NVIDIA Corp (NVDA) — Q4 2026 Earnings Call Transcript

Thank you. Good afternoon, everyone, and welcome to NVIDIA's conference call for the fourth quarter of fiscal 2026. With me today from NVIDIA are Jensen Huang, President and Chief Executive Officer, and Colette Kress, Executive Vice President and Chief Financial Officer. Our call is being webcast live on NVIDIA's Investor Relations website. The webcast will be available for replay until the conference call to discuss our financial results for the first quarter of fiscal 2027. The content of today's call is NVIDIA's property. It can't be reproduced or transcribed without prior written consent. During this call, we may make forward-looking statements based on current expectations. These are subject to a number of significant risks and uncertainties, and our actual results may differ materially. For a discussion of factors that could affect our future financial results and business, please refer to the disclosure in today's earnings release, our most recent Forms 10-K and 10-Q, and the reports that we may file on Form 8-K with the Securities and Exchange Commission. All our statements are made as of today, February 25, 2026, based on information currently available to us. Except as required by law, we assume no obligation to update any such statements. During this call, we will discuss non-GAAP financial measures. You can find a reconciliation of these non-GAAP financial measures to GAAP financial measures in our CFO commentary, which is posted on our website. With that, let me turn the call over to Colette.

Thanks, Toshiya. We delivered another outstanding quarter with record revenue, operating income and free cash flow. Total revenue of $68 billion was up 73% year-over-year, accelerating from Q3. Growth on a sequential basis was also a record as we added $11 billion in data center revenue across a diverse and expanding set of customers, including cloud providers, hyperscalers, AI model makers, enterprises and sovereign nations. Demand for our Blackwell architecture, extreme co-design at data center scale continues to strengthen as inference deployments grow in addition to training. The transition to accelerated computing and the infusion of AI across existing hyperscale workloads continue to fuel our growth. Agentic and physical AI applications built on increasingly smarter and multimodal models are beginning to drive our financial performance. On a full year basis, data center generated revenue of $194 billion, up 68% year-over-year. We have now scaled our data center business by nearly 13x since the emergence of ChatGPT in fiscal 2023. We look ahead, we expect sequential revenue growth throughout calendar 2026, exceeding what was included in the $500 billion Blackwell and Rubin revenue opportunity we shared last year. We believe we have inventory and supply commitments in place to address future demand, including shipments extending into calendar 2027. Every data center is power-constrained. Customers make critical architectural decisions based on performance per watt given these constraints and the need to maximize AI factory revenue. SemiAnalysis declared NVIDIA, Inference King, as recent results from InferenceX reinforced our inference leadership with GB300 NVL72, achieving up to 50x performance per watt and 35x lower cost per token compared with Hopper, and continuous optimization of CUDA software helped deliver up to 5x better performance on GB200 NVL72 just within 4 months. NVIDIA produces the lowest cost per token and data centers running on NVIDIA generate the highest revenues. Our pace of innovation, particularly at our scale is unmatched, fueled by an annual R&D budget approaching $20 billion and our ability to extreme co-design across compute and networking across chips, systems, algorithms and software. We intend to deliver X factor leaps and performance per watt average generation and extend our leadership position over the long term. Q4 data center revenue of $62 billion increased 75% year-over-year and 22% sequentially, driven primarily by sustained strength in Blackwell and the Blackwell Ultra ramp. With NVIDIA infrastructure in high demand, even Hopper and much of the 6-year-old Ampere based products are sold out in the cloud. Nearly a year has passed since the release of our Grace Blackwell NVL72 systems. Today, nearly 9 gigawatts of infrastructure on Blackwell are deployed and consumed by the major cloud service providers, hyperscalers, AI model makers and enterprises. Networking, a cornerstone of our data center scale infrastructure offering, was a standout this quarter, generating $11 billion in revenue, up more than 3.5x year-over-year. Demand for our scale-up and scale-out technologies reached record levels, both growing double digits sequentially, driven by strong adoption of NVLink, Spectrum-X Ethernet and InfiniBand. On a year-over-year basis, growth was driven primarily by NVLink 72 scale-up switches as Grace Blackwell systems accounted for roughly 2/3 of data center revenue in the quarter. NVLink scale-up fabric has revolutionized computing and demonstrates the power of extreme co-design across all of the chips of the supercomputer and the full stack. In Q4, we announced that we will enable AWS with NVLink to integrate with their custom silicon. Momentum is strong with our Spectrum-X Ethernet scale up and scale across networking as customers work to unify distributed data centers into integrated gigascale AI factories. For the full year, our networking business exceeded $31 billion in revenue, up more than 10x compared to fiscal 2021, the year we acquired Mellanox. Our demand profile is broad, diverse and expanding beyond just chatbots. First, there is a fundamental platform shift from classical machine learning to generative AI. Strong evidence of ROI as hyperscalers upgrade massive traditional workloads to generative AI, including search, ad generation and content recommender systems is encouraging our largest customers to accelerate their capital spending. For example, at Meta, advancements in their GEM model drove a 3.5% increase in ad clicks on Facebook and more than 1% gain in conversations on Instagram, translating into meaningful revenue growth. With the same NVIDIA infrastructure, Meta Superintelligence Labs can train and deploy their frontier agentic AI systems. Frontier agentic systems have reached an inflection point. Claude Code, Claude Cowork and OpenAI codecs have achieved useful intelligence. Adoption is skyrocketing and tokens are profitable, driving extreme urgency to scale up compute. Analyst expectations for 2026 CapEx across the top 5 cloud providers and hyperscalers who collectively account for a little over 50% of our data center revenue are up nearly $120 billion since the start of the year and approaching $700 billion. We continue to expect the transition of classic data center workloads to GPU accelerated computing and the use of AI to enhance today's hyperscale workloads and contribute toward roughly half of our long-term opportunity. Every country will build and operate some parts of its AI infrastructure, just like