5 shares department and several major U.S. city police agencies and state highway patrols, validate one simple truth: Veritone is a trusted AI partner for the public sector.

Finally, we are aggressively expanding our technical and partner reach. Our recently announced partnership with the Cold Case Foundation is a strategic force multiplier and demonstrates the same thing, using AI to unify decades of disparate investigative data and help agencies surface connections that would otherwise remain varied. This collaboration will not only accelerate our product capabilities, but will expose Veritone's technologies to entities across the U.S. and the globe as they partner with the foundation to solve their most difficult cases.

The conclusion is clear. We possess a unique AI native solution that solves the most critical challenges facing public sector organizations today. We are confident that the aiWARE technology stack is the essential foundation that will allow us to continue layering in agentic AI and automation capabilities well into the future. We are energized by this progress and the immense growth ahead.

Overall, we are very pleased with the progress we continue to make in our business. The momentum we are seeing across Commercial Enterprise, Public Sector and Broadbean continues to reinforce the strength and scalability of the Aware platform and our position at the center of the rapidly expanded AI and data economies. Importantly, the combination of an accelerating growth across VDR and Public Sector, together with the operating leverage initiatives we announced today, further strengthens our path to profitability and long-term value creation.

As organizations increasingly invest in AI infrastructure and seek to operationalize and monetize proprietary data, Veritone is uniquely positioned to capitalize on these long-term secular trends. We remain focused on disciplined execution, scaling our platform and data ecosystem, expanding strategic partnerships and converting our growing pipeline into durable revenue and profitability.

With that, I'll now turn the call over to Mike Zemetra to review our financial results and outlook in more detail. Mike?

Thank you, Ryan. As we previously discussed on our last call, Q1 '26 results are going to be somewhat in line with the prior year, in large part due to the timing of contractually onboarding several large hyperscalers in mid- to late Q1 '26 and of our Public Sector deals specifically to more pronounced expansions across the federal government and internationally. As I will explain later in my prepared remarks, we do expect VDR to generate its strongest quarter to date as early as Q2 '26 with several potential contract values individually ranging from several to tens of millions of dollars and as a result, we remain confident in our annual revenue guide of $130 million to $145 million.

Before I detail our Q1 performance, I would first like to discuss several important strategic initiatives. First is our recent strategic deal with Oracle announced in Q1 '26. We believe this partnership is a game changer and will initially provide over 20% savings on compute with nondilutive cash-based incentives from Oracle over time to facilitate the future scale and growth in our AI platform, including VDR, and expansion and acceleration of the Veritone Data Marketplace in our public sector.

In addition, we will be able to leverage Oracle's high-performance AI to power our aiWARE platform. This collaboration also allows Veritone's customer base to use AI with a superior price performance, security and data sovereign team provided by Oracle's distributed cloud. We plan to share more details on the progress on this initiative with Oracle as it progresses throughout fiscal 2026.

Second, we've made great strides in securing more digital data with the introduction of Veritone Marketplace, and further expanded the supply of digital content with adding many petabytes of readily accessible data from everything from cruise lines to fast food and major furniture outlets through key partnerships entered into at the close of fiscal 2025.

Why is this important? Every hyperscaler has specific requirements depending on what exactly they are trying to train their AI models on. For multi-camera angles of point-in-time situations to specific movements and actions in sports and real life to 4K nature videos, all with hundreds to many thousands of hours of bespoke, indexed and curated digital content for a single instance of AI training. Our VDR platform powered by aiWARE is uniquely positioned to solve this need at the scale and meet time lines these hyperscalers require.

In fiscal 2026, our goal is to be able to fulfill all of the hyperscalers needs in video and audio digital content and we believe that with partnerships we have forged through today and the build-out of the Veritone marketplace, our competitive moat just got larger in our ability to secure the necessary content and time frames of the hyperscalers improved substantially as compared to 2025.

Lastly, we are in the process of reevaluating our cost structure and believe we have the ability to unlock substantial savings of up to 30% in existing operating expense as early as the end of Q2 2026. In part to improve our operating margin but more importantly, to subsidize areas where we need to continue to invest for growth. This initiative reinforces our target of operating profitability as early as Q4 2026. We plan to share the details of our plan in the coming months.

During my prepared remarks, I will discuss our Q1 year-over-year performance in KPIs, balance sheet and liquidity position and provide updates on our financial progress in Q2 2026 and fiscal 2026 guidance.

Now I'd like to discuss our Q1 2026 performance in more detail. Q1 revenue was $20.3 million, down $2.2 million from Q1 2025, driven by Managed Services, which was down $1.5 million or 19.2% from prior year. Overall, our Software Products & Services was relatively flat year-over-year, including Broadbean by Veritone, which was down slightly year-over-year despite a very challenging macro environment across hiring in Q1 2026.

Note that Q1 results were tempered by the fact that deals with several large hyperscalers, including Google and NVIDIA, were delayed and not signed until mid Q1 2026. As a result, several larger VDR deals from newer hyperscalers were pushed into the remainder of 2026 as opposed to Q1 2026. I'm also happy to report that we currently have a near-term VDR sales pipeline and bookings of over $68 million, up over 150% from our guidance in mid-2025 and over 500% for our guidance a year ago. In addition, we have over $20 million of active sales pipeline, which could all close in Q2 2026 and includes several deals in the many to $10 million range. As I will discuss in more detail, Q2 2026 could be one of our best quarters on record. Assuming these larger deals close and at a minimum, 2026 revenue could be in the range of $25 million to over $30 million or in excess of 25% growth year-over-year at the high end.

Turning to the Public Sector, which grew 69% year-over-year, driven by the continuing rollout of larger deals executed in the first half of 2025, including the Department of Defense and certain larger public safety agencies. We did experience some delays in our larger federal deals, including the planned rollout of OSI but expect these deals to resume their planned rollouts in Q2 2026. Given these delays, which were partly driven by resource and prioritization across the DoD, we expect the public sector to continue to grow throughout fiscal 2026. However, this growth should be more pronounced beginning in fiscal 2027 with the expected rollout of iDEMS across the DoD, including OSI.

Turning to Q1 Managed Services, which decreased $1.5 million year-over-year, principally due to a decline in representation services, coupled with the year-over-year decline in content licensing due in large part to the timing of certain licensing revenue pushed to Q2 2026 and slight declines year-over-year from the NCAA March Madness. We expect this negative trend to reverse as early as Q2 2026 as we're already seeing improvements in our representation and licensing services in Q2.

Turning to key performance metrics across our software products and services in Q1 2026. ARR of $64.2 million, up 9% from Q1 2025 of $58.7 million. The improvement in the ARR was largely driven by increased consumption-based revenue from onetime software revenue or DDR and stable recurring SaaS-based revenue. Overall, ARR from consumption-based customers increased 50% year-over-year. Recurring subscription-based SaaS customers were flat year-over-year. As of Q1 2026, 73% of our ARR was from subscription versus consumption-based customers as compared to 81% in Q1 2025.

New bookings of $16 million, up slightly year-over-year. Gross revenue retention continued to be above the 90th percentile. In total, Software Products & Services customers of 2,897 was down 8% year-over-year, predominantly from our commercial enterprise sector, which includes lower consumption-based customers across broad beam by Veritone, principally due to a macro-driven churn from smaller customers as we focus on larger ARR opportunities. As the hiring market continues to struggle, we expect this trend of smaller ARR customers to continue throughout fiscal 2026.

Q1 GAAP gross profit was $12.7 million as compared to $13.7 million in Q1 2025. The decline was primarily driven by the decline in revenue, principally across our Managed Services. Q1 GAAP gross margin of 62.7% as compared to 61.1% in Q1 2025, an improvement of 166 basis points. Excluding noncash depreciation and amortization expense, Q1 2026 non-GAAP gross margin was 67.7% as compared to 65.1% in Q1 2025, an improvement of 260 basis points.

Note that we continue to forecast 2026 non-GAAP gross margins to be closer to 60% to 65% throughout the year, which will vary depending on the timing and mix of VDR revenue in a given period. Q1 operating loss of $19.4 million improved by $2.2 million or 10% year-over-year, primarily driven by lower operating expenses across G&A due in part to head count efficiencies coupled with lower professional and banking fees year-over-year.

Net loss was $19.5 million, a slight improvement from $19.9 million in Q1 2025. Driving this year-over-year improvement was the $2.2 million improvement in operating loss and a $2.4 million improvement in net interest expense year-over-year as a result of the paydown and retirement of 100% of the company's senior secured debt in November 2025. And offset by a onetime gain of $3.7 million in Q1 2025 for a change in the fair value of the company's estimated earn-out from the Veritone One sale in Q1 2024. Excluding this onetime gain, net loss would have improved to $4.1 million or 21% year-over-year.

Overall, non-GAAP net loss was relatively flat at $11.9 million as compared to $11.1 million in Q1 2025. The year-over-year variance was mostly driven by lower capitalized software in Q1 2026 as compared to Q1 2025.

Turning to our balance sheet. As of March 31, 2026, we held cash and restricted cash of $15.4 million as compared to $27.7 million at December 31, 2025. The $12.3 million net change in cash reflects: net cash outflows from operations of $11.5 million, principally driven by our non-GAAP net loss of $11.9 million; net cash outflows from investing and financing activities of $1 million, driven by net cash outflows of $0.5 million in capital expenditures and $0.7 million in net share settlement of equity awards. Excluding the $19.9 million capital raise, in Q1 2025, we improved our net cash outflows by over 40% or $8.5 million year-over-year.

Turning to liquidity today. As of March 31, 2026, we held $15.4 million of cash and restricted cash as compared to $16.4 million as of March 31, 2025. Moreover, all of today's cash is unencumbered and free of any restricted debt covenants. Unlike in the prior year, when we had a $15 million minimum cash requirement under our legacy senior secured debt. In addition, we have approximately $45 million of total debt outstanding at March 31, 2026, accruing interest at an annual rate of 1.75% as compared to over $130 million at March 31, 2025, a year-over-year improvement of more than $85 million in debt principal and more than $13 million reduction in annualized debt carry costs.

This improved balance sheet allows us to focus on reaching our growth potential to meet the market opportunities ahead of us. That said, we will continue to be opportunistic with continued focus to further improve our current liquidity position and balance sheet as well as the previously discussed plan to reduce our consolidated operating expenses up to 30% over the next several months. At March 31, 2026, we had 93 million shares issued and outstanding and 2.5 million warrants outstanding to certain legacy term debt holders.

Now turning to full year 2026 guidance. As a reminder, we will only be providing financial guidance for the full fiscal year 2026, given the complexity of forecasting the timing of VDR deals which tend to be larger in dollar value and entirely consumption-based, coupled with the complexity of government decision-making, especially during the war time. That said, and as I explained earlier, we are seeing a large backlog of more than 20 million in active VDR deals that could close in Q2 2026 and we have given a soft range in Q2 2026 revenue to be between $25 million to in excess of $30 million, which at the high point, would be a year-over-year improvement of over 25%.

As a backdrop to our annual guide, our Software Products & Services revenue pipeline and long-term outlook continue to be at all-time highs. More specifically, we continue to see strong demand across commercial VDR in the public sector. In 2026, hyperscalers, including Google, Amazon, Meta, NVIDIA, which are all current customers, have individually forecasted to spend hundreds of billions of dollars in fiscal 2026 to progress their AI initiatives, including further investment into their large language models.

According to Fortune Business Insight, the global AI training data set market size was valued around $3.6 billion in 2025 and is projected to grow from $4.4 billion in 2026 to $23.2 billion by 2034, a CAGR of 23%. From a model training perspective, we believe we are well positioned to exploit this potential revenue opportunity at the forefront of this future spending with our VDR solution as the more mature models are now investing heavily in rich video data where we believe Veritone has a clear competitive advantage. As of today, our near-term sales pipeline in VDR alone is over $68 million and continues to grow.

One of the largest learnings from fiscal 2025 was to improve upon the speed and expand the range of data set availability of content demand from our VDR customers and to improve our ability to deploy those data sets quickly. As previously discussed, we were unable to secure millions of dollars of potential VDR revenue in fiscal 2025, simply due to the fact that we cannot readily source the content requested from our VDR customers in a timely fashion.

To address this in 2026, we are focused on the most efficient and cost-effective ways to increase the supply of data, and we will also be investing in the engineering and product around VDR, including Veritone Marketplace where our aim is to deepen our competitive moat with exclusive access to thousands of more data providers. As previously discussed, we now have access to partners who control more than 50 million hours of valuable video data sets.

In Q1 2025, we entered into a highly strategic deal with a third party that locked down universal access to millions of hours of new video and audio data sources including with some of the largest retail, travel, entertainment and fast food providers in the world. We believe these near-term strategic decisions will enable us to continue to grow our VDR revenue in fiscal 2026 and beyond, at or above the current 23% projected CAGR for spending on large language models through fiscal 2034.

In the Public Sector, the market for digital evidence management solutions today exceeds north of $10 billion and is growing at double-digit rates. In fiscal 2026, we are targeting our large Public Sector growth to be between 60% to 70% year-over-year. This growth is forecasted to come from expanded offerings from existing federal contracts, including those with the DLA and OSI and from new international deals across Western Europe. That said, once we begin formally rolling out more instances of items across the broader DoD, including OSI and the DLA, we expect the growth rate to be much higher starting in the first half of 2027.

Collectively, our backlog and sales pipeline across our core AR platform remains in excess of $200 million today. And as Veritone remains uniquely positioned to capture even more opportunity in the data as a currency market, we expect that pipeline and our potential to monetize our trove of tokenized audio and video content to increase further.

On the OpEx side, we are forecasting relatively flat sales and marketing and G&A expenses year-over-year with forecasted spending across these areas as a percentage of revenue expected to show improvements year-over-year. However, they should be down year-over-year beginning in the second half of 2026, following our cost reductions initiative discussed earlier. We are also projecting research and development expenses to be slightly higher year-over-year throughout fiscal 2026 as we continue to invest in and build out our VDR and public sector initiatives, including the Veritone Marketplace and planned new software product features and enhancements in 2026 and beyond.

Including our revenue guide for 2026, we remain on track towards our projected operating profitability as early as Q4 2026. Note that consistent with 2025, we expect revenue to grow sequentially quarter-over-quarter in 2026, with Q1 to be lower in revenue from Q2 through Q4 2026 with progressive growth each quarter. This is partly driven by the public sector, where we see a higher revenue ramp starting late in the first half of 2026 from our existing larger federal deals, coupled with the timing of certain international contracts we expect to announce in the coming year. In addition, and based upon discussion and timing of certain VDR deals and the delayed signing of several large hyperscalers to late Q1 2026, we expect to start seeing a more pronounced revenue ramp in VDR starting in Q2 and throughout the second half of 2026.

The key risk to our revenue projections are the consumption-based nature of VDR, coupled with the timing of government-based contracts and decision-making. In addition, the visibility into our VDR pipeline is typically 2 to 3 months in advance of delivery and decision-making on the nature and volume of content they change depending on the customers' need and anticipated impact on the training models.

More specifically, in fiscal 2026, we are maintaining our previous guidance of revenue to be at $130 million to $145 million, which at the midpoint represents a 49% increase year-over-year. We are expecting the public sector revenue to continue to grow over 60% year-over-year and the remaining growth to come from our Commercial Enterprise sector, predominantly from VDR. Our Broadbean by Veritone products and services are included in this growth, and we expect Broadbean by Veritone to be slightly down year-over-year given the current macroeconomic hiring environment. Our Managed Services is expected to be up year-over-year by 10% to 15%, principally due to the recent improvements we are seeing on the representation side of our business.

We expect gross margins to fluctuate between 60% to 65% and driven by the forecasted mix of revenue in the period and non-GAAP net loss to be between $13.5 million to $22.5 million, which at the midpoint represents a 56% improvement year-over-year as compared to fiscal 2025. The change is reflective of the timing shifts in revenue, the previously discussed planned increase in research and development, coupled with the compressions in gross margins due to the mix of VDR in 2026. As previously discussed, we believe we are still on track towards operating profitability, which at the earliest would be Q4 2026.

Before closing the call, I'd like to remind everyone listening that Veritone will be in New York City this week attending Needham's 21st annual technology, media and consumer conference. That concludes my prepared remarks.

Operator, we would like to now open up the call for questions.

[Operator Instructions] The first question comes from Joshua Reilly with Needham.

All right. Great. I have a few. Maybe just starting off on the expanded digital media hub opportunity. Can you give us some more color around what verticals you'll be targeting there and the timing of the launch for that offering?

Josh, yes, we're excited about this one. Right now, as we kind of detailed on the call, the digital media hub was primarily and originally built and designed for, I'll say, more complex enterprise-level integrations and onboarding of data for some of the largest media companies. And although, obviously, we're very sticky with those customers. And again, we're in no way going to be compromising our service attention to them. But what we're going to be introducing here just over the next couple of months, and again, to be very clear, over the next couple of months, not a few quarters, is an ability to preserve that same level of market-leading DMH capabilities but to allow us to almost provide a near self-service onboarding solution for these entities. And so that is going to allow us to more seamlessly onboard individual teams, even individual creators, but also corporate enterprises.

What we are all seeing is almost every company is a media company now, meaning they are creating unstructured audio video they need a more effective way of storing those files above and beyond just, I'll call it, more traditional like say, generic storage facilities, frankly, like a Google Cloud or, say, like Google Drive or a Box or something like that, they need the proficiency and expertise that Veritone has been providing our media and entertainment customers for years. And so again, we believe that this is going to greatly expand the vertical focus that's going to span between smaller businesses but also corporate enterprises who are sitting on a tremendous amount of audio, video and other structured data, and we do expect to launch this renewed solution just in here in the next couple of months.

3

Got it. All right. On the Oracle agreement, can you just remind us the timeline for integrating the OCI, how you'll be marketing to customers the use of that and then maybe how are you going to leverage that as a sales tool and a little bit of a competitive advantage going forward?

So first is we are very near starting to do some porting of -- I think we're well along the path of integration. It's been a great collaborative working environment, working with their team. To be clear, what's making all this possible is our ability and our success of transforming aiWARE to a complete platform-agnostic solution. So our ability to make this transition finally and start working with Oracle really starts with us, right, and our -- and what we've done in terms of transition in Kubernetes and creating a more containerized top-to-bottom offering of aiWARE and applications.

That being said, the Oracle team has been fantastic. And we are looking and we are doing integrations, and we're looking to actually start moving some major payloads over to them as early as early August. That being said, we have already started the co-selling opportunity with them. I've been invited and I'll be speaking for example, the keynote in June with their national sales kickoff and we expect to really be working in a very collaborative operating model with them as they're going very aggressive across several verticals. Obviously, media entertainment, sports and news is a big one where obviously, Oracle is very motivated to continue to drive, but also across the public sector as well.

So again, I think we're on track or even ahead of schedule in terms of technical integration and planning to start moving initial, I'll say, storage payloads over first and then compute payloads. Number 2 is we're already working and working together on co-selling executions. And then third ultimately is -- and once we are up and running, we do believe that these incremental cost savings based upon due to the efficiencies of how we're going to be running at a lower cost structure with Oracle Cloud will be passed on to our customers, which again, I think is going to be resulting in a more competitive offering than our competition.

Got it. That's helpful. And then as we think about the guidance of $130 million to $145 million in revenue for the year, if you could just lay out some of the key variables plan to hit that would lead you to hit the high end of the guidance or exceed the number? And what -- I'm assuming it would primarily be driven by VDR and Commercial Enterprise deals, but any additional details on how you're thinking about the setup of the guidance?

I think for sure, it's going to be dominated by VDR. As Mike kind of articulated in more detail, not just I'll say the overreaching pipeline, but kind of the short-term visibility we have on multiple different deals. Obviously, we were hopeful, as we've communicated a few times to bring some of the major -- the additional hyperscalers, Google and NVIDIA on board in Q4. Those were delayed until February of this year. But the bottom line is we finally got those done and we're servicing orders now. So we're really excited about that opportunity. So again, VDR will lead and be the bellwether to achieve and hopefully surpass even that the high end is an opportunity.

Public sector, as we mentioned on the call, we are completely unlocked now with a couple of our bellwether accounts that we landed last year, the DLA and Air Force OSI, and we expect those to considerate. We're on track again, they actually had -- there was actually some personnel changes over there. We're past those blockers. And so as we have articulated, we're up and running and live with DLA. Again, DLA, the logistics agency manages and staff all of our bases around the world, and we're looking to greatly expand our rollout through the balance of '26 and beyond for DLA and also OSI. So I think those will be 2 bellwether leads that are going to help substantiate and hopefully get us to the high end of our guide by the end of the year.

Got it. One last question for me is on the Public Sector business, curious, when you're selling to these customers, how much is it a budget of -- getting budget allocated a factor relative to just understanding the capabilities of what your platform can do for them? And how does that impact the timeline of closing deals with both U.S. federal customers and state and local in terms of awareness relative to budgets?

Yes. Great question. So let's start with the -- let's break them apart. So for state and local enforcement, I think the key is being ready. So like everybody, they'll have different cycles where certain opportunities for, okay, I'd say, larger reviews of platforms and systems come up for renewal. So the key is to make sure that you have those relationships, you are communicating with the procurement officers at times even the chiefs or the captains at the respective areas. So it's important for Veritone to continue to build up our brand relationships with these groups, which I think we're doing a great job at.

Second is that when these opportunities come arise is how can you enter or land an additional contract with large agencies such as, again, which we've kind of teed on larger entities in California or New York whether they're sheriffs or police agencies. The key is having and what makes kind of our offering so unique is we can land with one application. We don't need to come in and have them buy the entire stack of all the offerings of iDEMS in every single application. For example, they may have budget already available out -- potentially out of cycle for just programmatic redaction, right, document and audio video redaction or another agency in their homicide division may have immediate budget for investigate, right, trying to accelerate and speed up case closures.

So again, the key is being ready when they do have a budget -- when it comes up for, I'll say, scheduled cycles, but also being opportunistic that if they do have -- and then they vary, but let's just say, hypothetically, a budget threshold without having to go to City Counsel is $30,000 or less per year. I'm just giving the hypothetical. We do have offerings. We have offerings because, again, everything is kind of built on aiWARE, we can land right at low entry cost points and then scale up with those entities and we can actually land in different departments, not just from landing an entire police agency, but again, if they have budgets in their records department for redaction or in their homicide division for an investigate, that's another way, makes us very unique in that capacity as well.

So again, on the federal side, again, it's all across the map. Again, I think what we've shown here is making sure that we are continuing to be disciplined and bidding on every single RFI and RFP comes out that we think is relevant to aiWARE applications, which we are doing. But it's also aligning yourselves directly with the mission leaders. As we mentioned on the call, when we're working with the Department of Homeland Security and Project Genesis, those are us working with individuals directly. It's important that you start to build those personal relationships in addition to going through the [indiscernible] and others for, I'll say, the more traditional procurement bidding an RFI and proposal process. So I would say it's probably even more involved on the federal front. But obviously, you're starting to see the fruits in which we discussed a little bit of finally landing some of the bellwether accounts with the Air Force and DLA and continue to build those and sort of make sure we're continuing to have the right products and services and service layer to scale those when they're ready to scale up.

[Operator Instructions] this concludes our question-and-answer session. I would like to turn the conference back over to Ryan Steelberg for any closing remarks.

Thank you for joining today. Veritone is poised for strong disciplined growth through the balance of 2026 and beyond by securing additional leading hyperscalers under contract and fueling a robust VDR pipeline, bolstered by our strategic Oracle partnership, we have solidified our role as the essential infrastructure for next-generation AI training data and enterprise scale deployment. Our momentum and pipeline within the Public Sector, coupled with the resilient performance of our Broadbean hiring division, reinforced our high level of conviction.

We are not just expanding our growth in reach. We are resharpening and we are sharpening our execution. The proactive measures we are taking to streamline operations and optimize our cost structure through internal reorganization, automation and AI initiatives are already yielding results. We expect to see these efficiencies and productivity gains accelerate starting in this current second quarter, firmly underpinning our path to operating profitability as early Q4 2026.

Veritone is leaner, faster and uniquely positioned to capture the massive AI opportunity ahead. Thank you for your continued support as we deliver on this mission. Have a good day.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.

Welcome to the Veritone Incorporated Preliminary Unaudited Fourth Quarter 2025 Financial Results Conference Call. [Operator Instructions] Please note, this event is being recorded.

I would now like to turn the conference over to Cate Goldsmith, Investor Relations. Please go ahead.

Thank you, and good afternoon. After the market closed today, Veritone issued a press release announcing its preliminary unaudited financial results for the fourth quarter and full year ended December 31, 2025. The press release is available on the Investor Relations section of Veritone's website.

Joining us for today's call are Veritone's President and Chief Executive Officer, Ryan Steelberg; and Chief Financial Officer, Mike Zemetra, who will provide prepared remarks.

Please note that certain information discussed on the call today will include forward-looking statements. This includes, without limitation, statements about our business strategy and future financial and operating performance. These forward-looking statements are subject to risks, uncertainties and assumptions that may cause the actual results to differ materially from those stated. Certain of these risks and assumptions are discussed in Veritone's SEC filings, including its annual report on Form 10-K. These forward-looking statements are based on assumptions as of today, March 26, 2026, and Veritone undertakes no obligation to revise or update them, except as may be set forth in a subsequent press release related to the company's audited Q4 and full year 2025 financial results.

During this call, the actual and forecasted financial measures we will be discussing include non-GAAP measures. Reconciliations of these measures to the corresponding GAAP measures are included in the press release we've issued today.

Finally, I would like to remind everyone that the call today is being recorded and will be made available for replay via a link on the Investor Relations section of Veritone's website at www.veritone.com.

Now I would like to turn the call over to our President and Chief Executive Officer, Ryan Steelberg.

Good afternoon. I'm excited to provide an update on our preliminary Q4 financial results, discuss the strategic deal announced today with Oracle and provide more strategic insight into some of our newer and exciting business opportunities, including our material progress with certain hyperscalers. Because we are discussing preliminary results, we will not hold a question-and-answer session following our prepared remarks today.

Our preliminary Q4 results, which we furnished today have a fairly wide revenue range of between $18.1 million to $30 million. The range is almost entirely driven by a single transaction, which we completed and signed in Q4 2025. The transaction was a complex multiparty nonmonetary transaction, which included an on-prem sale of our aiWARE stack and application software at a price of $12.9 million. In exchange, we received certain intangible rights and direct and preferred access to a significant number of customers, who control a variety of digital data sets for future use in VDR and AI model training at a fixed revenue share of 50%, which is a significant improvement over our current margin on VDR today.

While the contracted price of the software was $12.9 million, from a pure GAAP accounting perspective, it is very challenging to arrive at the appropriate fair value of the sale as the realization of the data rights is prospective, considering the relatively new, albeit fast-growing market of data sales and VDR. As a result, the standalone selling price of the software could be discounted substantially from the revenue perspective. It is important to note that the $12.9 million price was included in our previous range of Q4 revenue guide, but it could end up in the lower end of this range, it is very binary. That said, we are actively working with internal and external resources to ensure the value is fairly reflected in our Q4 2025 results, which we expect will be completed by the time we file our 10-K.

Irrespective of the final GAAP accounting conclusion, I would like to explain why this deal is so strategic for Veritone, and why we are confident that we will ultimately turn this $12.9 million face value active deal into something far greater. Veritone currently holds exclusive and nonexclusive rights and relationships to monetize some of the most valuable and iconic sports, entertainment, public safety and news data sets from content and IP owners, such as the NCAA, CBS News and other franchise customers. However, the demand for VDR-specific content has increased substantially, and the content demands have varied widely. We do not have universal access to unlimited amount of data suppliers under contract with Veritone today.

Case in point, in fiscal 2025, we were forced to turn down more than $10 million worth of bona fide data orders because we could not source the volume or specific type of content or data in the time frames our hyperscaler and model developer partners required. Hence, why this strategic deal is part of the solution going forward in addition to our normal business development efforts. As a result of this deal, we now have preferred access to potential VDR customers, who control more than 50 million hours of monetizable data sets today. To put this into perspective, the entire marquee catalog of the NCAA video library is less than 1% of this size. And just to cite one example of these new customers from this transaction, which we recently signed an agreement with, is a major fast food franchisee to provide Veritone with access to their catalog of multi-camera video surveillance footage, data that is in high demand by hyperscalers for frontier and world model development. Our forecasts have conservatively estimated the overall library of data sets from this transaction for potential VDR customers could generate over $100 million in VDR revenue over the next 3 years.

In addition, we have a fixed margin on this data that is significantly better than our historical VDR margin. While the corresponding revenue may ultimately recognize over time from an accounting perspective, we now have access to a significant amount of valuable data sets that can be monetized through our VDR platform with no upfront cash expenditure to obtain them. We expect to begin monetizing this data set as early as Q2 2026. I know I have spent a lot of time discussing this transaction, but we want our investors to have a detailed understanding of the situation as well as the significant opportunity.

Next, and I am so excited about this, I would like to introduce and discuss the groundbreaking agreement we just announced with Oracle, a deal that has been in the works for almost a year. The Oracle deal is a multiyear strategic partnership to accelerate the deployment of Veritone's aiWARE platform, application and data services via the Oracle Cloud Infrastructure, or OCI, with built-in and substantial financial cash incentives that allow Veritone to scale its cloud infrastructure at a more efficient cost for compute and storage. Under the terms of the agreement, Oracle Cloud Infrastructure will become the cornerstone of Veritone's next generation of AI solutions for commercial, public sector and high-growth Veritone Data Refinery markets. As the demand for scalable and secure AI infrastructure reaches unprecedented levels, Veritone will leverage OCI's high-performance AI super clusters to power its aiWARE platform and data solutions. This partnership ensures that Veritone's customer base can harness the transformative potential of AI with superior price performance, security and data sovereignty provided by Oracle's distributed cloud.

Signing and collaborating with Oracle represents a pivotal milestone in Veritone's mission to help enterprises find the truth in their data, and Oracle's financial commitments in this partnership validate our leadership in managing unstructured data and our commitment to providing some of the most robust AI solutions in the market. By migrating critical workloads to OCI, we are unlocking new levels of scalability for Veritone Data Refinery and providing our public sector and commercial customers with the performance and security they require to stay ahead in an AI-driven world. Please check out the joint press release on our investor site to learn more.

On to our core operating business. Over the past two years, we have been on a disciplined journey to realign Veritone around a singular clear vision, reestablishing aiWARE as the essential orchestration and intelligence layer for enterprise and public sector AI and unstructured data. Today, I am proud to announce that, that transformation of Veritone is complete. We have reshaped this organization into a focused platform-driven company, and we are now executing from a position of operational, financial and technical strength. In 2024, we started to reposition the company. In 2025, we validated that model, and today, we are no longer managing through a transition. We have exited non-core assets, we've simplified our operating structure and aligned our resources behind scalable, platform-driven revenue streams. We are now deploying capital, expanding our platform and driving the kind of operating leverage that defines the market leader.

The most significant evidence of our progress is the Veritone Data Refinery or VDR and more importantly, those hyperscale leading customers that we are now serving. What began as a unique platform capability organically built, has evolved into a scaled token production engine that is now powering the world's major hyperscalers and foundational model developers. Our strategy was to convert proprietary unstructured data sets into reoccurring revenue, and that strategy has been validated at the highest levels. We are now actively engaged with under contract too and generating revenue with all major players, including NVIDIA, Amazon, Google and Meta, among others. Now that we have increased our access and supply of rich data and have established contracts with all major hyperscalers, we see a clear opportunity to deepen and expand our engagement across each and every one of these partners.

With hyperscalers expected to spend approximately $700 billion in combined CapEx in 2026 according to S&P, Veritone stands to capture an increasing share of that investment as a leading provider of clean, model-ready training data. On the supply side, we are effectively enabling enterprises to treat their unstructured audio video images and text as a renewable, monetizable asset class to meet this AI demand. As previously discussed, it is imperative that we remain very focused and cost-efficient on data acquisition.

On the VDR supply side, in addition to data sets now available to us as a result of the strategic transaction I detailed above, we continue to sign and expand our portfolio of available data sets through other channels as well as to improve our capabilities to refine, process and prepare those data sets for licensing, distribution and sales. Historically, the sales and representation process has been heavily dependent on business development and manual efforts, which, despite being effective, has limited our ability to significantly accelerate the acquisition and distribution of our data set offerings and portfolio. That changes I'm excited to follow on the announcement we did a couple of weeks ago about the launch of the Veritone Data Marketplace built on aiWARE.

Veritone has officially taken the next step in evolution of the AI supply chain. Leaning into the data economy, we recently announced the launch of the Veritone Data Marketplace, a platform designed to improve and streamline the data ecosystem without compromising on control, transparency or quality. This marketplace is the essential partner to our Veritone Data Refinery. While VDR acts as the tokenization engine that makes unstructured data AI-ready, VDM, the Veritone Data Marketplace serves as the trusted and accredited distribution hub, ensuring those assets are monetizable, transactional and governed. We are particularly excited about the value this brings to both sides of the data licensing market. Rights holders with valuable archives and data sets now has a clear path to monetization with complete asset control. Simultaneously, AI developers gain access to a deep catalog of proprietary data spanning industries from entertainment to human behavior to robotics. These data sets can even be evaluated for metadata density and model fit before acquisition, ensuring they are optimized for foundational architectures.

Previously available only to a limited group of customers, the Veritone Data Marketplace is now available for content owners and AI developers to accelerate responsible AI development. We are proud to lead the charge in sourcing high-quality data ethically and delivering it through a governed, transparent marketplace that meets the needs of the hyperscaler community.

Our public sector division is starting to execute on all cylinders delivering a strong 68% year-over-year growth. This success is built on three clear strategic pillars: First, high-impact applications. Our Veritone iDERMS suite is revolutionizing productivity, enabling mission-critical workflows that simply aren't possible without Veritone. This has dramatically expanded our addressable market across SLED, higher education, Fed SIV and international agencies. Second, unmatched deployment flexibility. aiWARE and iDERMS are designed for the most sensitive environments. We meet the strictest security and sovereignty requirements, whether deployed in government clouds or completely network isolated air-gapped environments. And now with Oracle, we will take that to an even higher level of performance and security, both domestically and globally. And third, a true open architecture. Unlike many of our competitors, aiWARE is a completely open platform, our ability to ingest data at scale and connect with any application or data set without vendor lock-in makes us the infrastructure of choice for federal AI monetization and the Department of War's AI First strategy.

This foundation has led to deep integration within the Air Force OSI and the JPS Trust modernization program. With our pipeline at record levels, our Q4 wins, including a major U.S. University, a top 5 sheriff department and another major state highway patrol truly validates that we are a trusted AI partner for the public sector.

In commercial enterprise, we have successfully operationalized our data to AI flywheel. By connecting proprietary supply or surging AI demands, we built a scalable architecture where volumes drive value. More data attracts more buyers, which directly fuels our margin expansion and relicensing efficiency. The results speak for themselves. In Q4, our licensing team executed 224 orders, growing nearly 10% over the prior year. Our reach is truly global, providing critical media assets to top-tier firms like Google and Goldman Sachs, major studios like NBCUniversal and premier sport entities, including ESPN, the NFL and Tom Brady's Religion of Sports.

This momentum directly feeds our high-margin software business. Software deal volume grew 14% year-over-year to 33 deals in Q4. We are seeing a powerful combination of retention and expansion, renewing core partners like Sony Pictures and SummitMedia, while simultaneously landing new accounts like Snap and deepening our relationships with prestigious events like the London Marathon and Augusta National Golf Club. I'm incredibly proud of our Hire division, now rebranded as Broadbean by Veritone, which delivered yet another strong year. Despite macro hiring headwinds, Broadbean maintained and contributed robust profit margins critical to Veritone's overall growth and success. To appreciate the scale that our Broadbean division manages, in 2025, Broadbean distributed over 7.6 million unique jobs, powering more than 40 million unique job ads. The result, we drove an impressive 132 million engagements in application and clicks directly into our customers' ATS and recruitment systems. Every month, an average of 30,000 unique HR professionals rely on our software to manage their talent acquisition.

Other highlights include our Global Media Services unit, achieving a record-breaking year with double-digit year-over-year net revenue growth. In fact, Q4 was our strongest on record, making the first time we crossed the $10 million threshold for media under management. We've carried that energy directly into 2026. I'm thrilled to announce a major SaaS win with the U.K.'s Department of Work and Pensions. This partnership establishes Veritone as a cornerstone of U.K. Government recruitment, as our software will now power job advertising for the Ministry of Justice, DEFRA and the Home Office.