with electricity and Internet today. In fiscal year 2026, our sovereign AI business more than tripled year-over-year and over $30 billion, driven primarily by customers based in Canada, France, the Netherlands, Singapore and the U.K. Over the long run, we expect our sovereign opportunity to grow at least in line with the AI infrastructure market as countries spend on AI proportional to their GDP. While small amounts of H200 products for China-based customers were approved by the U.S. government, we have yet to generate any revenue. And we do not know whether any imports will be allowed into China. Our competitors in China bolstered by recent IPOs are making progress and have the potential to disrupt the structure of the global AI industry over the long term. To sustain its leadership position in AI compute, America must engage every developer and be the platform for choice for every commercial business, including those in China. We will continue to engage with the U.S. and China government and advocate for America's ability to compete around the world. We unveiled the Rubin platform last month at CES comprised of 6 new chips, the Vera CPU, Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPUs and Spectrum-6 Ethernet switch. The platform will train MOE models with 1/4 number of GPUs, reducing inference token costs by up to 10x compared to Blackwell. We shipped our first Vera Rubin samples to customers earlier this week, and we remain on track to commence production shipments in the second half of the year. Based on its modular cable-free tray design, Rubin will deliver improved resiliency and serviceability relative to Blackwell. We expect every cloud model builder to deploy Vera Rubin. Moving to gaming. Gaming revenue of $3.7 billion increased 47% year-on-year, driven by strong Blackwell demand and improved supply. GeForce RTX is the leading platform for PC gamers, creators and developers. In Q4, we added several new technologies and advancements, including DLSS 4.5, which uses AI to bring game visuals to a new level. G-SYNC Pulsar, bringing incredibly clear graphics even in motion, and 35% faster LLM inference across leading AIPC frameworks. Looking ahead, while end demand for our products remains strong and channel inventory levels are healthy, we expect supply constraints to be the headwind to Gaming in Q1 and beyond. For Professional Visualization, it crossed the $1 billion mark for the first time, with revenue of $1.3 billion, up 159% year-over-year and 74% sequentially. During the quarter, we launched the RTX PRO 5000 Blackwell workstation with 72 gigabytes of fast memory for AI developers running LMs and agentic workflows. Automotive revenue of $604 million was up 6% year-over-year and was driven by robust demand for self-driving solutions. At CES, we introduced Alpamayo, the world's first open portfolio of reasoning vision-language-action models, simulation blueprints and data sets, enabling vehicles that can think. The first passenger car featuring Alpamayo built on NVIDIA DRIVE will be on the road soon in the new Mercedes-Benz CLA. Physical AI is here having already contributed over $6 billion in NVIDIA revenue in fiscal year 2026. Robotaxi rides are growing exponentially with commercial fleets from Waymo, Tesla, Uber, WeRide and Zoox, and many others are expected to scale from thousands of vehicles in 2025 to millions over the next decade, creating a market poised to generate hundreds of billions of dollars in revenue. This expansion will demand orders of magnitude more compute with every major OEM and service provider developing on NVIDIA's platform. We continue to advance robotics development. With the new NVIDIA Cosmos and Isaac Group, open models, frameworks and NVIDIA's powered robots and autonomous machines for leading companies, including Boston Dynamics, Caterpillar, FranKaufman Hall Robotics, LG Electronics and NEURA Robotics. To accelerate industrial physical AI adoption, we also announced new expanding partnerships with Dassault Systemes, Siemens and Synopsys to bring NVIDIA AI infrastructure Omniverse digital twins, World Models and CUDA-X libraries to millions of researchers, designers and engineers building the world's industries. Let's move to the rest of the P&L. GAAP gross margin was 75% and non-GAAP gross margin was 75.2%, increasing sequentially as Blackwell continued to ramp. GAAP operating expenses were up 16% sequentially and up 21% on a non-GAAP basis related to new product introductions and compute and infrastructure costs. Non-GAAP effective tax rate for the fourth quarter was 15.4%, below our outlook for the quarter, primarily due to the impact of a one-time tax benefit. Inventory grew 8% quarter-over-quarter, while purchase commitments also increased significantly, and we have strategically secured inventory and capacity to meet demand beyond the next several quarters. This is further out in time than usual and reflects the longer demand visibility we have. While we expect tightness in the supply for our advanced architectures to persist, we remain confident in our ability to capitalize on the growth opportunity ahead with our scale, expansive supply chain and the long-standing partnerships continuing to serve us well. We generated free cash flow of $35 billion in Q4 and $97 billion in fiscal year 2026. For the year, we returned $41 billion or 43% of free cash flow to our shareholders in the form of share repurchases and dividends. We continue to invest in technology and our ecosystem to cultivate market development, drive long-term growth and ultimately yield total shareholder returns superior to the market or our peer group. Importantly, we will continue to run a strategic and disciplined process as it relates to our investments and we remain committed to returning capital to our shareholders. Let me turn to the outlook for the first quarter. Starting this quarter, we will be including stock-based compensation expense in our non-GAAP results. Stock-based compensation is a foundational component of our compensation program to attract and retain world-class talent. Let me first start with revenue. So revenue is expected to be $78 billion, plus or minus 2%. We expect most of our growth to be driven by data center. Consistent with last quarter, we are not assuming any data center compute revenue from China in our outlook. GAAP and non-GAAP gross margins are expected to be 74.9% and 75%, respectively, plus or minus 50 basis points. For the full year, we continue to see gross margins in the mid-70s. We will keep you updated on our progress as we prepare for the Vera Rubin transition. GAAP and non-GAAP operating expenses are expected to be approximately $7.7 billion and $7.5 billion, respectively, including stock-based compensation expense of $1.9 billion. For the full year, we expect non-GAAP operating expenses to grow in the low 40s on a year-over-year basis as we continue to invest in our expanding opportunity set. For the full year fiscal year '27, we expect GAAP and non-GAAP tax rates to be between 7% and 19%, excluding any discrete items and material changes to our tax environment.