Furthermore, our first year in the Workday Platinum Partner Program was a triumph, securing 59 new deals, a 30% increase over the previous year. With an expanding pipeline of Fortune and Forbes Global 500 brands, we are just getting started. As we look ahead, we are preparing to unveil our next-generation job management modules and our groundbreaking Agentic AI Broadbean Framework. This isn't just an upgrade, we believe it's a productivity revolution, again, for our tens of thousands of monthly Broadbean users.

Looking forward to 2026. As we enter the year, our focus is simple, disciplined scale. We are focused on converting our $50-plus million VDR pipeline into recognized reoccurring revenue and expanding our public sector deployments. We will continue to allocate capital towards platform expansion, and we'll continue to evaluate selective strategic opportunities and partnerships that strengthen our data and orchestration moat. Veritone has moved past the transition. We have the platform, the partners and the financial foundation to lead the AI-driven data economy.

Thank you. Now I'd like to turn it over to our Chief Financial Officer, Mike Zemetra. Mike?

Thank you, Ryan. Given the preliminary nature of our financial results, I will only be discussing our guidance for fiscal 2026 today as well as a few balance sheet updates. Our software products and services revenue pipeline and long-term outlook continue to be at all-time highs. Specifically, we continue to see strong demand across commercial VDR and the public sector. In 2026, hyperscalers, including Google, Amazon, Meta, NVIDIA and others have individually forecast to spend hundreds of billions of dollars in fiscal 2026 to power their AI initiatives, including further investments into their large language models. According to Fortune Business Insight, the global AI training data set market size was valued around $3.6 billion in 2025 and is projected to grow from $4.4 billion in 2026 to $23.2 billion by 2034, a CAGR of 23%.

From a model training perspective, we believe we are well positioned to exploit this potential revenue opportunity as more mature models are now investing heavily in rich video data, where we believe Veritone has a clear competitive advantage. As of today, our near-term sales pipeline in VDR alone is over $50 million and continues to grow. One of our most important learnings in 2025 was to expand the range of data set availability for our VDR customers and to improve the velocity to deploy these data sets. As Ryan mentioned, we were unable to secure millions of dollars of potential VDR revenue in fiscal 2025 simply due to the fact that we could not readily source the content requested by our VDR customers in a timely fashion. To address this in 2026, we will be focused on the most efficient and cost-effective ways to increase the supply of data, and we will also be investing in the engineering and product around VDR, including Veritone marketplace, where our aim is to deepen our competitive moat with exclusive access to thousands of more data providers.

As Ryan discussed, as a result of our Q4 strategic data set transaction, we now have access to customers who control more than 50 million hours of valuable video data sets, including with some of the largest retail, travel, entertainment and fast food providers in the world. We believe these near-term strategic decisions will enable us to continue to grow our VDR revenue in fiscal 2026 and beyond at or above the current 23% projected CAGR for spending on large language models through fiscal 2034. In the public sector, the TAM for digital evidence management solutions today exceeds $10 billion and is growing at double-digit rates. In fiscal 2026, we are targeting our public sector to grow between 60% to 70% year-over-year. This growth is forecasted to come from expanded offerings under existing federal contracts including those with the DLA and OSI and from new international deals across Western Europe.

Collectively, our backlog and sales pipeline across our core AI platform is in excess of $200 million today. And as Veritone remains uniquely positioned to capture even more opportunity in the data as a currency market, we expect that pipeline and our potential to monetize our trove of tokenized audio and video content to increase further.

On the OpEx side, we are forecasting relatively flat sales, marketing and G&A expenses in fiscal 2026 as compared to 2025 with forecasted spending across these areas as a percentage of total revenue expected to show improvements year-over-year. We are projecting research and development expenses to be slightly higher year-over-year on an absolute dollar base, starting in Q1 and throughout fiscal 2026 as we continue to invest in our VDR and public sector revenue initiatives, including the Veritone marketplace and planned new software product features and enhancements in 2026.

Note that consistent with 2025, we expect revenue to grow sequentially quarter-over-quarter in 2026 with Q1 2026 revenue approximating Q1 2025. This is partly driven by the public sector, where we see a higher revenue ramp starting late in the first half of 2026 from our existing larger federal deals, coupled with the timing of certain international contracts we expect to announce in the coming year. In addition and based upon the discussion and timing of certain VDR deals, including the signing of several large hyperscalers in late Q1 2026, we expect to see a more pronounced revenue ramp in VDR starting in Q2 and throughout the second half of 2026.

The key risk to our revenue projections are the consumption-based nature of VDR, coupled with the timing of government-based contracts and decision-making. In addition, the visibility into our VDR pipeline is typically 2 to 3 months in advance of delivery and decision-making on the nature and volume of content may change depending on the customers' need anticipated impact on its trading models. Given these factors, coupled with the complexity of government decision-making, especially during wartime, we will only be providing financial guidance for the fiscal year 2026, which we plan to update on our next earnings call.

More specifically, in fiscal 2026, we are expecting revenue to be at $130 million to $145 million, which at the midpoint represents a 47% increase year-over-year from the low end of our 2025 preliminary financial range. We are expecting the public sector revenue to grow between 60% to 70% year-over-year and the remaining growth that comes from our commercial enterprise sector, predominantly from VDR. Our Veritone Hire products and services are included in this growth, and we expect Veritone Hire to be flat to slightly down year-over-year given the current macroeconomic hire environment. Our managed services is expected to be up year-over-year by 10% to 15%, principally due to the recent improvements we are seeing on the representation side of our business. We expect gross margins to fluctuate between 60% to 65%, driven by the forecasted mix of revenue in the period and non-GAAP net loss to be between $13.5 million to $22.5 million, which at the midpoint represents a 54% improvement year-over-year at the low end of our preliminary 2025 financial range. The change is reflective of the timing shifts in revenue, the previously discussed planned increase in research and development coupled with the compression in gross margins due to the mix of VDR in 2026. We believe we are still on the path to profitability, which at the earliest would be in Q4 2026 and is highly dependent on the planned compound growth of VDR in the public sector throughout fiscal 2026.

Finally, I want to highlight several material improvements we have made to our balance sheet. In Q4, we retired 100% of our senior secured term debt and repurchased approximately 50% of our then outstanding convertible notes. This has resulted in a 90% reduction in our annual debt carrying costs from roughly $14 million to just $800,000. We closed fiscal year 2025 with unencumbered cash and cash equivalents of $27.7 million, free of any restricted covenants, $45 million and 1.75% convertible debt and 92.6 million shares outstanding. We expect to provide a full financial release of Q4 and the full year 2025 once we finalize our fiscal 2025 results, which we expect to furnish when we file our 2025 annual report on Form 10-K.

That concludes my prepared remarks. Ryan?

Thank you, everyone, for your time today. Veritone has come a long way in just a very few short quarters. Over the past two years, we have focused the business around aiWARE, our core platform, which powers really our entire corporate and product offering strategy. We are very excited about the scale that we are beginning to experience. For example, VDR has evolved in a true production engine, and we are engaged with all of the major hyperscalers, which is a tremendous accomplishment. The strategic data set transaction and our partnership with Oracle are additional proof points of our success, and they're expanding our access to high-value data and giving us the infrastructure to scale more efficiently going forward. As we move into 2026, we are focused on disciplined scale by converting our pipeline into revenue, expanding our data supply and continuing to build on the operating leverage we have created. We have the platform, the partners and the foundation in place. Now it's about execution and scaling into what we believe is a very large opportunity. Demand remains strong. Our pipeline continues to grow, and our engagement with hyperscalers is as strong as it has ever been.

I appreciate everyone's time today, and we look forward to speaking with you next quarter. Thank you.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.

Before we begin, I'd like to note that certain statements we may make today, December 1, 2025, could be considered forward-looking. These statements are based on our current expectations and assumptions, and actual results may differ materially due to risks and uncertainties. We do not undertake any obligation to update these statements after today's discussion.

For a full list of risk factors, please review the disclaimer on this slide and our most recent filings available at sec.gov. A replay of this webcast will be available later this week via a link on the Investor Relations section of Veritone's website at www.veritone.com.

[Operator Instructions] We will review the submissions and address the selection of questions towards the end of this session.

Welcome, everyone, and thank you for joining Veritone's AI and Data Economy Forum. We have a full agenda for you today where you'll learn more about Veritone's vision and strategic framing of this exciting market opportunity. We're also going to go through in detail our technologies, products and solutions, including demos. And then finally, we're going to go through our go-to-market and business strategy to continue to drive high growth.

In addition to myself, you will also hear from other Veritone leaders, including Sean King and Chris Doe as well as a few comments from a few of our esteemed customers.

First, let's talk about the title of this forum today: The AI and Data Economy. I specifically did not state just AI, but data as well and Veritone is very focused on both.

Succinctly, data is the lifeblood of AI. And in today's world and economy, data is both a new asset class and is becoming a new form of currency. Veritone is a native AI company from our inception in 2014, but we are also a data company defining the future of data tokenization, utilization and ultimately, monetization.

Going back to our founding. Our company name, Veritone, means truth in the signal. It embodies our foundational belief that the world's unstructured data contains objective truths that would properly refined fuels intelligence, insights and value. We build the systems and applications that process and activate this data to uncover and operationalize this truth and also its value.

This data processing at scale has established Veritone as a key enterprise leader in both the AI and data economies. Enabled by our unmatched expertise in unstructured data, and our proprietary AI operating system, aiWARE, Veritone delivers market-leading applications and workflows at scale focused on transforming unstructured data into actual intelligence and tangible value.

For more than a decade, Veritone has been cognitively processing or tokenizing video and audio, which is the fastest-growing segment of unstructured data. We've been converting this unstructured data into metadata rich tokens, aligned with faces, objects, speech, sentiment, events and many more contextual embeddings. Our aiWARE operating system provides a scalable, secure and AI model agnostic foundation for the ingestion and operationalizing of both structured and unstructured data across disparate enterprise systems securely, reliably and at scale.

More importantly, we are processing and tokenizing this data in a transactional utility-driven format that delivers immediate value and measurable ROI for our customers, which includes many of the market leaders for media, entertainment and sports, including CBS, who you're going to hear from later; to public sector and government, including the United States Air Force and hundreds of state and local law enforcement agencies and sheriff departments.

To help educate the audience on what exactly is this unstructured data opportunity and its journey, let me walk you through an example customer.

In this case, CBS News. CBS News has been producing millions of hours of unstructured audio and video going back decades. What Veritone does is we work with them as a customer. We ingest all of that raw audio and video into an instance of aiWARE. And then immediately, we start applying cognitive AI processing against those audio and video files. This turns it into indexed and organized tokens. Those tokens, think of them as understanding of what's inside every single frame of video and every second of audio, is actually the input that drives these powerful applications. It creates the utility value.

And sometimes, it's as simple as search and discovery, helping archivists or researchers find specific points in time or specific elements of content. It helps their sponsorship and advertising groups optimize product placement and also helps drive research decisions, what programming is working and what's not working. But the point here is we are the full end-to-end stack from ingestion to AI cognitive processing and tokenization to actually providing the utility value of this indexing through our applications.

Today, Veritone stands on a trajectory of strong and strategic growth, powered by, again, our AI operating system, aiWARE, which fuels the global data economy generating trillions of tokens per quarter across all of our customers. As the AI economy accelerates, the data economy is expanding right along with it, and Veritone's strategic positioning and market timing could not be more perfect.

Building on our scaled data processing and tokenization of unstructured data, with the example that I just gave about CBS News, we are also emerging into an exciting new field, working with hyperscalers and AI model developers to actually help train and tune their next-generation models. In fact, we're using the same audio and video files that we've prepared and clean and indexed, and we are now working and licensing and selling that data directly to the largest hyperscalers and AI model developing companies. We call this new solution, the Veritone Data Refinery, or VDR.

Through VDR, we have established ourselves as a premier data and model training partner for many of the largest hyperscalers and foundational model developers. We transform, again, raw, unstructured audio and video into high-quality tokenized data sets that can be used to train and fine-tune the world's most advanced AI models.

VDR is a new product line that is being sold into both public sector and commercial enterprise customers. With a qualified near-term pipeline and bookings already surpassing $40 million, VDR is emerging as a critical partner to many global hyperscalers and the next generation of AI model developers.

As we've stated on previous calls, we are confident by the end of the year, we will hold active contracts or projects with nearly every major hyperscaler in the market, further solidifying VDR's position as a critical enabler in the unstructured AI training data ecosystem.

Our tokenization platform not only powers our AI workflows and customer applications today, but now serves as the foundation for a powerful new monetization framework. The data as a currency era has arrived and Veritone is uniquely positioned to capitalize. Once data is tokenized through aiWARE, the opportunities are endless for our customers. We're able to help them increase profitability and work more efficiently with data that is instantly searchable, analyzable and actionable.

As content and data libraries expand and data volume surge, Veritone is uniquely positioned to unlock value from commercial enterprises and IP owners. Our technology enables enterprises, many of whom who have been loyal customers of ours for over 10 years to now fully monetize their content and data archives, transforming dormant assets into active revenue-generating resources while simultaneously producing high-quality model-ready training data that fuels the next generation of AI innovation.

Now enough of me talking. Let's pass it over to Chris Doe and Sean King, who are going to actually walk you through our products and services and our go-to-market strategy.

Hello, everyone. I am Chris Doe, Head of Product at Veritone. And today, I'm very excited to walk through our technology platform, our product portfolio and a preview of what is coming next year.

Before we begin, a quick introduction about myself. I've spent the last 20 years in product management across data, media and online advertising. I've always gravitated towards the bleeding edge of technology, and that is ultimately what brought me to Veritone 5 years ago.

Today, I oversee our commercial and public sector applications as well as aiWARE, the platform that underpins all of our AI software and data services. These products are business critical for some of the largest media companies, sports leagues and law enforcement agencies in the world.

First, I want to ground us all in the technology platform that makes everything you'll hear about today possible. aiWARE is the core end-to-end platform that powers all of our commercial and public sector applications. It brings in large volumes of audio, video, images and documents, applies AI processing at scale and converts that raw data into data tokens that can be searched, analyzed, transformed and monetized.

The platform excels at 4 key areas: ingestion, AI cognition and orchestration, tokenization and activation. First is ingestion. aiWARE is built to handle data at scale, whether that is a movie studio archive, a live broadcast feed, a series of CCTV networks or drone footage, all of it can be brought into a single secure environment with full data provenance. So customers always know exactly where every asset originated from and how it has been processed. The platform also features advanced metadata support and synthetic metadata enrichment options.

Second is AI cognition and orchestration. The platform manages a large and growing ecosystem of models, including our own proprietary models and those from the best-in-class providers. All processing runs in a secure, controlled and explainable manner to support the requirements of both the commercial and government customers. As part of this, we developed our own token standard called AION, which provides a consistent way to represent token outputs across all these models. This ensures that every piece of intelligence generated by aiWARE is structured, traceable and interoperable.

Third is semantic tokenization. aiWARE normalizes the intelligence it produces, creating transcripts, labels, timestamps, entities, embeddings and other forms of structured metadata. This is what enables clean and consistent search across all of our products including natural language search, which you'll see throughout the demos today.

Finally, activation. The intelligence generated by aiWARE can be activated in multiple ways. In most cases, the activation is within our purpose-built applications. But in other cases, the value is achieved in dashboards, alerts, data exports and third-party integrations.

Activation is where business outcomes are achieved. As we move throughout the demos, remember that all the applications you see today are not point solutions. They're built on 1 platform and 1 single architecture, aiWARE.

Having introduced aiWARE and how our underlying platform ingests and cognitively processes the unstructured data, we are now going to dive into our application layer, which ultimately turns this value into our customers. Starting in commercial.

The initial application I want to discuss is Digital Media Hub, or DMH. DMH is our intelligent media asset management, archive and monetization solution for all of your creative content. DMH is used by many leading sports, news and entertainment clients worldwide, including CBS, NBCUniversal, Augusta National and ESPN.

DMH got a major upgrade this year and is faster, smarter and more integrated than ever. We've enhanced our AI-powered tagging, streamlined the user interface and introduced powerful new collaboration tools.

The real magic is the AI, automated tagging and processing allows the user to instantly begin searching across the indexed data. Users can simply enter a term, phrase, keyword and automatically be taken into the exact moment they are looking for, no more manual tagging.

Users can also leverage our advanced moment search built on top vector database that allows for a more semantic-type search experience. DMH also now includes upgraded commerce tools that simplify rights management, pricing and transaction processing, making it easier to monetize your archive and with automated workflows and cloud transcoding, delivering assets to the right destinations is faster and more seamless than ever.

We have also introduced new customer-driven features, including bulk tools for metadata updates, ordering and organization as well as a redesigned collections experience that makes sharing and collaboration easier than ever. As you can see, DMH delivers a deep set of capabilities that support an impressive range of customer needs.

Next, let us look at our long-standing Discovery product, which is used to ingest and analyze thousands of television, OTT, radio and audio streams 24/7. Here, you can see how users can search radio and broadcast television streams using natural language. This enables media broadcasters to derive programming insights in a matter of seconds. Discovery also includes one-click tools that generate concise data summaries and actionable campaign recommendations.

Most recently, Discovery has been fully integrated with our chatbot, Veri, giving users interactive search and deeper content intelligent insights. Veri is the result of more than a year of innovation from our labs team. It's an amazing tool to summarize all the AI processing results that are attached to the ingested files.

Users can ask Veri to summarize large watch list, break down key moments and answer follow-up questions in real time. They can refine their results by adjusting confidence levels, request deeper context or even ask for the sentiment of an entire video. Veri delivers these insights in seconds, turning complex analysis into a simple conversational experience.

Now let us shift into our public sector applications. This is one of the fastest-growing areas of our business and for good reason. Digital media analysis has become essential for modern policing and investigative work. In many ways, it is becoming the new DNA analysis. Agencies are dealing with enormous volumes of video, audio, documents and digital evidence. And there's just simply no way they can process all that information manually at the scale required.

With that said, let's talk about Veritone's Intelligent Digital Evidence Management Systems, also known as Veritone iDEMS, our solution that helps public safety and justice agencies accelerate their investigations. Veritone iDEMS is a suite of AI-powered software products, which I'm excited to show you now.

The first application in iDEMS is a Investigate. This product helps investigators quickly find key moments across large sets of evidence, connect related items and collaborate more effectively across cases. It brings every file into one place, along with all the associated metadata and AI-generated insights so that everything is searchable and easy to navigate. Investigate also connects seamlessly with all the rest of the iDEMS products, ensuring that all the evidence and workflows stay linked across the entire investigative process.

Next is Veritone Track. Track uses advanced computer vision to help investigators locate a person of interest across large collections of video. This goes far beyond simple face recognition. Track evaluates appearance, clothing, movement patterns and other visual attributes to find matches across different cameras, angles and environments. As you watch the demonstration, imagine an investigator needing to search across dozens of CCTV feeds to identify where a person may have been.

Traditionally, this could have taken days or weeks. With Track, that work can happen in minutes. The system presents potential matches, timelines and movement paths. So investigators can quickly understand where a person may have traveled and which footage is most relevant. It accelerates investigations while reducing manual review time in a significant way. The end result of this workflow is a shareable person of interest timeline to help solve active cases faster.

Now back in the Investigate application, I can search for a transcription keyword and show you how Redact works. Redact is one of the most widely used public sector applications and solves a very real-world challenge. Agencies are required by law to protect PII, personally identifiable information, when releasing body-worn camera footage, dash camera footage or other evidence. Historically, this meant someone had to go frame by frame and manually blur sensitive information. Redact removes that burden. As the video plays, you can see how Redact automatically detects sensitive keywords and applies accurate, frame-consistent masking. The same workflow is possible for faces, screens, license plate and coming soon, documents.

In the next example, after running another simple keyword search and navigating to our data detail page, you will see the automatic video summarization applied to the body camera footage. This is a straightforward but powerful demonstration of how we save investigators significant time while showcasing the strength of our video understanding capabilities. We will be adding additional police report generation capabilities next year and also adding Veri, so detectives can talk to their evidence if they so choose. Together, these capabilities highlight how iDEMS delivers the workflows and intelligence required to drive better investigations. In the end, it's all about catching the bad guys as fast as possible.

One of our most exciting advancements in our product portfolio over the last year is the Veritone Data Refinery, or VDR. Introduced late last year, VDR represents the next evolution of our aiWARE platform. Taking everything we've built in multi-mobile ingestion, cognitive AI processing and semantic tokenization and elevating it to a new strategic tier.

For years, aiWARE has powered our applications by transforming raw video and audio into structured machine-readable intelligence for our customers. With VDR, we've extended that same architecture to a new frontier, preparing and refining tokenized video and audio as high-quality training data for modern AI models.

This offering is in theory, is a direct application of our core expertise. aiWARE's proven ability to ingest, structure and cognitively analyze unstructured media now allows us to produce training-ready data assets at scale with the consistency, transparency and operational rigor that model developers require.

VDR is both a product and a service engine, a full stack pipeline that transforms enterprise archives into AI-ready intelligence, unlocking new revenue streams for rights holders and accelerating model development for AI companies. It is one of the clearest expressions yet of how Veritone's technical foundation, proprietary workflows and domain experience position us at the center of the data economy emerging around next-generation AI.

So what are we working on for VDR for next year? Well, we're going to be introducing a lot more advanced token generation services via advanced clipping and annotation tooling. Our goal is simply to provide the best quality training data to our partners and over-service them throughout all engagements.

Looking ahead, we are entering one of the most transformative periods at Veritone. Several years of deep platform and application investment are coming together, and we're about to enter a new era of innovation at Veritone. A major theme will continue to be natural language. Users across all of our products will be able to leverage our chatbot Veri to talk to their data for search, analysis, workflow execution like we showed earlier in Discovery.

To be clear, we will always support the structured user experience our products offer today, but we will complement those experiences with a natural language alternative if clients so choose. You will also see more advancements in video understanding. As more models are introduced into the market, you can ensure Veritone will be testing, onboarding and benchmarking all of those for our clients.

Next, real-time video analytics is something I'm very excited about. Veritone has a long track record of delivering real-time AI insights that help our customers make business-critical decisions. And doing that reliably at scale is incredibly difficult. But this is what we excel at. And we have both the experience and the vision to continue leading in this area.

Finally, agentic workflow. Today, the super majority of our utility value realized by our end customers is through the application layer. Yes, these AI-enabled applications, including DMH, Redact, Investigate and others have greatly improved the speed and efficiency. But the ultimate yield potential of all this centralized tokenized data is still constrained. This is where Veritone agentic AI will exponentially increase performance and yield.

Imagine the same homicide detectives using Investigate can invoke a swarm of Veritone AI agents to not only enable them to analyze multiple crime scenes at the same time, but these agents can immediately start taking action on potential evidence, hits and elements of interest to advance or follow up on.

Now imagine a police department using the same technology to monitor dozens of city cameras receive real-time alerts or person of interest and immediately pull relevant case evidence through natural language search. Investigators collaborate on the same data in minutes instead of days.

On the commercial side, imagine a major television network running hundreds of live feeds and simply asking Veri to track brand exposure in real time or identify story moments, generate highlight clips or flag compliance issues, all automated, all instant. These examples are why we are hard at work, adding Veri to all of our applications, many of which are scheduled to launch in this upcoming quarter, Q1 of 2026.

In sum, next year's road map is the most exciting we've had in my 5-year tenure. I'm incredibly optimistic about 2026 and the evolution of both our product portfolio and our service offerings, all powered by our comprehensive aiWARE platform.

You may wonder why I'm so optimistic. Well, look at 2025. Veritone created a rocket ship business line called VDR pretty much out of thin air. And this was because they used the aiWARE platform, a platform that allows builders to launch new products and solutions quickly and efficiently in a secure and compliant fashion. I can't predict the future, but I do know we have a lot of entrepreneurs on staff, and I guarantee there will be more lines of business sprouting up in years to come. That's why I'm so optimistic and so I'm so bullish about aiWARE.

Thank you for spending the time with me today. I want to thank all of our customers, partners and employees that made this possible today.

Hi. I'm Sean King, Chief Revenue Officer here at Veritone. Our overarching message today is clear. Veritone has built the market-leading tokenization platform, and our decade of revenue growth has resulted in one of the largest and most comprehensive libraries of tokenized audio and video data. These tokenized data assets are what power our applications, as Chris demonstrated, providing operational efficiencies, production gains and net new revenue opportunities for our customers across both commercial and public industries.

This foundation allows us to serve an industry-agnostic customer base around the world. Our aiWARE platform powers organizations across media, government, sports and enterprise sectors from global broadcasters and hyperscalers to law enforcement and Fortune 500 companies. Because aiWARE is adaptable to any data type and workflow, we deliver value across diverse industries and use cases whether data needs to be analyzed, managed or monetized.

Today, you'll hear from a few of our customers, demonstrating how we help leading organizations unlock the full value of their data. As our advanced AI and data capabilities have strengthened, we are seeing not only new business wins and renewals but contract expansions, allowing our customers to fully harness their data assets.

As Ryan mentioned at the start, the data as a currency era has arrived, and Veritone is helping our customers lead it. Every organization sits on vast underutilized data, and we help them turn it into living monetizable assets. Through Veritone Data Refinery, alongside our licensing services, our customers transform video, audio and text into tokenized model-ready data sets as well as commercially ready media assets that unlock new revenue streams, help enrich and improve the creative process and accelerate AI innovation.

Our customers, whether in media, government or enterprise gain the ability to control, enrich and transact on their data safely and at scale. As data becomes the world's newest currency, Veritone enables our customers to fully participate in the emerging data economy and create measurable value from every token.

Now let's hear from one of our long-standing customers CBS News.

I'm Maggie Dakin, Director of Archive Sales for CBS News Archives. For over 90 years, CBS News has documented the most pivotal moments both in our nation's history and across the globe. Over time, it became clear that these stories would hold immense value for generations to come. With that, the CBS News Archives is born.

We have been partnering with Veritone licensing for the last decade to make our content available for storytellers, creating moving and informative films, documentaries, episodic series and podcasts. As we digitize our archives, Veritone's aiWARE provides a unique opportunity to solve one of the biggest hurdles in our business, search and discovery. As we look toward digitizing our entire archive, Veritone's technology will play a pivotal role in improving the licensing experience for creatives.

Now let's shift to one of the fastest-growing, most critical areas of our business: public sector. Today, digital media analysis has become essential to modern policing and investigative work. It is, in many ways, the new DNA analysis. Law enforcement and public safety agencies are facing a data challenge of immense scale, dealing with enormous volumes of video, audio, documents and digital evidence from body cameras, surveillance videos and other sources that are simply impossible to process manually.

Veritone's Intelligent Digital Evidence Management Systems, or iDEMS, is the direct solution. It's a suite of AI-powered tools designed to not just manage this data but to accelerate investigations and deliver better outcomes. Our iDEMS suite showcases compelling use cases that directly solve the most painful points for our customers. Veritone Investigate helps teams find the crucial signal and the noise connecting evidence and enabling collaboration.

Veritone Redact addresses the mandatory compliance challenge with releasing public records, replacing weeks of manual frame-by-frame blurring with automated, accurate masking of faces and personal identifiable information, or PII. This is an unmatched win for efficiency, privacy protection and transparency.

Lastly, Veritone Track leverages advanced computer vision, turning days or weeks of manual video review into mere minutes, locating persons of interest across vast video collections. Collectively, iDEMS delivers faster investigations, more successful case outcomes and strengthens the vital relationships between agencies and the communities they serve.

Now let's take a moment to hear from our customer, Eden Prairie Police Department.

My name is Alyssa Benkowski, I am the records and evidence supervisor with Eden Prairie Police Department. The Eden Prairie Police Department maintains a large amount of digital data captured by over 90 body-worn cameras, 30 squad cameras and several interview room cameras. Each year, our agency responds to over 40,000 calls for service. With only 3 record specialists, in addition to their day-to-day responsibilities, they respond to approximately 2,500 public data requests each year, so it is imperative that our tools are accurate, efficient and dependable.

My first body worn camera project was for a 15-minute video. And with our old system, it took me over 8 hours to complete that project. When we tried Redact, I ran that same video as a comparison, and I finished that in about an hour.

The transcript feature and head detection immediately stood out to me, especially since facial recognition alone doesn't tend to keep up with people when they're turning their heads or when officers are moving within the frame.

Redact's upload speed, transcript processing, object detection and integration with our evidence management system have made a big difference for us. We no longer waste time downloading and reuploading files and multiple staff members can work on those projects at any time, which is essential for the volume that we are dealing with.

In addition, we recently learned that muting and beeping audio does not meet our state requirements, which are 2 of the only options that most systems offer. When I reached out to Veritone after learning this, I received a Veri prompt response that audio masking had recently been implemented, and they worked quickly to turn that feature on for our department.

This leads me to one final thought. One of the main things that stands out about Veritone is their customer service. Their team is quick to respond, always helpful and makes it known that our needs truly matter. Even if I contact the wrong person, I'm always routed to the correct one right away, and I never feel like I'm inconveniencing anyone over there. They really take a lot of pride in their products.

Redaction will always be a big and sometimes daunting task, but with Veritone Redact, it has made it much more manageable. It helps us stay compliant, maintain transparency and keep our workflow moving. We are truly grateful for the difference that this has made in our processes.

It's absolutely incredible to be able to service such customers. It takes a robust go-to-market strategy to drive the kind of revenue results we saw in third quarter. Our customer contracts are generally subscription-based, consumption-based or a combination of both, enabling flexibility to meet a wide range of customer needs. We serve thousands of existing customers through these models, while leveraging both direct sales efforts and a strong partner ecosystem to attract new prospects and build a healthy, sustainable pipeline.

Our go-to-market strategy centers around direct sales motion supported by a strong diversified partner ecosystem. By prioritizing direct opportunities, we can ensure tight alignment with our customer needs, shorter feedback loops and deepen engagements across the industries we serve. This motion is amplified by our robust partner network that extends our reach and accelerates adoption.

Today, we leverage hyperscalers, system integrators and channel partners like AWS, MissionRT, Technology North, Key Code Media, Carahsoft, OPEXUS and Nuix, just to name a few. We also align with necessary technology alliances like the AWS Generative AI Center of Excellence to meet customer needs where they are. Together, these complementary motions enable Veritone to scale efficiently, understand new and developing needs and deliver AI and data solutions that are operationally transformative.

Ultimately, our message is clear. Veritone is the company that transforms vast underutilized data into living usable, monetizable tokens. These tokens power our customer applications and all monetization opportunities.

Through aiWARE, a market-leading tokenization platform and its intelligent applications, we are empowering organizations across every sector for media to law enforcement, accelerating innovation, achieving measurable value from every token and fully leading the data as a currency era.

Thank you all for your time today. Over to you, Ryan.

All right. We are going to have a little live Q&A here. Let us kind of assemble our team, make sure I see everybody. And then I'll turn it over to Cate, who is helping organize the questions. But first, I just want to let thank you, everybody, for joining today. Obviously, we have a pretty exciting and dynamic team over here. Thank you, Chris. Thank you, Sean, for being representatives of a very large -- much larger organization. There's a lot of people who contributed here. But about -- but before we kick it off, I want to just to impress upon people really the power of our core technology platform that makes all this not just possible but so cost effective for us to continue to iterate and build these effective workflows any solutions for our customers.

All right. With that, Cate, can you assemble the questions, and let's get through these.

Sure. Thanks, everyone, for submitting questions throughout the session, and we also have a few analyst questions that came in before the session today.

So Ryan, just to get started, this came in from the chat. The White House officially launched the Genesis Mission last week. Can you talk about how this will impact the business for Veritone and public sector opportunities that we've spoken about?

Yes. I think what you're referring to last week, the White House again issued an executive order named the Genesis Mission. I hope everybody has a chance to go through that. It's not too long. We'll follow up to all the attendees today and give you a link to that. But candidly, this was an exciting advertisement for Veritone in our core thesis and mission. And it really sets out 2 things. One is, we obviously know about everything that OpenAI and NVIDIA and everybody else are doing as part of this huge AI infrastructure build-out.

But I think finally, Michael Kratsios, the CTO of the White House and respective team, Gil included, they really set forth, which I find a lot of passion in, and I think is very important is what is our nation's core foundational model thesis as well, right, not just being overly reliant candidly, on just third-party companies. And I think this was a major step forward. But here is the 2 main points of this executive order that I think are so relevant and powerful to Veritone.

First and foremost is, if you read this, this is as much about data provenance and leveraging proprietary data as it is about having our nation build its own proprietary AI foundational models. So please read through that. Probably the most impressive thing is for our nation and for with this executive order is the President's mandate to activate and leverage the legacy existing research and science foundational data sets that the United States has been producing for decades and decades, which includes petabytes and petabytes of audio and video as well.

So again, this for us is a testament to our maniacal focus on all things data and the tokenization of that data to turn it into clean both semantic utility value, as Chris went over in depth today as well, but also providing future AI model training. So again, I think this is a validation of our business model.

And yes, if you're going to ask the question is, we hope to be part of this exciting initiative, obviously, is one of the companies that has been processing petabytes of audio and video for decades now and producing high-yield tokenization, we hope to be a part of this exciting initiative with the government directly in the future.

So again, I think this was a phenomenal thing. I'm personally proud of it. I'm proud of it as an order for our nation. And I do think it's very relevant and helps validate Veritone's core business model and thesis.

Thanks, Ryan, very helpful color. Moving to the next question. I know during today's event and especially during Veritone's last earnings call, data tokenization was kind of at the forefront of your prepared remarks and some of the industry insights that we've been speaking to. Why is this such a major focus for Veritone. And can you help us kind of contextualize what broader tokenization trends you're seeing across the market and how this impacts our market opportunity?

Sean, why don't you start off and then I'll try to follow up and add to this where it deems fit.

Sure. Happy to. As kind of both Ryan and Chris said throughout this, for more than 10 years, Veritone has been turning audio and video data on behalf of our customers into tokens and ultimately becoming one of the top-tier semantic token factories that exist, frankly. And our solutions, as kind of Chris demonstrated, we're built to transform that raw audio and video into valuable metadata rich tokens that are basic building blocks for powering our applications, powering certain discovery of assets, CBS News kind of demonstrated earlier. So this really puts Veritone in a great spot because there's still the growing need for structured data that can actually be utilized and then utilized and monetized. So these tokenization processes and trends that you're seeing are a massive opportunity for Veritone and our customers to make the most out of their assets and be able to provide high-quality data.

We've talked a lot throughout this as data as a currency era is upon us. I believe I read from Mordor Intelligence, that they global tokenization asset is expected to hit somewhere north of $10 trillion by 2030. And Veritone, as we sit here today, we're perfectly positioned here to take advantage of this and we -- that with our strong track record and subject matter expertise.

And really, at the end of it, it's our aiWARE platform that can be the engine for that global data economy as well as our Veritone data refinery as a key of turning that into a powerful new way to grow organizations and Veritone and our customers. So as the data and AI industry continues to take off at lightning speed, as I mentioned, aiWARE alongside VDR and our decades of subject matter expertise, places Veritone squarely at the center of this expanding opportunity.

Great. Thank you, Sean. I don't need to add anything to that.

Thanks, Sean. And then I think we'd love to hear from you too as well, Chris, Ryan, on this one. When you talk about the versatility and flexibility of the aiWARE platform that powers Veritone's products, are there any limits to the platform and its capabilities?

aiWARE is truly an impressive technology. It's what attracted me to Veritone years ago. This vision for an AI operating system before AI was mainstream. First and foremost, it fuels all our applications like we've talked about before. Both the purpose-built ones for our BUs as well as all the custom applications, workflows and dashboards that we've built over the years through our professional service offerings.

Secondly, as I've said multiple times, and you've heard today, aiWARE shines at high-scale ingestion and processing across all formats and across all environments. We talk a lot about audio and video, but we're also very competent in structured data as well as document understanding. For structured data, we get log files from advertising campaigns or telemetry from drone footage, and we ingest that and map that to our systems. On document understanding, opening up these PDFs and understanding all the structure of the tables, graphs and text, it's not trivial, and we do that all day long as well.

Thirdly, aiWARE is hardened and meets all the security, reporting and compliance requirements across all of our client engagements. These are very complex operational demands that we have been consistently meeting for years.