This quarter, we significantly deepened and expanded our partnerships with leading frontier model makers. We recently celebrated OpenAI's launch of GPT-5.3-Codex trained with and inferencing on Grace Blackwell NVLink 72 systems. GPT-5.3-Codex can take on long-running tasks that involve research, tool use and complex execution. 5.3-Codex is deployed broadly inside NVIDIA. Our engineers love it. We continue to work with OpenAI toward a partnership agreement and believe we are close. We are thrilled with our ongoing partnership with OpenAI, a once-in-a-generation company we've had the pleasure of partnering with since their first days. Meta Superintelligence Labs is scaling up at lightning speed. Last week, we announced that Meta is deploying millions of Blackwells and Rubin GPUs, NVIDIA CPUs and Spectrum x Ethernet for training and inference. This quarter, we announced a partnership with Anthropic, and a $10 billion investment in their company. Anthropic will train an inference on Grace Blackwell and Vera Rubin system. Anthropic's Claude Cowork agent platform is revolutionary and has opened up floodgates for enterprise AI adoption. Between Claude Cowork and OpenAI Codex, Anthropic's Claude Cowork agent platform compute demand is skyrocketing and the moment of agentic AI has arrived. With partnerships spanning Anthropic, Meta, OpenAI and xAI, NVIDIA is deployed across every cloud and with our ability to build full-stack AI infrastructure from the ground up, we are uniquely positioned to partner with frontier model builders at every stage, training, inference and AI factory scale-out. Finally, we recently entered into a nonexclusive licensing agreement with Grok for its low-latency inference technology and welcome the team of brilliant engineers to NVIDIA. As we did with Mellanox, we will extend NVIDIA's architecture with Grok's innovations to enable new levels of AI infrastructure performance and value. We look forward to sharing more at GTC next month. Okay, back to you.

We will now transition to Q&A. Operator, please poll for questions.

I think you mentioned that you now have growth visibility into calendar '27 also, and I think your purchase commitments kind of reflect that confidence. But Jensen, I'm curious, when you look at your top cloud customers, cloud CapEx close to $700 billion this year, many investors are concerned that it would be harder for this level to grow into next year. And for several of them, their cash flow generation capability is also getting compressed. So I know you're very confident about your roadmap, right, and your purchase commitments and whatnot, but how confident are you about your customers' ability to continue to grow their CapEx? And if their CapEx doesn't grow can NVIDIA still find a way to grow in that envelope?

I am confident in their cash flow growing. And the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents across the world and enterprises everywhere. You're seeing incredible compute demand because of it. In this new world of AI, compute is revenue. Without compute, there's no way to generate tokens. Without tokens, there's no way to grow revenues. So in this new world of AI, compute equals revenues. And I am certain that at this point with the productive use of Codex and Claude Code and the excitement around Claude Cowork and just the incredible enthusiasm about OpenClaw and the enterprise versions of them. All of the enterprise ISVs who are now working on agentic systems on top of their tools platforms. I am certain at this point that we are at the inflection point, we've reached the inflection point and we're generating profitable tokens that are productive for customers and profitable for the cloud service providers. And so the simple logic of it, the simple way to think about it, is computing has changed. What used to be software running on computers, a modest amount of computers, call it, $300 billion or $400 billion worth of CapEx each year has now gone into AI. And AI in order to generate tokens, you need compute capacity. And that translates directly to growth and that translates directly to revenues.

Congratulations on the numbers. You talked about some of the strategic investments that you've made into Anthropic and potentially OpenAI, CoreWeave as well, but also partners, Intel, Nokia, Synopsys. You're clearly at the center of everything. Can you talk about the role of those investments? And kind of how do you view the balance sheet as a tool to kind of grow NVIDIA's position in the ecosystem and participate in that growth?