Lastly, aiWARE is open. What I mean by this is, it's deployment and model agnostic. On the deployment front, we host our applications in both Azure and AWS, but we're also deployed in GovCloud as well as FedRAMP for our secure federal clients.

On the model front, we have an open model framework. What this means is you have a seamless integration and management across hundreds of pretrained AI models. This allows clients to pick and choose the model that works best for their business and evolve over time as their business evolves. AI is changing and businesses are changing and aiWARE supports this evolution.

Given all this, aiWARE can really support endless AI solutions, and it's really architected for innovation. Builders can get up and running quickly, prove out the use case and start scaling operations from there, all possible from the years of technology investment into aiWARE. It's really an extraordinary platform and it's only accelerating.

I think I'll add just one comment. I think this was kind of an add-on from Joshua Reilly at Needham, who was asking about our focus, our ability to continue to develop prebuilt applications for specific verticals like Redact for public safety, like DMH for commercial.

Well, first is, hopefully, you've got a good insight on to the aiWARE platform. What is so powerful, it allows us really in a low-code workflow type of environment to spin up and activate against new market needs or problem sets very quickly. Once we have the data, once we've integrated it, once we've tokenized it, frankly, we can build new custom workflows incredibly fast.

Our end users may see a huge difference when they're using the redaction software. But candidly, that's just a different workflow. Frankly, to us, from a technology perspective, there's not much of a difference than an archivist at CBS News using the aiWARE applications to solve their problem sets.

So I think we're in a great position to continue to build these point-specific or solution-specific applications, frankly, that are easy to sell and they're easy to frankly train a new end user to use quickly but we are also adept at helping build custom solutions or custom workflows for much larger and more dynamic customers like the Air Force. So hopefully, that answers your question. But again, it all terminates and begins and ends with the aiWARE platform.

Thanks, Ryan. And then turning to the chat, and I believe this was a question that came up during the demos. So from the audience on the DMH product, who would be using this? Is it the owner of the content or others looking to access the content? And if you could expand on the utility of this as well as the revenue model.

Chris, I'll take this and then maybe you can add a couple of comments if you want. First, DMH, Digital Media Hub, is primarily used by our commercial customers, heavy usage by media and entertainment, sports and news organizations. And to the question is pretty astute. It is actually used by both internal teams, right, such as the team at CBS News or at CNBC or others. But it also has permissioning, so third-party customers or those who have credential rights can access specific content objects as well.

So in the context, let's say, the Masters golf tournament, they internally use it for management of their files, the speed of building highlights, the speed of preparing content for sponsor activation, but on the flip -- but on the opposite side, with those who have permission, they can log in and get access to those files almost immediately.

So when you see like Tiger Woods footage from the Masters or more recently, Rory McIlroy from the Masters show up in an advertisement in like just a few hours, that's because Wieden + Kennedy and the advertising agency or other third-party customers have also permission-based access.

And again, that permissioning is critical. Again, this goes back to the core fundamental functions of aiWARE, where once the data is ingested, it has object-level permissioning down to very specific, I'll say, elements of audio and video. So again, this is a highly -- high-use critical component for both internal use cases by a lot of our DMH customers but also has permission-based support for third-party entities who are gaining access to those files as well. But thank you for the question.

And then this was originally a question from Joshua Reilly at Needham and seeing a few other questions in the chat just around the hyperscalers and their use of the VDR product. So for VDR, it seems initially the largest hyperscalers are proving to be customers of Veritone but that seems to be expanding into a much wider group now, Ryan, as you mentioned. How big is this customer set for VDR over time?

Sean, why don't you take this?

Sure. I just want to reiterate kind of as Ryan said earlier on, we're still on track to secure contracts with the majority of all the hyperscalers by the end of the year, which is great because let's face it right now, they're the big spenders in the AI space. But I just want to reiterate that here, this is still very early days.

So today, we may see 30 or so major players in this space, but we expect that number to explode into the hundreds over the next couple of years. New groups will be building fresh applications for unique use cases. Some may be layering on top of existing technologies, fine-tuning models, creating very domain-specific use case models.

There's a lot of different opportunities. But what's at the base and foundation for all these opportunities is the need for quality data to get started. And that's where Veritone has the opportunity to come in. That's where -- that's our expansive opportunity where we have that chance to be their trusted partner to helping them quickly obtain annotate, prepare that necessary data, we very quickly have the opportunity to really become mission-critical in their processes.

I think I would say if the total ecosystem where the super majority of the VDR customers is customer -- 50 companies or less today, I expected that to be an order of magnitude bigger over the next couple of years or even into the thousands. And again, that -- we have that conviction because candidly, the cost of building models continues to drop. Obviously, with this huge AI infrastructure build-out, the cost of compute, the cost of storage continues to fall and thus, a lot of these companies, many of which are venture-backed or private equity-backed startups have more -- now more capital that they could use to actually train the models.

So again, we are already seeing new players in the space, and we expect the number -- the unique number of customers that we are selling and licensing this clean data to them to increase over the next couple of years.

And then on the heels of that, another question from the audience and the chat around the VDR product offering. Can you please explain how the VDR contracts are structured? So are these consumption-based, subscription based or onetime fees and add a little bit more color on how those revenues come in?

Today, the majority of the VDR contracts are consumption-based. There are elements that you could deem as subscription-based. And we do see sort of the business models change over time. Remember, in the VDR business, we sit on both sides of the equation, which is exciting and creates an interesting moat for us.

On one side, we are often the exclusive, I would say, representation technology partner of the content provider, so let's say, CBS News. But on the flip side, we are also the vendor or partner with the buyer, one of the hyperscalers, right, who is looking to procure and license that proprietary data. But again, as it is today, the majority of these contracts is consumption based. And in certain times, depending on what the needs are, some of these contracts are actually onetime fees. We deliver the fully clean prepared data set, and that's it, and we recognize that revenues in quarter.

But again, you will see that model change over time. As we process multiple orders, if you want to call it that way, or multiple data programs for a single existing customer, you're starting to see specialization happen already, which by the way, is really unique because we can offer specialization, meaning I can deliver video files or audio files with different structured formatted metadata or tokens to different hyperscalers to suit their needs to make their process more efficient. And thus, you could see more subscription like of revenues in the future. But to be very clear today, the majority of our VDR revenue is -- and the recognition of such is primarily consumption-based, which -- and also includes elements of onetime fees.

Why don't I just take a couple? I know we're running out of time. Cate, how much more time do we have before we need to wrap up here, just a couple of minutes?

I have just a couple of minutes, around six minutes but Ryan, if you want to take...

There's a few interesting questions. I apologize. I'd love to answer all these, and we can try to get to some follow-ups with our IR team and our team to reach out to these people. But there are a lot of great questions.

One obviously says is, "are we expecting any new hyperscaler contracts could be released before end of year?"

To be clear, we're signing those in quarter, and we have signed new ones in quarter for fourth quarter. I think you're looking for more public disclosure. Candidly, we would love to be able to screen the mountain tops. I mean, obviously, people can infer the type of companies and which specific companies we are working with and selling to. But obviously, we're going to honor the confidentiality of those agreements.

And this kind of ties into my answer on the President's Executive Order. Again, it's not just the models, it's the data. And often, what data these companies are using and how they're using it is frankly like state secrets, okay? State secrets.

So again, all I can reaffirm what Sean said is we are working with the major players today. And hopefully, we'll be able to mention their name specifically more publicly in the future but be reassured that we are working with the majority of the players already today some of the biggest names you know. And we -- again, we expect that to have relationships and active engagements with the super majority of all of them by the end of the year.

And then I'll answer one more question and then we can kind of wrap it up for this session today. But again, as much of our fundraising has been secured from institutional investors, do you see prospects for strategic investments in the company in the future?

Absolutely. We've had opportunities to enter into more strategic relationships with different major players over the years. Historically, we've elected not primarily for being independent and frankly, remains Switzerland.

What I mean by that is, as Chris articulated earlier, our platform is completely agnostic. We have customers right now using AI and applications that are using those and those are deployed on Azure, on AWS, in FedRAMP and other providers.

We also work with countless AI model developers as well as us producing our own AI models internally. And so we want to make sure that when we evaluate strategic opportunities, that it does nothing to disrupt our independent open thesis which has proven to be very successful. But again, we are always opportunistic. We are always looking at what type of strategic relationships would be a force multiplier for the company, which may or may not include a direct investment as well. But thank you for a great question.

Thanks, everyone, for all the questions. I'm going to put in the link to our IR e-mail alerts in the chat. So would love for you all to subscribe, and that's where we will be sharing further updates, whether it's press releases or product updates as well as the upcoming conferences Ryan and team will be attending. So really appreciate everyone attending today's session, and we'll turn it back to you, Ryan, just to close this out.

Again, thank you, everybody. A few things I want everybody to get from the session on webinar today is: first, just alignment, if you can hear how Chris speaks to our opportunity through the lens of product and engineering, how Sean speaks to it from a go-to-market and revenue generation and obviously, how I speak to it from a strategic perspective and market perspective is we are all in on what we've been doing for a decade, right? Our ability to be the experts on leveraging AI and fueling AI with unstructured and structured data, tokenizing it and bringing almost immediate ROI and utility value to the opportunity and to the marketplace and to our customers. So we're very, very excited about that.

Please pay attention to this phenomena of the tokenization of real-world assets. That's what we do. Audio and video is a real-world asset, and we are definitively a market leader in that space. We're very bullish about VDR. We're very bullish about what we're doing in the public sector and in commercial. But again, I think we have many different levels of moats. People always ask what's proprietary, what's different?

To be clear, it's all of it combined, right? Achieving this level of scale and achieving this number of real customers that are going to -- most of which have been customers of ours for years, we have almost like the network effect where we can leverage our same core operating system, aiWARE, to service radically different types of customers in completely different market segments, but all running on the same stack. That's incredibly hard to do. It's incredibly challenging. And Veritone has not only been doing that successfully for years, but we're also servicing some of the largest enterprise customers. Look at the names of our customers that have been with us for a long time.

Anyway, I want to thank my team. I especially want to thank representatives from CBS and police departments for joining us today and supporting us. And again, we will try to answer all your questions and follow-up via e-mail, but I appreciate your time today, and we look forward to closing out a very strong year. Thank you.

Good day, and welcome to the Veritone Inc. Third Quarter 2025 Financial Results Conference Call. [Operator Instructions] Please note this event is being recorded.

I would now like to turn the conference over to Cate Goldsmith, Investor Relations. Please go ahead.

Thank you, and good afternoon. After the market closed tonight, Veritone issued a press release announcing results for the third quarter 2025 ended September 30, 2025. The press release and other supplemental information are available on the Investor Relations section of Veritone's website. Joining us for today's call are Veritone's President and Chief Executive Officer, Ryan Steelberg; and Chief Financial Officer, Mike Zemetra, who will provide prepared remarks and then open the call up for a live question-and-answer session.

Please note that certain information discussed on the call today including certain answers to your questions will include forward-looking statements. This includes, without limitation, statements about our business strategy and future financial and operating performance. These forward-looking statements are subject to risks, uncertainties and assumptions that may cause the actual results to differ materially from those stated. Certain of these risks and assumptions are discussed in Veritone's SEC filings, including its annual report on Form 10-K. These forward-looking statements are based on assumptions as of today, November 6, 2025, and Veritone undertakes no obligation to revise or update them.

During this call, the actual and forecasted financial measures we will be discussing include non-GAAP measures. Reconciliations of these measures to the corresponding GAAP measures are included in the press release we issued today.

Finally, I would like to remind everyone that the call today is being recorded and will be made available for replay via link in the Investor Relations section of Veritone's website at www.veritone.com. Now I would like to turn the call over to our President and Chief Executive Officer, Ryan Steelberg.

Thank you, Cate, and thank you, everyone, for joining us today. As the iconic line from one of my favorite childhood shows, the A team goes, I love it when a plan comes together, simple yet profound and never more fitting than it is for Veritone today. I look forward to sharing the details of our exceptional quarter and the strategic momentum now propelling Veritone forward.

But before we dive in, I want to take a moment to highlight the substantial progress we've made over the past 24 months, progress that has fundamentally strengthened our company and position us for sustained growth and success. In early 2023, we set a bold and deliberate course to realign Veritone with its aiWARE first mission, fortify our financial position and ignite durable strategic growth. We've executed with precision and purpose every step of the way, transforming vision into measurable achievement. Today, I'm proud to declare success across these core initiatives. Veritone has not only regained its footing, but stand stronger, more focused and more strategically positioned than ever before.

First, let's start with the balance sheet, where we have completely flipped the script. A few years ago at their peak, we carried nearly $80 million in high-cost term debt and roughly $200 million in convertible debt. Today, I'm proud to announce that we've completed an agreement to retire the entirety of our term debt and repurchase approximately 50% of our outstanding convertible debt, while maintaining material net cash on the balance sheet to fully fund operations through 2026 and achieve our profitability goal in the later half of next year. As a result, our annual debt service burden has now dropped from over $14 million a year to approximately $800,000 a year, a transformative shift that fundamentally strengthens our financial foundation and future trajectory.

Our balance sheet, historically anchor relating with debt now provides renewed flexibility and stability, enabling us to fully capitalize on this hyper-growth market opportunity.

Second, after years of disciplined financial execution and strategic reorganization efforts that have generated tens of millions of dollars in savings. Our operating model today is now tightly aligned with both our scale and our strategic focus on our high-growth AI software products and services. With this foundation in place, our confidence in achieving operating profitability in late 2026 has never been stronger.

Third, we have refocused Veritone squarely on our AI routes. Our unmatched expertise and unstructured data and our market-leading intelligent AI applications. Today, Veritone stands on a trajectory of strong, strategic and profitable growth, powered by our proprietary AI operating system, aiWARE, which fuels the global data economy by generating trillions of tokens every quarter. As the AI economy accelerates, the data economy is expanding right alongside it, and Veritone's strategic positioning and market timing could not be more perfect.

Veritone has emerged as a leading semantic token factory for video and audio, a service we call the Veritone Data Refinery or VDR, built entirely on our own proprietary aiWARE platform. Our tokenization engine not only powers our own AI workflows and customer applications, but now serves as the foundation for a powerful new monetization framework. Our VDR product offering is leading the way as the first of several major monetization initiatives, which we plan to roll out beginning in the first quarter of next year.

For more than a decade, Veritone has been tokenizing video and audio the fastest-growing segment of unstructured data at massive scale. More importantly, we're doing so in a transactional utility-driven format that delivers immediate value and measurable ROI for our customers and ultimately, for our investors through our expanding suite of innovative AI products, services and applications. Our tokenization and monetization strategy is designed to exist and operate both on chain in the context of blockchain and independently as it has done profitably and efficiently for more than a decade. The data as a currency era has arrived, and Veritone is uniquely positioned to capitalize, executing from a position of strength, expertise and leadership in a rapidly expanding multi-billion dollar market.

In fact, according to [ Mordor ] Intelligence, the global tokenized asset market is projected to reach $13.5 trillion by 2030. It is going to be a very exciting next few years for Veritone and our customers.

Now turning to our strong quarterly results. Veritone delivered revenue of over $29 million. This performance underscores the accelerating demand for our market-leading aiWARE solutions, data products and intelligent applications, representing a 32% year-over-year increase in revenue for the overall business. Looking specifically at software products and services, which includes VDR, revenue grew by impressive 55% during the quarter. Excluding Veritone Hire, our software products and services revenue surged by more than 200%.

As global investment in AI infrastructure intensifies, demand for high-quality training data and for our AI applications continues to rise in parallel. This powerful combination is driving sustained high margin expansion across our software products and services business. And as Michael will outline shortly, our bottom line performance this quarter was equally strong, delivering more than a 50% year-over-year improvement.

Now turning to the partnerships, contract wins and products that underpinned our strong results. This quarter, our Veritone Data Refinery, VDR business continued to deliver exceptional performance. VDR transforms raw unstructured audio, video images and text in the high-quality tokenized data sets that power and fine-tune the world's most advanced AI models. We have established ourselves as a premier data and model training partner. And during the quarter, we secured several significant new VDR customers, including contracts with multiple major hyperscalers, further solidifying VDR's position as a critical enabler in the unstructured AI training data ecosystem.

VDR's accelerating momentum underscores not only the surging buy-side demand for clean, model-ready training data, but also the expanding monetization potential of premium video and audio assets themselves. Our qualified VDR pipeline and bookings now exceed $40 million, reflecting 100% growth quarter-over-quarter. Importantly, that figure represents only current bookings and near-term opportunities. Our total VDR pipeline now spans multiple sectors, regions and time horizons, positioning Veritone for sustained growth as AI developers increasingly move beyond Open Web data in favor of proprietary multimodality data sets, precisely the domain VDR was built to serve.

Looking forward, we are confident that by the end of 2025, Veritone will hold active contracts or projects with every major hyperscaler in the market. Our VDR solutions are indeed growing quickly and provide great strategic leverage for future growth, but we are equally excited about our AI applications business, again built and deployed on the same aiWARE platform. For Veritone and our customers, our AI applications are the delivery vehicles for data-centric high-value use cases, driving efficiency gains and ROI.

In the quarter, we achieved significant progress across our commercial enterprise applications business. We signed 27 commercial agreements, including partnerships with ESPN, the NCAA and News Max. And just last week, we announced an expansion of our long-standing relationship with CBS, now encompassing many of CBS Media Ventures flagship programs, including Entertainment Tonight and Inside Edition, which are now available for licensing through Veritone. These partnerships exemplify how Veritone empowers the world's leading media organizations to unlock and monetize their vast content and data archives.

Furthermore, through VDR, we are transforming these assets into clean, searchable model-ready data sets, creating high-value training material for AI models and in turn generating meaningful reoccurring revenue streams for both Veritone and our partners.

As content libraries expand and data volume surge, Veritone is uniquely positioned to unlock maximum value for commercial enterprises and IP owners. Our technology enables partners to fully monetize their content archives, transforming dormant assets into active revenue-generating resources while simultaneously producing high-quality model-ready data that fuels the next generation of AI innovation.