As you know, fundamentally, at the core of everything NVIDIA is our ecosystem. That's what everybody loves about our business. The richness of our ecosystem. Just about every startup in the world who's working on NVIDIA's ecosystem is built on top of NVIDIA's platform. We're in every cloud. We're in every on-prem data center. We're all over the world's edge and robotic systems. Thousands of AI natives are built on top of NVIDIA. We want to take the great opportunity that we have as we're in the beginning of this new computing era, this new computing platform shift to put everybody on NVIDIA. Everything is already built on CUDA, and so we're starting from a really terrific starting point. But as we build out the entire AI ecosystem, whether it's AI for language or physical AI or AI physics or biology or robotics for manufacturing, we want all of these ecosystems to be built on top of NVIDIA. And this is such a wonderful opportunity for us to invest in the ecosystem across the entire stack. Our ecosystem is also richer today than it used to be. We used to be largely a computing platform on GPUs, but now we're a computing AI infrastructure company, and we have computing platforms on every aspect of that. And everything from computing to AI models to networking to our DPU, all of that has computing stacks on top of it. And as I mentioned before, whether it's in enterprise or in manufacturing, industrial or science or robotics, each one of these ecosystems have different stacks. And we want to make sure that we continue to invest in our ecosystem. So our investments are focused very squarely, strategically on expanding and deepening our ecosystem reach.

Networking continues to rise as a percentage of your overall data center profile, right? Through fiscal '26, your networking revenues accelerated on a year-over-year basis every single quarter, right, with 3.6x growth, as you guys mentioned, year-over-year growth in Q4. Obviously, on the strength of your scale-up and scale-out networking product portfolio, I would seem to remember that first half of last year, your annualized run rate on your Spectrum-X Ethernet switching platform was around $10 billion annualized. It looks like that may have stepped up to around $11 billion, $12 billion in the second half of last year. Jensen, looking at your order book, especially with Spectrum-XGS, upcoming 102T Spectrum-6 Switching platforms launching soon. What is the Spectrum run rate trending now and as you foresee exiting sort of this calendar year?

Yes. We consider ourselves an AI infrastructure company, and our AI computing infrastructure encompasses CPUs, GPUs, and our invention of NVLink, which allows us to expand a single computing node into a large computing rack. We've pioneered the concept of a rack-scale computer, providing racks instead of individual nodes. The NVLink Switch scale-up system extends this capability using Spectrum-X and InfiniBand, both of which we support. Additionally, we can scale across data centers with Spectrum X. Our approach to networking is fundamentally about openness, enabling customers to mix, manage, and integrate solutions in ways that suit their unique data center needs. Ultimately, this is all part of our cohesive platform. The introduction of NVLink has significantly boosted our networking business. Each rack includes nine nodes of switches, and each node has two chips, with plans for more in the future, making our switching capacity per rack quite substantial. We have also become the largest networking company globally, particularly in the Ethernet market, where we've established ourselves as a leading player in Ethernet switching over the last couple of years. Spectrum X Ethernet has been a major success for us. We aim to accommodate different networking preferences; some customers appreciate the low latency and scalability that InfiniBand offers, and we will continue to support that. Others prefer Ethernet integration, which we've enhanced with artificial intelligence capabilities for better data center processing, where we excel. Our Spectrum-X performance reflects that. In large-scale AI setups, a 10% to even a 20% increase in network effectiveness and efficiency translates into significant financial savings. Consequently, NVIDIA's Networking business is expanding rapidly, fueled by our effective AI infrastructure, which is also experiencing impressive growth.

I guess with CPX for large context Windows and Grok likely adding a decode-specific solution. Curious how we should think about your future roadmap? Should we be thinking about customized silicon either by workload or customer as an increasing focus by NVIDIA particularly helped by your move to dilate architecture?

We want to extend the dilate as long as possible because crossing a dilate means crossing an interface, which adds unnecessary latency and power consumption. We're not opposed to using dilates, as we do when there's no other choice. In the Grace Blackwell and Rubin architectures, we've utilized two large reticle-limited dies, which helps minimize architecture crossings. The presence of dilates often impacts our competitors' architectural effectiveness. Many attribute our competitive edge to software, but the relationship between software and architecture is complex. Our software's efficiency stems from our strong architecture. The CUDA architecture is clearly more effective and efficient, providing superior performance per FLOP and watt compared to any other computing architecture due to our unique design approach. Regarding Grok and the low latency decoder, I have some exciting ideas to share at GTC. Our infrastructure's versatility is greatly enhanced by CUDA, and we will continue leveraging that. All our GPUs are architecturally compatible, so when I optimize models for Blackwell, that work also benefits Hopper and Ampere. This compatibility is why the A100 remains relevant and high-performing even years after its launch. It allows us to invest heavily in software engineering and optimizations, ensuring benefits across our entire installed base from various GPU generations. We will maintain this approach, enabling us to extend product life, drive innovation, and enhance performance per dollar and watt for our customers. You'll see what we plan for Grok at GTC; we aim to extend our architecture with Grok as an accelerator, similar to how we integrated NVIDIA’s architecture with Mellanox.

Colette, I wanted to dig a little bit into the call for sequential growth through the year. So I mean, you grew this quarter more than $10 billion sequentially in data center and the guidance seems to imply the bulk of the increased $10 billion sequential in data center and so on. How do you see that as we go through the year, especially as Rubin ramps into the back half? Blackwell has been a pretty massive acceleration for sequential growth. Should we expect something similar as we get to Rubin? And then I was also just hoping you could comment on your expectations for Gaming. I understand the memory issues and everything else. Do you think gaming can still grow year-over-year in fiscal '27? Or will that be under more pressure given memory? So those two questions, please.

Thanks, Stacy. Let me start with the revenue going forward. Again, we're trying to look at revenue quarter-by-quarter. As you think about the full year, we are absolutely going to be still selling and providing Blackwell, probably at the same time that we're also seeing Vera Rubin come to market. This is a very great architecture that helps them just today quickly standing up and have already planned on many different orders across the different customers to provide that. It's too early yet to determine how much in terms of that Vera Rubin, that beginning ramp will start in the second half, and we'll get through it. But no confusion in terms of the strong demand and the interest. We do expect pretty much every single customer to be purchasing Vera Rubin. The question is how soon are we in market and how soon are they able to stand that up in terms of in their data centers. That was your first part. The second part was focusing on our gaming. As much as we would love to have additional more supply, we do believe for a couple of quarters, it is going to be very tight. If things improve by the end of the year, there is an opportunity to think about what that is from a year-over-year growth. But it's still too early for us to know at this time, and we'll get back to you as soon as we can.

Jensen, I'm curious if you can touch on the importance of CUDA as now more of the investment dollars in AI are coming from inference workloads.

Without CUDA, we would be uncertain about how to handle inference. The entire stack from TensorRT LLM, which we introduced a few years back and remains the most efficient inference stack globally, needs us to develop new parallelization algorithms on top of CUDA to effectively distribute the workload and utilize the combined bandwidth of NVLink 72. NVLink 72 has given us a remarkable increase of 50 times more performance per watt. This is a significant advancement. Moreover, NVLink 72 is a remarkable innovation that was challenging to accomplish. The development of switching technology, disaggregating switches, and constructing the system racks were all undertaken openly, and it was clear how challenging it was for us. The results are outstanding. Performance per watt is 50 times, and performance per dollar is 35 times higher. The advancements in inference are remarkable and crucial. It's essential to understand that inference generates revenue for our customers since agents are producing many tokens, and the outcomes are highly effective. When agents code, they can generate thousands to even hundreds of thousands of tokens over a span of minutes to hours. These systems, characterized by their agentic nature, are creating different agents that collaborate as a team. The rate of token generation has significantly increased. Hence, we need to perform inference at a much faster rate. Given that each token has a dollar value, this directly impacts revenue. Consequently, performance in inference translates directly into revenue for our customers. For data centers, the tokens generated during inference per watt directly relate to the revenues of cloud service providers. This is significant because all entities face power limitations. Regardless of the number of data centers, each one has power constraints, whether it's 100 megawatts or 1 gigawatt. Therefore, the architecture that delivers the best performance per watt is crucial because each token, based on performance per watt, converts to dollars. The relationship between tokens per watt and dollars per watt, in a gigawatt context, directly correlates to revenue. It's apparent that every cloud service provider comprehends this, and every hyperscaler recognizes that capital expenditures convert to compute, and compute, when architected correctly, maximizes revenue. Ultimately, compute and revenue are inherently linked, and I believe this understanding is widespread now.