Turning to the public sector. In Q3, we closed 82 contracts from new and existing customers across federal, state and local agencies. We also added 30 new agencies during the quarter, including the IDMs win at a top 5 law enforcement agency and an annual items renewal with a Department of Law Agency deployed in their private cloud. Despite the federal government shutdown, we continue to be actively engaged with our customers and prospects in the Department of [ Law and ] DHS as we expand the footprint of Veritone solutions. Our public sector pipeline now approaches $218 million in qualified opportunities, up from $110 million earlier this year, a testament to both our accelerating demand for AI-driven solutions and Veritone's reputation as a trusted technology partner to law enforcement, defense and [ Fed SIV ] agencies.

We also saw encouraging momentum internationally, including closing a partner-led transaction for a national police agency in the EU and advancing multiple opportunities in the U.K. for Veritone Redact, Veritone items and a new workflow solution on aiWARE. Our international pipeline now exceeds $28 million and continues to grow rapidly, reflecting rising product market fit and strong demand among agencies engaged in government initiatives in key global markets, all supported by our secure and GDPR compliant infrastructure.

Our awardable status on the Department of Wars P1 marketplace positions Veritone to capitalize on the expanding wave of government AI investment. This designation enables us to engage directly with DoD and civilian agencies, shorten procurement cycles and deliver mission-critical AI solutions faster, helping government partners tackle their most pressing operational and analytical challenges.

Subsequent to the end of the quarter, we introduced a suite of new advanced capabilities within Veritone Redact, our aiWARE powered SaaS application that automates the redaction of sensitive information in audio, video and text. The latest features include AI-powered voice masking, inverse blur and multi-language transcription in 64 languages, all designed to enhance privacy, compliance and efficiency. These advancements are already driving expansion within existing customers and creating new opportunities for agency and enterprise partnerships.

Before turning things over to Mike, I want to reflect again on how far we've come and where we're headed. I could not be more excited about Veritone's future, and I remain deeply grateful to our investors, partners and employees who have supported us through this remarkable transformation. I have never been more confident in the future of Veritone, our business, our people and the market opportunity before us. We are leading the tokenization and monetization of valuable unstructured data, and have firmly established ourselves as a key enterprise leader in both the AI and data economies. With a strengthened balance sheet supported by high-quality equity raises completed in the past few months and the material debt reductions announced today, Veritone is entering a new phase of execution, one defined by rapid profitable growth. With our exceptional talent dynamic AI platform, market-leading applications and a robust expanding pipeline, Veritone's future has never been brighter. Over to you, Mike.

Thank you, Ryan. We started the second half of 2025 with one of our strongest quarters to date with Q3 revenue surpassing our recent guidance, led by our software products and services growth of over 55% year-over-year and 48% year-over-year improvements in our bottom line non-GAAP net loss. In addition to this momentum in our results, we secured over $100 million in equity capital in September and October 2025, substantially improving our longer-term liquidity position.

As I will explain in more detail, I'm thrilled to announce today that we will be paying off 100% of our term loan and paying down 50% of our convertible debt, further improving our liquidity position. The results we achieved this quarter are the culmination of years of hard work and strategy coming to fruition. During my prepared remarks, I will discuss Q3 year-over-year performance in KPIs, which exclude the results of our media agency, which are presented as discontinued operations and the corresponding historical financial periods, balance sheet and liquidity position, including the recent capital raises and paydown of our debt and Q4 and fiscal 2025 guidance, starting with Q3 2025 performance.

Q3 revenue was $29.1 million, up $7.1 million or 32% from Q3 2024, driven by an $8.1 million increase from our software products and services, offset by a $1 million decline in our managed services. The $8.1 million revenue growth in software products and services was driven by commercial enterprise, which improved $7.8 million year-over-year and public sector, which grew over 25% year-over-year. The growth in commercial enterprise was led by Veritone Data Refinery or VDR. VDR, which launched in Q4 2024 is one area where we continue to see very strong growth and today has a near-term sales pipeline in bookings of over $40 million, up over 100% from our guidance in Q2 2025.

Overall, Veritone Hire remained relatively flat year-over-year, driven largely by the hiring softness in the macro economy. Excluding Veritone Hire, our software products and services grew more than 200% year-over-year. The growth in the public sector was driven by the rollout of larger deals executed in the first half of 2025, including the Department of Defense and larger public safety agencies. We expect these larger public sector deals, coupled with our expanding public sector pipeline to generate substantial growth beyond 2025, which I will explain in more detail later.

The $1 million decrease in Q3 managed services revenue was principally due to a decline in representation services by a decrease in our VeriAds services, offset by a slight improvement in content licensing. As we previously discussed, we expect this negative trend in representation services to continue through 2025 or until the macro economy shows demonstrated improvements over 2024.

Turning to key performance metrics across our software products and services in Q3 2025. ARR of $68.8 million, up 9% from Q2 2024 of $63.4 million and 12% sequentially from Q2 2025, from increased consumption-based revenue largely driven by VDR and stable recurring SaaS-based revenue. Overall, ARR from consumption-based customers increased 26% year-over-year and 74% sequentially from Q2 2025. Recurring subscription-based SaaS customers were up slightly by 3% year-over-year. As of Q3 2025, 73% of our ARR was from subscription versus consumption-based customers as compared to 76% at Q3 2024.

Total new bookings of $21.5 million, up $5 million or 30% year-over-year, primarily due to larger VDR bookings across our software customer base. Gross revenue retention continued to be above the 90th percentile. Total software products and services customers of 3,021, which was down 9% year-over-year, predominantly from our commercial enterprise sector, which includes lower consumption-based customers across Veritone Hire and the continuing impact of sunsetting legacy CareerBuilder customers post the June 2023 acquisition of Broadbean and smaller customers as we focus on larger ARR opportunities, offset by an increase across public sector, largely from the growth in public safety customers.

Q3 GAAP gross profit was $18.7 million, compared to $14.7 million in Q3 2024. An improvement of $4 million, largely driven by growth in software products and services, including VDR, with GAAP gross margin of 63.3% as compared to 66.6% in Q3 2024. Excluding noncash depreciation and amortization expense, Q3 2025 non-GAAP gross margin was 70.6% as compared to 71.2% in Q3 2024, a decline of 60 basis points. Note that included in Q3 2025, a certain onetime software revenue that has very high gross margins, while VDR gross margins remain close to approximately 40%.

As I will discuss later, we do not expect the same level of onetime software revenue to recur in Q4 2025. And as a result, are forecasting Q4 2025 non-GAAP gross margins to be closer to 60%.

Q3 operating loss of $15.8 million improved by $6.7 million or 29% year-over-year, primarily driven by the increase in gross profit, offset by lower operating expenses. Net loss from continuing operations was $26.9 million, an increase of $4.4 million or 20% compared to Q3 2024. The year-over-year increase was principally driven by an $8 million noncash charge in the estimated fair value of the earn-out expected from the divestiture of Veritone One recorded in Q3 2025 and a $2.2 million change in our Q3 tax provision, offset by the $6.7 million improvement in operating loss.

Non-GAAP net loss from continuing operations was $5.8 million as compared to $11.1 million in Q3 2024, a $5.3 million or 48% improvement. The improvement was principally due to the year-over-year growth in non-GAAP gross profit, coupled with lower operating losses driven by increased discipline on cost management. As I will explain further, these reductions will provide us with a more efficient cost structure as we manage towards our planned growth throughout the remainder of 2025 and targeted profitability in the latter part of 2026 and beyond.

Turning to our balance sheet. As of September 30, 2025, we held cash and restricted cash of $36.5 million as compared to $16.9 million at December 31, 2024. The net change in cash reflects net cash outflows from operations of $41.2 million, principally driven by our non-GAAP net loss of $25.6 million, deferred purchase consideration of $1.2 million, and interest paid on debt of approximately $5.8 million, coupled with the timing of working capital in the quarter, driven largely by the increase in AR due to the growth in revenue in the period. Offset by net cash inflows from investing and financing activities of $64.9 million, driven by net cash inflows of $70.9 million from equity offerings through Q3 2025, partially offset by $5.8 million in debt principal payments and $3.6 million in capital expenditures.

Turning to liquidity today. In September 2025, we completed an underwritten equity offering, selling 9.5 million shares of common stock priced at $2.63 per share and an overallotment of 1.4 million shares granted to the underwriter, which was exercised in full for total gross proceeds of $28.8 million. In October 2025, we completed a registered direct offering, selling 12.9 million shares of common stock priced at $5.83 per share for total gross proceeds of $75 million. Immediately following the October offering, we held cash and cash equivalents in excess of $100 million.

At September 30, 2025, our consolidated debt is down from a peak of $201 million in December 2021 to approximately $126.7 million. Subsequent to September 30, we paid down $3.6 million of our term debt through deferred purchase consideration received in October 2025, bringing our debt to $123.1 million, comprised of $31.8 million of term debt and $91.3 million of convertible debt.

Today, we announced that we have agreed in principle with certain debt holders to pay off 100% of our senior secured term debt and buy back 50% of our convertible notes for a total of approximately $77.5 million of consolidated debt principal in exchange for approximately $77.5 million of cash and 625,000 shares of our common stock valued at today's closing price. Immediately following this debt payoff, our unencumbered consolidated cash is approximately $34 million, which is sufficient to fund our operations over the next 12 months at a minimum. Post this paydown, our remaining debt will be approximately $45 million, comprised solely of our 1.75% convertible notes due November 2026.

By completing this transaction, we free up an estimated $13 million of annualized debt carry costs and substantially improve our liquidity position and future cash flow outlook.

I want to underscore what an impressive and important step reducing our debt is. The improved flexibility and stability we now have as a result of our strengthened balance sheet will allow us to focus on reaching our growth potential to meet the hyper growth market opportunity we face. That said, we will continue to be opportunistic with continued focus on further improving our current liquidity position and balance sheet.

At September 30, 2025, we had 7.9 million shares issued in outstanding and 2.5 million warrants outstanding to our debt holders.

Now turning to updated fiscal Q4 2025 and full year 2025 guidance. Our software products and services revenue pipeline and long-term outlook continue to be at all-time highs. More specifically, we continue to see strong demand across the approximate $10 billion global digital evidence management market, our public sector and VDR pipelines continue to grow. Collectively, our backlog and sales pipeline across our core AR platform is in excess of $200 million today. And as Veritone remains uniquely positioned to capture even more opportunity in the data as a currency market, we expect that pipeline and our potential to monetize our trove of tokenized audio and video to further increase.

More specifically, in Q4 2025, revenue is expected to be between $33.4 million and $39.4 million as compared to $22.4 million in Q4 2024, a 63% increase at the midpoint and 25% sequentially from Q3 2025. In Q4, we expect our software products and services to increase more than 75% year-over-year, led by the growth in public sector and commercial enterprise. Specifically, we expect our public sector revenue to grow close to 50% year-over-year, and our commercial enterprise revenue led by VDR to grow more than 75%. Our Veritone Hire products and services are included in this growth, and we expect Veritone Hire to be slightly down year-over-year given the current macroeconomic environment.

Consistent with Q3 2025, our managed services is expected to be down year-over-year, principally due to the representation side of our business, which is experiencing some slowness as a result of the more challenging macroeconomic environment. We expect Q4 non-GAAP gross margins to be approximately 61% to 60%, driven by the forecasted higher mix of VDR revenue in the period. Q4 non-GAAP net loss is projected to be between $1.5 million to $5 million as compared to $9.7 million in Q4 2024, representing a 66% improvement at the midpoint and a 44% improvement sequentially from Q3 2025.

Turning to fiscal 2025 outlook. We are updating our prior guidance for fiscal 2025, which we are expecting revenue to be between $109 million to $115 million, which at the midpoint represents a 22% increase year-over-year and non-GAAP net loss to be between $31.6 million to $26 million, representing a 29% improvement year-over-year at the midpoint. The change is reflected of the timing shifts in revenue, coupled with the compression in gross margins on VDR in 2025, which we expect to improve upon in fiscal 2026.

Before closing the call, I'd like to remind everyone listening that Veritone will be attending Needham's Virtual Sixth Annual Tech week, November 20 through the 24 and UBS' Global Technology and AI Conference, December 1 through the 4 in Scottsdale, Arizona. That concludes my prepared remarks. Operator, we would now like to open up the call for questions.

[Operator Instructions] And your first question comes from Joshua Reilly with Needham.

Nice job on the quarter here. Maybe just starting off on the Q4 revenue guidance, there's a $6 million range, obviously there between the high and the low end. Maybe we could just review what are the puts and takes that would get your expectations in the business to the high end of the revenue guidance for the quarter? And then maybe what would drive it to the lower end of the guidance for the quarter?

Yes. I think it's just timing and velocity on some of the larger VDR deals. Again, as I sit here today, obviously, I was very bullish in both the words that I chose and sort of my tone and disposition. So obviously, we're going to push to get to the highest. But again, relative to the size and magnitude of some of these VDR deals and some of the commercial deals and some of the, I'd say, to a lesser third degree, the timing on some of the public sector Fed deals, that's really going to be sort of the inputs to the ranges. But as I sit here today, very, very optimistic, very excited and most importantly, we have, I'll say, a very mature pipeline to substantiate that range and give us the opportunity to hit the higher end of the range.

Got it. And then on the 100% quarter-over-quarter increase in the VDR pipeline, can you just help us understand what is -- what are you doing from a go-to-market perspective to kind of drive that pipeline growth? And then as we look forward into 2026, are you expanding the go-to-market efforts there to further expand the VDR pipeline? Or do you kind of have the people in place to manage the upside opportunity there? Maybe kind of help us understand the dynamics there?

So this is a really exciting one. And I want to be clear, I mean, a lot of our growth, obviously, this conversation today was dominated by VDR, but there's the other side of VDR, which is the supply side. And that's the side that, frankly, before we even introduced the concept of VDR, we've been servicing and selling AI-based software to a lot of media and entertainment and content groups for years. So again, where we see the great opportunity, and what makes us very, very unique as compared to really anybody who's in the AI training data ecosystem is that we -- to be clear, we represent and generate revenue from both sides, right? Again, both from the buy side, those are the hyperscalers and the model developers we're selling to, but also the representation side, those are the media entertainment and other data suppliers that we represent, but also have been selling software, too.

So again, your question specifically, as I stated in my prepared remarks, we do believe that we will be engaged in doing the active projects and business with nearly all, if not 100% of the major buyers today in the space, but the space is growing quickly. And so what we're very focused on is to make sure that not only are we continuing to take care of the larger transactions with the bigger AI model development shops, but also, we do believe because of the cost basis for compute, storage, et cetera, continues to come down, we do believe that there's going to be more entities, different companies to sell to, to sell these training data sets too.

So we do believe that we want to continue to strategically expand our sales force. We are building a pretty reputable brand as it relates in the AI training data market. But we -- but to be clear, we want to make sure we don't simply focus on the buy side. We continue to want it and what we have seen. We do really didn't touch upon it that much in my prepared remarks. But we are also seeing an increased velocity of the data providers coming to Veritone. We signed a multitude of different -- we obviously mentioned a couple of the bigger brands on my call, but we saw a multitude of different customers now coming to us not only for the VDR solutions, but for our AS software side of the equation.

So again, I wanted just to summarize that, we are unique in the sense that we sit and represent and sell to both sides of the equation, the sell side and the buy side. And yes, we will be investing strategically more into the go-to-market to increase velocity to make sure, again, that that's not a limiting function going into next year.

Got it. That's very helpful explanation there. And then maybe on the Q4 guidance for the public sector, I believe you said it was going to be up 50% year-over-year. How much of the -- what are you factoring in on the federal side with the government shutdown, obviously, still in place here to kind of hit that number? And are you making any assumptions that some federal -- U.S. federal deals closed and kind of hit that number, which would require the government to reopen? Or just kind of how are you kind of calibrating those assumptions?

I think we've taken that into the handicap for that guide. Obviously, as a percentage of the overall revenue base, it's not that large. And so meaning to sway us to take a more pessimistic view of hitting that guide. To be clear, we still are generating growth. We still are closing new businesses and revenue in the federal space. But have we seen potentially some delays in some of the revenues that would have contributed in this quarter, we have. Thankfully, the other sides of the business have grown just sort of outperformed, and so that's where we're extremely bullish overall. But again, the short answer is, like everybody, we hope this government shutdown ends sooner than later. And just what if all these things start hitting all cylinders.

So again, overall, we're very excited. And I think that a lot of investors need to when they're thinking about some of our market peers in the market, we're not a one-trick pony in a certain vertical, right? We have the exact same powerful technology stack aiWARE that's being sold effectively into both commercial and public sector, and I think investors should take note of that. So again, this is another great example of having a focused yet diversified business can be very, very attractive.

And your next question comes from Glenn Mattson with Ladenburg Thalmann.

I just know VDR is the bigger story, of course, but I wanted to just drill down on that public sector stuff for a minute. What I'm curious is -- is it that the federal shutdown is causing a bit of a slowdown. And in particular, is it related to that Air Force contract only because I want to understand if it's -- as I think about my forecast for next year, if there's like a snapback as a temporary thing or if it's maybe that something else going on? Just if you could elaborate.

No, I think this is a short-term blip. Again, in terms of at least my perspective for your modeling, that this was weeks, if not a couple of months delay. But again, it's not binary, to be very clear. So it will have a negligible overreaching effect on -- for your modeling for next year. And when you say snap back, again, a lot of the projects are active. But again, there are people that we were working with. Not the majority, but definitely individuals who were furloughed and put off, which, I'd say, may have contributed to some of the slowdown. But again, I'm not -- so I would say no, that should not impact in my mind, what you are -- in terms of your modeling or our excitement for overreaching public sector. But again, like many others, we did see some hit over the last several weeks.

Thanks for that color. And then as I think also kind of a model question. I don't know if you want to handle it, Mike. But when I think about next year. I mean, historically, the front half of the year for software and services was kind of a bit lighter in a stronger second half. If I -- if I have that right. And so that might be being overwhelmed by the VDR growth. How should we just think about seasonality next year? I know you're not necessarily guiding for next year, but just kind of the trend, the timing?

[indiscernible] want to give any guide -- yes, I'll take it. We haven't given any guide for next year. But to your point, a lot of the growth was in the back half of this year, and that should continue in the first half relative to year-over-year comparisons. If that's to kind of give you some direction.

[Operator Instructions] Your next question comes from Stephen Banta with Banta Asset Management.

Great quarter, and it looks like you guys are executing well against what you stated back in '23. I'm just curious if you can provide maybe a little bit of color around your strategy with Veritone Hire. It seems like the business in general is firing on all cylinders and is looking to be optimistic in the future. But when it comes to Hire, how are you guys looking at that? Do you have a strategy with regard to that business?

It is a very stable business. Obviously, as we've stated in the market -- on the calls relative to the other, I'd say, more hyper growth areas of the business. It's more or less flat or slightly down. But it is a very stable business. It is a meaningful contributor to the business in terms of cash flow. And so we'll keep all of our options on the table. Again, I think most importantly, on a relative basis to its peers in the marketplace in an industry where a lot of players in the space have been down anywhere from 10% to 20%, for business that's been flat, slightly down, we're outperforming. So as of all things, it is an important part of the overall Veritone portfolio and revenue base.

But like all things, we're going to keep an open mind about the future. But right now, we're pleased with the overreaching. I think the most important thing we want investors to look at is overreaching the improvement across the board at an aggregate basis of what we've been able to do both top and bottom line. And as of today, the Veritone Hire piece is an important part of that equation.

This concludes our question-and-answer session. I would like to turn the conference back over to CEO and Chairman, Ryan Steelberg for any closing remarks.

Well, thank you, everybody, for the call today. Obviously, we're looking where -- the company, we sort of fit in -- we are an AI company, but ironically, relative to what I think we've done is massive improvements on some of the legacy overhangs. We have been kind of operating, I feel, historically, with 1, almost 2 arms tied behind my back. And I think us as a collective company with our product portfolio being in this market, both -- I'll call the data and AI economy, we should all be very, very optimistic, excited and bullish about our prospects. Again, hopefully, we can continue to improve our multiple as a company relative to our peer group. But there's a lot of things that we needed historically to clean up. And as I stated clearly, I believe that we have cleared those up. And I think we're in a phenomenal situation.

Also I want to double tap before I sign off on that very unique situation we are, where we sit in the middle of providing great utility value both from data suppliers and for data acquirers, very unique. And I want -- and we're going to continue to talk about that and continue to press on that. It just provides tremendous growth and revenue diversity for the business. So thanks, everybody, for their time, and we'll speak to you soon.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.

Good day, and welcome to the Veritone Inc. Second Quarter 2025 Financial Results Conference Call. [Operator Instructions] After today's presentation, there will be an opportunity to ask questions. [Operator Instructions] Please note this event is being recorded. I would now like to turn the conference over to Kate Goldsmith. Please go ahead.

Thank you, and good afternoon. After the market closed today, Veritone issued a press release and presenting results for the Second Quarter 2025, which ended June 30, 2025.

The press release and other supplemental information are available on the Investor Relations section of Veritone's website. Joining us for today's call are Veritone's President and Chief Executive Officer; Ryan Steelberg; and Chief Financial Officer, Mike Zemetra, who will provide prepared remarks and then open the call up for a live question-and-answer session.

Please note that certain information discussed on the call today, including certain answers to your questions will include forward-looking statements. This includes, without limitation, statements about our business strategy and future financial and operating performance. These forward-looking statements are subject to risks, uncertainties and assumptions that may cause the actual results to differ materially from those stated.

Certain of these risks and assumptions are discussed in Veritone's SEC filings, including its annual report on Form 10-K. These forward-looking statements are based on assumptions as of today, August 7, 2025, and Veritone undertakes no obligation to revise or update them.

During this call, the actual and forecasted financial measures we will be discussing include non-GAAP measures. Reconciliations of these measures to the corresponding GAAP measures are included in the press release we issued today.

Finally, I would like to remind everyone that the call today is being recorded and will be made available for replay via a link on the Investor Relations section of Veritone's website at www.veritone.com.

Now I would like to turn the call over to our President and Chief Executive Officer, Ryan Steelberg.

Thank you, Kate, and thank you, everyone, for joining us this afternoon. I'm excited to speak with you about our recent quarter and our overall progress we've made against our strategic business priorities. I will start with an update on our exciting results and progress against our growth plans and then Mike will cover our financials in more detail.

We are thrilled to report that our revenue of over $24 million for the quarter came in at the high end of our updated guidance in June and is a testament to the demand for our aiWARE platform and our market-leading AI applications and solutions. Our results demonstrate strong organic non-Veritone hire software revenue growth of over 45% in the quarter, and we expect this growth rate to continue through the balance of the year.

Veritone is definitively growing again, and this growth is being led by our core AI software solutions, spanning both commercial and public sector business lines. Since cofounding Veritone in 2014, I can confidently say that there has never been a more exciting moment for our company than right now. We are translating demand into tangible and sustainable growth. And what makes this even more compelling is that our fastest growth areas including Veritone Data Refinery or VDR and public sector not only delivered strong results, but also represent our largest pipelines and most expansive addressable markets in our history.

We are building momentum in all the right areas and the opportunity ahead of us has never been greater. In the quarter, we secured 104 new software customers and grew our VDR pipeline by over 100% from Q1 and over 33% just since our late June business update.

The near-term VDR pipeline now surpasses $20 million as the demand for high-quality training data across both our commercial and public sector verticals remains very strong. We have also broken through with yet another major DoD agency with the signing of our sole source contract with the U.S. Air Force in June. This deal is already contributing revenue in 2025 and and we expect it to ramp significantly in 2026 and beyond.

Our public sector pipeline is now up to $189 million, up from $110 million at the end of the first quarter. Overall, it has been a fantastic momentum building quarter as we transition into the second half of the year. Our recently announced cost-saving initiatives, which are expected to generate $10 million in annualized savings together with our $10 million equity offering we completed in June, have strengthened our financial position and enhanced our ability to execute our strategy and focus on driving growth.

Mike will provide more detail on these efforts and their impact across our core commercial and public sector verticals shortly. Now I want to provide an updated perspective on the AI landscape and our market opportunity. The AI landscape is indeed evolving rapidly. While enterprise-wide generative tools like Copilots and Chatbots have scaled quickly, function-specific applications remain mostly in pilot mode.

This Gen AI paradox, as McKinsey describes it, underscores a gap between broad adoption and truly transformative use cases. AgenticAI aims to close this gap shifting from reactive LLM centric tools to proactive goal-driven agents capable of autonomous workflow execution. These agents combine planning, memory and reasoning capabilities, but also pose new challenges around governance, data fragmentation and effective monitoring and control.

Scaling them effectively requires an infrastructure designed for trust, interoperability and flexibility across different vendors. This has been and remains Veritone's clear opportunity. Our aiWARE platform provides a scalable, secure and model agnostic foundation for ingesting and operationalizing both structured and unstructured data across disparate enterprise systems securely and at scale.

This architecture aligns directly with the emerging AgenticAI mesh where multiple AI agents interact, collaborate and continuously learn while maintaining visibility and control. As organizations move towards more adaptive, workflow-driven AI Veritone offers the infrastructure and purpose-built aiWARE platform to power that evolution, unlocking operational agility, smarter operations and new revenue opportunities. AiWARE also uniquely positions Veritone to capture 1 of the most compelling opportunities in the evolving AI value chain, training data.

As reported just last week by -- the Wall Street Journal, Big Tech will spend over $400 billion this year on AI CapEx and OpEx in what has become a new tech arms race. As next-generation LLM and multimodality models become more sophisticated, demand for high-quality domain-specific training data sets has surged. Industry estimates suggest that approximately $3 billion will be spent this year alone by fewer than 50 companies including the major hyperscalers on acquiring and preparing training data and is expected to grow to over $17 billion by 2032.

Veritone is uniquely equipped to serve this market through our VDR solution, which transforms massive volumes of unstructured video, audio and text into centralized license ready data sets for internal use or external model training. In the first half of 2025 VDR greatly exceeded our expectations in both adoption and revenue contribution. Data customers across both our commercial and public sector verticals ranging from major media networks to public institutions are using VDR to extract new value from both current and legacy data archives, which often are underutilized, some dating back over a decade.

On the buy side, VDR is now directly supplying clean structured training data to some of the world's largest hyperscalers and AI model developers. In many ways, VDR represents the next-generation solution for unstructured training data, building on what data labeling companies like Scale AI recently acquired by Meta for over $30 billion and Shutterstock have done for the training data economy.

Veritone's differentiation lies in our ability to process complex and diverse media types and modalities like audio and video at tremendous scale. In the second quarter alone, Veritone aiWARE processed millions of hours of video and audio or in data science speak, over 5 trillion tokens. Through multimodal tokenization, Veritone efficiently transforms these millions of hours of unstructured video and audio assets into the foundation training blocks of intelligent systems. Our aiWARE powered pipeline enables transformer-based sequence modeling and other training methods using discrete audio and video tokens, bringing enterprise-grade media into the era of AI.

Whether it's leveraged for third-party model training, fine-tuning enterprise customers' models or even for training Veritone's proprietary internal models, we are providing a modern day Agentic stack for customers around their audio, video and text data.

As I mentioned earlier, our qualified VDR pipeline now exceeds $20 million, up from $15 million at the end of June and more than doubling since early May. This growth reflects both expansion with existing customers as well as new agreements with leading hyperscalers and foundational model developers.

We expect to formalize partnerships with nearly all of the major hyperscalers by the end of 2025. The growth in our commercial business continues to accelerate, driven by our growing demand for our differentiated AI-powered solar software and managed services. We enable IP owners to unlock the full value of all of their media libraries by making them searchable, discoverable and monetizable across a range of channels including advertising, broadcasting, documentary production, TV and film projects and internal initiatives.

In the second quarter, Veritone Commercial successfully closed 11 software enterprise deals with clients such as Inner Milan, LaborCup, United States Soccer Federation, Alpha Media, St. Louis Zoo, ESPN and the Big Ten Network. These agreements underscore the continued expansion of Veritone's Software-as-a-Service offerings and highlight the critical role of our AI solutions and AI differentiated managed services in supporting our customers.

Turning to the public sector. The major highlight this quarter is our multiyear agreement with the U.S. Air Force to deploy our AI platform and intelligent digital evidence management system, or ITEMS. We will provide the Air Force with advanced investigative and information capabilities to enhance and accelerate data analysis for investigative activity across diverse mission areas.

This contract represents a material portion of our sales pipeline and represents a strong alignment between our capabilities and the mission-critical needs of our federal partners. We have already begun recognizing revenue from this contract in 2025 with revenue contributions expected to accelerate meaningfully in 2026. This latest sole source award with the Air Force represents our third contract with this agency and greatly expands the scope of our partnership.

Our work with the DoD's Defense Logistics Agency also continues to expand. Our task orders with the DLA funded under the Jet 2.0 IDIQ specifically require contractors to be confident in the use of Veritone applications, which has resulted in a number of contracted service providers entering into reseller agreements with Veritone IDMs for new opportunities.

In conjunction with our recently achieved awardable status through the Department of Defense's P1 solutions marketplace, we are realizing accelerated opportunities to expand our work with the DoD and other areas of federal, state and local government and agencies. In the second quarter, we signed 35 new public sector customers including the Riverside County Sheriffs Department and a top 5 police agency in the United States.

Additionally, we signed 95 renewal contracts in the quarter, further validating the mission-critical nature of our aiWARE software and strong customer retention. We are confident that Veritone is well positioned to take advantage of the surge in AI spend by our government as use cases and demand for our solutions and AI continue to grow.

Our direct public sector pipeline has grown to nearly $200 million, with defense technology spending projected to approach $1 trillion and the administration's prioritization of AI innovation, highlighted by the White House's recent AI action plan, we see significant additional business opportunities in the public sector, both this year and beyond.

Turning to our higher division. The second quarter delivered solid performance across multiple key areas. The strategy is implemented to navigate the challenging hiring market yielded positive outcomes, resulting in year-over-year growth for our SaaS and media services formerly known as Broadbean and exceeding our Q2 revenue targets.

We also surpassed annual sales targets and achieved record growth in Media service revenue in our inaugural year as a LinkedIn gold partner. With our programmatic business slated to conclude its LinkedIn pay-for-performance and Applied Connect integration in early Q3, we expect to benefit from these efforts across our entire portfolio. These investments with LinkedIn will enable us to continue delivering substantial value to clients and strengthen our collaboration with LinkedIn, the new global market leader.

We signed 58 new software deals in the quarter including some of our most significant deals to date while also building a more robust pipeline that could lead to even more positive results in the second half of the year, particularly in Q4, which traditionally is our strongest quarter for media deals in the higher division.

Another major hire initiative focused on expanding our SaaS revenue through enhanced ATS partnerships and integrations is also gaining momentum and shows considerable promise for the SaaS revenue growth. At the close of Q1, we executed our most critical partnership agreement to date with Workday. This elevated us to the highest platinum level of partners and created substantial opportunities for co-selling and lead generation with a global leader in the market.

In our first active quarter, our lead pipeline with Workday clients exceeds $1 million in contract value despite the nascent stage of lead and deal flow. We finalized 11 new Workday client deals this quarter. Furthermore, new integrations with major global ATSs and as well as our partnership with the integrations marketplace combo will provide Veritone access to over 100 new ATS integrations in Q3 and beyond.

We successfully concluded several notable deals this quarter with global corporations such as KPMG, Dauer, Faten, CBRE and Suncorp, among others. Before turning things over to Mike, I want to congratulate our team for their strong performance and perseverance. Veritone is growing again, and we remain very bullish on our future. Our pipeline is the largest it has ever been, led by public sector and VDR, and our AI software revenue growth is accelerating.

Now Mike, over to you.

Thank you, Ryan. We continued our strong momentum in the first half of 2025 with solid financial results in Q2. Revenue came in at the top end of our recent guidance with our software products and services, excluding Veritone hire, growing over 45% year-over-year, driven by strong performances across our public sector and commercial enterprise. We ended Q2 with solid customer metrics and contributions made across our software products and services and managed services.

As we enter the second half of 2025 we remain very confident on the future growth prospects across our core software products and services, which I will explain in more detail. During my prepared remarks, I will discuss Q2 year-over-year performance in KPIs which exclude the results of our media agency, which are presented as discontinued operations in the corresponding historical financial periods, balance sheet and liquidity position and Q3 and fiscal 2025 guidance.

Starting with Q2 2025 performance. Q2 revenue was slightly over $24 million, which was flat from Q2 2024 principally due to a $1 million increase from our software products and services, offset by a $1.9 million decline in our managed services. The $1.8 million revenue growth in our software products and services was driven by our public sector, which grew over 90% year-over-year, coupled with the commercial enterprise software products and services revenue that improved $0.8 million year-over-year. The growth in the public sector was driven by execution of larger deals in Q2 2025, including the Department of Defense and larger public safety agencies, including a top 5 law enforcement agency in the U.S. and Riverside County.

We expect these larger public sector deals coupled with our expanding public sector pipeline to generate substantial growth in the second half of 2025, which I will explain in more detail. The growth in commercial enterprise was led by Veritone Data Refinery or VDR. VDR, which launched in Q4 2024 and is 1 area where we anticipate substantial year-over-year growth throughout the remainder of fiscal 2025, and today has a near-term sales pipeline over $20 million up over 100% from our guidance in Q1 2025.

The $1.9 million decline in Q2 Managed Services was principally driven by a $2 million decline in representation services driven by declines in our very ad services and a onetime live event campaign of $1 million in Q2 2024, which did not recur in Q2 2025 offset by a $0.1 million improvement in licensing.

As we previously discussed, we expect this negative trend in representation services to continue throughout 2025, and or until the macro economy shows demonstrated improvements over 2024. Overall, Veritone hire remained relatively flat year-over-year, driven largely by the higher softness in the macro economy. Excluding Veritone Hire, our software products and services grew over 45% year-over-year.

Turning to key performance metrics across our software products and services in Q2 2025. ARR of $62.6 million, up 7% from Q1 2025 of $58.7 million and down year-over-year from the expected declines in consumption-based revenue from customers across our hiring software products and services over the trailing 12 months.

Overall, ARR from recurring subscription-based SaaS customers was up slightly by 2% year-over-year. As of Q2 2025, 81% of our ARR was from subscription versus consumption-based customers, up from 74% at Q2 2024 and flat sequentially from Q1 2025. Total new bookings of $15.8 million, up $1.8 million or 13% year-over-year, primarily due to larger renewals across our software customer base.

Gross revenue retention continued to be above the 98th percentile and total software product and service customers of 3067, which was down 9% year-over-year, predominantly from our commercial enterprise sector. which includes lower consumption-based customers from Veritone Hire and the continuing impact of sunsetting legacy CareerBuilder customers post the June 2023 acquisition of Broadbean and a smaller customers as we focus on larger ARR opportunities, offset by an increase across public sector, largely from growth in public safety customers.

Q2 GAAP gross profit was $15.3 million, compared to $16.4 million in Q2 2024, a decline of $1.1 million, largely driven by the higher mix of lower margin revenue in Q2 2025 with GAAP gross margins of 63.9% as compared to 68.2% in Q2 2024. Excluding noncash depreciation and amortization expense, 2025 non-GAAP gross margins were 68.9% as compared to 73.6% in Q2 2024, a decline of 470 basis points, largely due to the decline in higher-margin consumption-based revenue, coupled with a higher mix of lower margin revenue.

Note that in Q2 2025, VDR gross margins were approximately 40%. We expect that as the VDR product matures, margins will initially be similar to Q2, but should expand into late 2025 and 2026 as we grow and diversify the mix of our content offerings. Q2 operating loss of $19.3 million improved by $1 million or 5% year-over-year, primarily driven by lower operating expenses, offset by a lower non-GAAP gross profit from the decline in revenue over the same period. Net loss from continuing operations was $26.8 million, an increase of $3.4 million or 14.5% as compared to Q2 2024. The year-over-year increase was principally driven by a $3.4 million change in the estimated fair value of earnout from the divestiture of Veritone One recorded in Q2 2025.

Non-GAAP net loss from continuing operations was $8.7 million as compared to $9.7 million in Q2 2024 and $11.1 million in Q1 2025. The improvement was partially due to lower operating losses driven by increased discipline on cost management offset by lower non-GAAP gross profit.

Further, in June 2025, we initiated up to $8 million of a targeted $10 million annualized cost reduction through reductions in personnel and improvements in our operating structure, including our platform costs. These cost reductions were initiated in part due to the softness in our managed services coupled with delays in certain public sector deals that were expected to close earlier in 2025.