First, let me say kudos on including the stock comp in non-GAAP. I think that's a great move. But that isn't my question. My question is around gross margins and the sustainability of the mid-70s long term. Should we read into the visibility on supply being available into calendar '27 that it's sustainable until then? And then, Jensen, what about after that? Are there innovations in memory consumption you can unveil that makes us feel better about the ability to keep margins at that level for a long time?

The most crucial factor influencing our gross margins is our ability to provide exceptional leads to our customers. This is the top priority. If we can achieve superior performance per watt that significantly surpasses Moore's Law, along with providing performance per dollar that greatly exceeds the cost of our systems, we can maintain our gross margins. This concept is key. We're progressing rapidly because the global demand for tokens has experienced an exponential increase due to recent inflection points. Notably, even our six-year-old GPUs in the cloud are in high demand, driving prices up. We recognize that the computational requirements for modern software are growing exponentially. Therefore, our strategy is to present a complete AI infrastructure every year. This year, we launched six new chips, and the next generation will introduce many more. With every iteration, we are dedicated to delivering significant improvements in performance per watt and performance per dollar. Our pace and capacity for extreme co-design enable us to provide this value to our customers, which is vital to our overall value proposition.

I'd like to ask about space data centers, which some of your customers are considering. How feasible do you think that is and what kind of horizon? And what do the economics look like today? And how do you think that could evolve over time?

Well, the economics are poor today, but it's going to improve over time. As you know, the way that space works is radically different than how it works down here. There's an abundance of energy, but solar panels are large, but there's plenty of space in space. The heat dissipation, it's cold in space. However, there's no airflow. And so the only way to dissipate heat is through conduction, and the radiators that you need to create are fairly large. Liquid cooling is obviously out of the question because it's heavy and freezes. And so the methods that we use here on earth are a little different than the way we would do it in space. But there are many different computing problems that really wants to be done in space. And so NVIDIA is already the world's first GPU in space, Hopper is in space. And one of the best use cases of GPUs in space is imaging, to be able to image at extremely high resolutions using optics and artificial intelligence. And to be able to do that computation of reprojection of different angles and be able to up res and do noise reduction and just be able to see, be able to image at very large, very high resolutions at extremely large scales and very, very fast. It's hard to do that by sending petabytes and petabytes of imaging data back here on earth and doing that work. It's easier just to do it out in space. And then ignore all of the data collected and processed until you see something interesting. And so artificial intelligence in space will have very good, very interesting applications.

I want to follow up on your comment regarding revenue diversification. Colette, you mentioned that hyperscalers accounted for over 50% of revenues, but the growth was driven by your other data center customers. To clarify, does this indicate that your non-hyperscale customers experienced faster growth? If so, can you explain what the non-hyperscalers are doing differently? Are their strategies distinct from those of the hyperscalers, or are they similar but on a different scale? Additionally, do you foresee this trend continuing? Will your customer base evolve in a way that non-hyperscalers become a larger portion of your overall business?

Yes. Let's see if we can help on this question. So when you think about our top 5, as we articulated as being our CSPs, our hyperscalers, and they have right now sat at about 50% of our total revenue. There's a big organization, therefore, of diversity of all different types of companies that we are working with, that it goes through our AI model makers, that goes through our enterprises, that goes to supercomputing, it goes to our sovereigns. There's a lot of other different facts on there. But you are correct. It's a very fast-growing area as well. We have a strong position in terms of all of our different cloud providers on our platform. And now we also have an extreme diversity of different customers that we are seeing all the way across the world. And this will really benefit seeing that diversity and being able to serve all of those parts. I'm going to see if Jensen wants to add about that?