As I will explain further, these reductions should provide us a more efficient cost structure as we manage towards our planned growth in the second half of 2025 and profitability into 2026 and beyond. Turning to our balance sheet. As of June 30, 2025, we held cash and restricted cash of $13.9 million as compared to $17.3 million at December 31, 2024. The net change in cash reflects net cash outflows from operations of $25.2 million, principally driven by our non-GAAP net loss of $19.8 million deferred purchase consideration of $1.2 million and interest paid on debt of approximately $3 million, coupled with the timing of working capital in the quarter, offset by net cash inflows from investing and financing activities of $23 million driven by net cash inflows of $29.9 million from our January and June 2025, registered direct offerings, partially offset by $3.9 million in debt principal payments and $2.3 million in capital expenditures.

Turning to liquidity today. On June 30, 2025, we completed a registered direct offering, selling 6.5 million shares of common stock priced at $1.09 per share and $1.8 million of prefunded warrants priced at $1.08 per share for gross proceeds of approximately $10 million. Of the total funding, approximately $3 million of the gross proceeds was received in July 2025. Included in the funding was $1 million from our CEO, Ryan Steelberg, which will price at the greater of $1.41 per share, which was the closing price of Veritone's stock on June 27, 2025, or the closing price of Veritone stock 2 trading days following the filing of our Q2 Form 10-Q.

At June 30, 2025, our consolidated debt is down from a peak of $201 million in December 2021 to approximately $128 million today, comprised of term debt of approximately $37 million maturing in December 2027 and convertible debt of $91.3 million due November 2026. As of today, we have over $25 million available across our $35 million ATM, which was established in November 2024. That said, we are currently exploring potential financing structures, including discussions with our current debt holders, which we believe could improve our current liquidity position and balance sheet.

At June 30, 2025, we had 47.6 million shares issued and outstanding and 2.5 million warrants outstanding to our debt holders. Now turning to updated fiscal Q2 2025 and full year 2025 guidance. Our software products and services revenue pipeline and long-term outlook continue to be at all-time highs. More specifically, we continue to see strong demand across the approximate $10 billion global digital evidence management market.

In the public sector alone, we are beginning to march towards our 100% to 150% revenue growth target for fiscal year 2025. In Q2, we announced we were awarded a sole source contract with the Air Force Office of Special Investigations, or OSI. Under the contract, the company's AI power solutions, including items, will provide OSI with advanced investigative, intelligence and counterintelligence capabilities in support of the DoD and interagency mission requirements.

During Q2 2025, we began to recognize revenue on the award. This is the initial deployment with plans to raw our IDM solution across the broader DoD investigative and counterintelligence branches, over the next several years. While we cannot discuss the magnitude or exact specifics of this deal, it will serve as a substantial growth driver of our public sector revenue in 2025 and '26.

We also remain in near-term contract basis on several large projects with various facets of the U.S. federal government and international public safety customers with a near-term sales pipeline in excess of $180 million. As previously noted, on the commercial side, we are seeing strong and increasing demand for our BDR product. More specifically, we are in active discussions with the largest hyperscalers on various BDR initiatives some of which are near-term agreements that approach or exceed $10 million individually and others are longer-term partnerships where we are being positioned to serve as their provider of choice across their VDR initiatives.

Our near-term sales pipeline on VDR, which launched in the second half of 2024 is now over $20 million, which is an increase of 100% or $10 million since March 2025 and $5 million since the end of June 2025. More specifically, in Q3 2025, revenue is expected to be between $28 million and $30 million as compared to $22 million from Q3 2024, a 32% increase at the midpoint and 21% sequentially from Q2 2025.

In Q3, we expect our software products and services to increase over 45% year-over-year, led by growth in the public sector and commercial enterprise. Specifically, we expect our public sector revenue to grow from 50% year-over-year and our commercial revenue led by VDR to grow or 45%. Included in this growth is our hiring products and services, which we expect to be relatively flat year-over-year given the current macroeconomic environment.

Consistent with Q2 2025, our managed services is expected to be down year-over-year, principally due to the representation side of our business, which has experienced some slowness as a result of the more challenging macro environment. We expect Q3 non-GAAP gross margins to be around 61% to 63%, driven by the forecasted higher mix of EDR revenue in the period.

Q3 non-GAAP net loss is projected to be between $6 million to $6.5 million as compared to $11.1 million in Q3 2024, representing a 43% improvement at the midpoint and a 28% improvement sequentially from Q2 2025.

Turning to fiscal 2025 outlook. We are updating our prior guidance for fiscal 2025, which we are expecting revenue to be between $108 million to $115 million which at the midpoint represents a 20% increase year-over-year. The change in our outlook is principally driven by the confidence in some of our more larger growth initiatives across the public sector and commercial VDR coupled with a forecasted decline in managed services, reflecting the more challenging macro market today.

And non-GAAP net loss to be between $30 million to $25 million, representing a 33% improvement year-over-year at the midpoint. The change is reflective of the timing shifts in revenue, coupled with the compression in gross margins on VDR in 2025, which we expect to improve upon fiscal 2026.

Before closing the call, I'd like to remind everyone listening that Veritone will be attending H.C. Wainwright's 27th Annual Global Conference, September 8 through the 10 in New York City. That concludes my prepared remarks. Operator, we would now like to open up the call for questions.

[Operator Instructions] And your first question comes from Scott Buck with H.C. Wainwright.

And your next question comes from Jesse Sobelson with Boral Capital.

I think the big thing, this is a great quarter doing exactly what you guys said you were going to do you got this guidance for the rest of the year that points to some pretty significant acceleration in the growth rate. I mean impacted a midpoint near 30% year-over-year here. while you've provided a lot of details on growing the business to signing the contracts with -- it was in the public sector, the growth in the pipeline.

Can you guys elaborate on what specifically needs to convert whether it's revenue recognition from this Air Force contract or any VDR pipeline signings that need to happen to support this step-up in the acceleration that we're seeing guided for the top line? And I'm just kind of curious what gives you the level of confidence into the visibility you have for that acceleration?

Thank you, Jesse. I think as we sit here right now, we probably have our smallest gap of go get revenue to realize that guide for Q3. And I say a bunch of that already in terms of bookings and visibility for the balance of the year. So again, as we sit here today in the end of the first week in August, our delta of go get to hit that guide midpoint and hopefully, the high end is the smallest it's ever been. .

The other way of saying that is sort of the contracted opportunities and the businesses and opportunities that we're currently servicing will support the balance of that opportunity for revenue. So again, we're -- we have the customers. We are transacting as we -- as Mike and I both mentioned, we're already generating revenue from these new DoD contracts, we're already generating revenue from these new customers with VDR. So we're very, very excited. We've been sort of waiting for this to happen, and they're all kind of hitting on all cylinders right now.

Awesome. That's really great detail. It's really exciting to see we're excited here too. I guess, let's just back up and ask a broader level question here as well and then take a step back. But just with the -- within -- everything is moving very fast through very rapid advancements and even generative AI. How do you view Veritone's differentiation, particularly in these regulated industries like defense and law enforcement versus the broader AI platforms out there such as like Palantir, Microsoft or other open source options. .

I think first and foremost, and this goes back all the way to our founding. We kind of and thankfully had a vision that AI-based models were going to become commoditized over time. There will be thousands of vision models. There'd be thousands of speech models -- and even today, there's hundreds of next-generation LLM and multimodal models.

The key is having a platform, aiWARE, that, in effect, can manage the full end-to-end stack, but be agnostic to these models or independent of these different models, meaning customers like iHeartMedia and others, they've been customers for many years. We own, I'll say, the software platform application layer with these entities. And so as models advance, they don't have to go anywhere.

So they can rely on the full technology stack of Veritone and aiWARE as this -- as the models mature. The 1 thing, though, that was kind of a byproduct. And candidly, and VDR is kind of a culmination of the scale that we've been accumulating over the last decade. But as you've just read through and a good parallel is what Altman just described, when OpenAI opened a couple of their weighted models, as many of us have read about, what they did in open source was their training data, right, which is really interesting.

And so what that points to is when you think about Veritone and why we are such a strategic partner and able to not only maintain these long-lasting relationship with customers, but able to bring on net new big customers -- it is not just about the understanding the orchestration of the AI and the models, but it's also understanding the unstructured data at huge scale.

I gave a little ties in my speech, which we're going to follow up on and talk a lot more over the next few weeks. But our scale of tokenizing audio and video and other forms of unstructured data is at huge scale. I actually mentioned that we, in effect, tokenized or, in effect, realized over 5 trillion tokens of -- in terms of scale for audio and video processing just in Q2. If you based upon the research if you kind of look at what it took to build these other foundational models, that's massive scale.

And frankly, that was just in 1 quarter. So again, our differentiation is not just understanding and making a very early bet that there were going to be hundreds, thousands, millions of AI models. So we made a bet that it's going to be -- you need to have the orchestration of not just the models, but the true understanding of the end-to-end platform that can manage the unstructured and structured day at the same time.

The last thing I'll put a little bit on that is we're platform agnostic. We're not limited to a single cloud provider, our platform has now been successfully deployed in air gap network isolated environments. So not only can we run this readily available in public clouds in government clouds, but we can also run it in network isolated environments and even on-prem, which is a big differentiator.

And your next question comes from Glenn Mattson with Ladenburg.

First on IMS, -- in the past, you've spoken about multiple kind of -- I think you said 7- or 8-figure kind of multiyear sized deals of the the win in the Air Force is a significant one, clearly. I guess partly, I'm curious, do you need to close more of those style deals to meet your guidance for the year you reiterated, so I'm guessing you feel pretty comfortable there. But also just can you speak to like what winning that contract and what it means for these next contracts? Is it get easier to close them as you get more reference customers of that size any color that would be great.

I'll pick those off. First, of course, our ability to point to a very large multifaceted contract with the Air Force. -- spread quickly, and we can attest and what we kind of spoke to is that we are seeing a multitude of different opportunities and inbound demand spread quickly across the DoD and even outside the DoD because of the Air Force contract announcement.

And that's, I'd say, commensurate with any other type of business. So referral and having that public disclosure is very, very important. The second thing is relative to our pipeline, the OSI Air Force contract alone presents a great, huge opportunity for years to come. And remember, the Air Force and the group that we're working with, yes, it's today kicking off with OSI, but this is an entity that is really taking the charge in some areas, the lead on expanding this overhaul of the government's law enforcement and counterterrorism opportunities into other agencies as well.

So yes -- and this opportunity for us really is the tip of the iceberg for expansion just in the areas of use cases for investigations and counterterrorism, we expect this to expand into other agencies as well. So we do have a healthy pipeline that crosses over into many other areas of DoD. And -- but again, to sort of be specific to your question is, we have enough meat if we can just execute on, frankly, the kind of customers that we already have. I think we're pretty confident to hit many of our goals.

That being said, as we attested to a material increase in the size of our pipeline, -- we are bringing on more and more new customers, as we mentioned, 35 new customers in public sector in the previous quarter, and we expect that to continue to accelerate here over the next few quarters.

You also mentioned the top 5 public safety customer. I'm not -- I can't recall if you've mentioned that in the past or not, but can you give any more color around what the -- what kind of an agency and what the use cases are that company .

It's an IDM customer. It's 1 of the biggest law enforcement agencies in the world. We're not at liberty to give specific names. But let's just say, hopefully, we're closing some homicide and murder investigations, deleveraging our next-generation AI-based software here. But we're thrilled, and we're all excited about our DoD efforts, but our state and local enforcement business continues to grow and thrive as well. And the Sharif department, which we talked about and also this police agency is kind of a good representation of the demand for our solutions in SLED as well. .

One more if I could squeeze in on VDR. You talked -- I think the pipeline you think is $20 million now. Can you quantify just give some more color on how you come over that number and what that means is translate to revenue? I know in that pie chart on your slide deck, you upped the range, the higher end of the range kind of -- can you just kind of help us understand like the bridge between that and revenue and the timing .

Yes. So I think as Mike articulated in his remarks, that near-term pipeline is kind of qualified as $20 million with high visibility between the next 3 to 12 months. VDR could just continue to grow and grow and grow. I think we've really hit a perfect use case sweet spot. As I articulated in my remarks, I really do believe Veritone has created a more technology version for unstructured data that scale AI did and others in primarily legacy data labeling efforts to help train data.

We are working with many of the largest hyperscalers and model developer companies today. So we are representing them in providing and helping them train their next-generation AI models. And the budgets are massive. So again, we are being successful. I've -- it's probably -- I'm very excited about public sector. I'm very excited about DoD -- but VDR could be lightning in a bottle. Still a lot more work to do, but I think we've got great product market fit right now, and it's pedal down on sort of our bullishness for VDR.

And your next question comes from Seth Gilbert with UBS. .

The full year guide by $4 million and reduced the high end of the non-GAAP net loss by $5 million. So just curious if you could talk about where you're making investments? Or is there anything paid in HMOs. .

The first -- did you hear the first part of that, Mike, I didn't hear the first part of the question. But if you got it. .

I didn't hear the first part, but yes, I think I can articulated. Yes. Just in terms of the non-GAAP net loss, what I explained is just kind of the velocity of VDR and the compression on margins. And so while we raise the top end of the revenue guide, we tightened on the lower end or I guess, the higher end than the non-GAAP net loss as a result of that margin compression. .

Got it. Sorry, I was switching between a few calls. Maybe 1 more. If the public sector grows 50% in 3Q, which I believe I heard you say on the call, then in order to get to the 100% to 150% year-over-year guide for the year, you need to grow the public sector by about $3 million quarter-over-quarter. -- maybe almost 300%.

So I'm just wondering if that's fair to assume a big uptick in 4Q? And is it all from the DoD revenue kind of maybe hitting in 4Q?

Yes. On 4Q, I mean we're not giving specific guidance, but I think your math is probably directionally accurate. And it's not dependent on a single contract, but that contract obviously is a vehicle for a good portion of that growth.

And your next question comes from Scott Buck with H.C. Wainwright. .

The 3Q guide suggests a nice sequential step-up in revenue. I'm curious, what do you think you kind of have in hand versus what you have to go out and earn here over the next couple of months to to meet the midpoint of that guide? .

Scott, as I kind of mentioned earlier, you may have been off the call, but I said right now, we probably have the smallest delta of go get that we've really had in any quarter. We have a lot of the deal flow kind of in the works already either under contract and just delivering and executing against such as VDR or areas that, again, we had to get over the hump to get the Air Force contract done with initial phases deployed. But now that, that contract and award is live and our initial software has been deployed again, now it's just ramping and scaling. So again, to be very clear, I would say, our contracted opportunity provides us probably the best resolution for us to I'd say, perform against the guide with the smallest delta of GoGet that we've had in a very long time. .

That's great and very helpful. Did you disclose what revenue from the Air Force was during the quarter? .

No, we do not. .

Can you?.

No, we're not going to. We're not going to break down specific contracts or specific line items. .

Fair enough. On the VDR pipeline, are there significant customer concentrations in there? Or is it fairly evenly dispersed? .

I think the space in general, as I've said in our -- in my prepared remarks, the space is dominated by really the top 50 major players -- so it is a massive category, but it is dominated by 50 players around that area. So again, this is unlike other areas of our business where we have thousands of customers you can sort of interpret that -- again, to be clear, we don't have a single customer who's dominated it.

But in terms of the current ecosystem out there, this is both good and bad, it is -- the super majority of the spend is concentrated to about 50 major big tech companies today and AI model development companies. On the flip side, the positive side is I probably -- as we stated, we're engaged with almost all of them already. right? We haven't -- we're not actively doing business with all of them yet. I think we did say is our goal is to be actively engaged in working with all of the major hyperscalers by the end of the year.

So on 1 side, I would say there is the potential of concentration within that 50 group. But on the opposite side is it afford us the ability because of the killer offering that we have and the great assets that we have we're already engaged with near a majority of the 50 already.

Okay. And as you guys have gone out and sold VDR to your spoken to customers on VDR, are you getting better in that sales process? I mean I imagine each incremental opportunity, your sales team was able to learn a little something new and might be more efficient moving forward.

Yes. I'll say empirically, the answer is yes, based upon the data we provided where we continue to increase our pipeline, right? So -- and a lot of that pipeline increase is not just net new customers, but it's repeat business or expanded business with existing customers in VDR. So I would say that is a good testament to not just us getting better on the sales process, but people getting through trials, building confidence and trust in our VDR products and solutions to accelerate. So I'd say that's exciting.

On the flip side, this audio and video and unstructured data as a training data class is relatively newer. So again, part of it is just familiarity with a lot of different groups who are, let's say, first time really trying to think about what their next-generation multi-modality video-based models may look like or groups who are trying to fine-tune legacy multimodality models.

So not only, I would say it's 2 sides. Not only are we learning more about what these different groups are trying to achieve with their next-generation models. But on the flip side, they are learning more about us. And I can say confidently is we're building great trust. People know that Veritone can deliver at scale. And I did touch on kind of the scale and size of tokens that we are dealing with. And we're talking about hundreds of millions, billions and obviously, potentially trillions of tokens in the context of just the scale of the audio video that we're dealing with. So we're bullish, and I'd say it's really 2 sides.

One, we're getting more confident. And two, our customers who we're working with and selling to are getting more confident in -- with a higher level of trust in us.

This concludes our question-and-answer session. I would like to turn the conference back over to President and Chief Executive Officer, Ryan Steelberg, for any closing remarks. .

Well, first, I'd like to thank everybody. I'm obviously very excited about this quarter. It's taken us a long time to clean up a lot of things. But as sort of we've been sort of indicating now for a few quarters, there's really a lot of excitement on some killer core assets in our AI software business in VDR that we knew were coming, and this quarter kind of culminated in a couple of things. One, we got over the hump with the big OSI Air Force contract and really started to sort of jog into a run with the growth of VDR, which is going to be a great accelerant for the balance of the year and into the future. 45% year-over-year growth in the second quarter in our core AI software is phenomenal.

And I think there's a lot -- there's a high ceiling there. So again, we're exiting this quarter and moving into the second half of the year with a purview into next year, feeling great. We still got to execute, but our pipeline is large. We're dealing with some of the largest tech companies out there who were selling with and sort of finding the sky limit for Veritone.

Mike covered it, but please check our IR sites or please register. We are going to be attending a multitude of different financial conferences here over the balance of the year and into the first quarter of next year. Hope to meet and talk to both our existing and new investors. But I appreciate everybody's time today. Thank you.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.