Yes, one of the advantages of our ecosystem built on CUDA is that we are the only accelerated computing platform present in every cloud and available through every computer manufacturer, as well as at the edge. We are now also focusing on telecommunications. It's clear that future radios will be AI-driven and future wireless networks will function as computing platforms. Someone needs to invent the technologies to enable this, and we've created a platform with Aerial for that purpose. Our presence is widespread, from robots to self-driving cars. CUDA benefits from the performance of specialized processors, particularly with Tensor Cores in our GPUs, while also offering the flexibility to address language, computer vision, robotics, biology, physics, and a variety of AI and computational problems. The diversity of our customer base is one of our greatest strengths. Additionally, without our ecosystem, even a programmable processor would struggle to expand beyond captured design wins tied to someone else's ecosystem. Our growth is naturally supported by the platform we developed. Furthermore, our partnerships with OpenAI, Anthropic, xAI, and Meta, along with nearly every open-source project, bolster this. With 1.5 million AI models on Hugging Face all running on NVIDIA CUDA, the open-source community likely represents the second-largest model ecosystem worldwide, following OpenAI. Our capability to support all of this makes our platform highly adaptable, user-friendly, and a secure investment. Consequently, this diversity in customers and platforms ensures we can serve every country and support the global ecosystem.

Yes. I guess sticking with the idea of the platform and extreme co-design. Some of the news over this last quarter has obviously been NVIDIA's ability or push to bring Vera CPUs to market on a standalone solution basis. So I guess, Jensen, I'm curious, of what's the importance of Vera plays in this architecture evolution as we move forward? Is this being driven more by the proliferation or the heterogeneity of inference workloads? I'm just curious of how you see that evolving for NVIDIA, particularly on a stand-alone CPU basis?

Yes. Thanks. And I'll tell you some more about it at GTC. But at the highest level, we made fundamentally different architecture decisions about our CPUs compared to the rest of the world's CPUs. It's the only data center CPU that supports LPDDR5. It is designed to be focused on very high data processing capabilities. And the reason for that is because most of the computing problems that we're interested in are data-driven, artificial intelligence being one. And the single-threaded performance in this ratio with bandwidth is just off the charts. And we made those architectural decisions because in the entire phase, the different phases of AI from data processing, before you even do training, you have to do data processing. So you have data processing, pre-training and in post-training now, the AIs are learning how to use tools. And the usage of tools, many of those tools run in CPU-only environments or they run in CPU with GPU-accelerated environments. And Vera was designed to be an excellent CPU for post-training. And so some of the use cases in the entire pipeline of artificial intelligence includes using a lot of CPUs. We love CPUs as well as GPUs. And when you accelerate the algorithms to the limit as we have, Amdahl's Law would suggest that you need really, really fast single-threaded CPUs, and that's the reason why we built Grace to be extraordinary to be great at single-threaded performance, and Vera is off the charts better than that.

Colette, could you discuss the deployment of capital? I understand you have increased the purchase commitments significantly, but it seems like you might have surpassed the toughest part of this process, and you anticipate generating around $100 billion in cash this year. Despite strong results, the stock hasn't seen much of an increase, so it appears you might consider this a good opportunity to buy back shares. Could you elaborate on this and explain why not make a substantial move now for a significant share repurchase?

So thanks for the question. We look at our capital return very, very carefully, and we do believe that one of the most important things that we can do is really supporting the extreme ecosystem that's in front of us, that stems from everywhere from our suppliers and the work that we need to do to assure that we can have the supply that's needed and help them from a capacity all the way that we are in terms of the early developers of the AI solutions that will be on our platform. So we will continue to make this a very important part of our process and strategic investments. But of course, we are still repurchasing our stock. We are still with our dividend as well, and we will continue to find the right unique opportunities within the year for doing those different purchases.

Jensen, you've mentioned the potential for data center capital expenditures to reach $3 billion to $4 trillion by 2030, suggesting an accelerated growth rate that you hinted at for the next quarter. What key application areas do you think will drive this growth? Is it physical AI, agentic, or something different? Do you still feel confident about that $3 billion to $4 trillion range?

Let's take a closer look at this from a few perspectives. First, our core belief is that the future of software, particularly with AI, is centered around token-driven models. Discussions on tokenomics highlight how data centers are generating tokens and how inference also revolves around token generation. For instance, the advancements with NVIDIA NVLink 72 allow for token generation at 50 times the performance per unit of energy compared to previous models. If we reflect on past computing demands, the requirements for software were minimal compared to what is anticipated in the future. AI is here to stay, and it will continue to improve. Historically, the world invested around $300 billion to $400 billion annually in traditional computing, but with the rise of AI, the need for computation has increased dramatically—by a factor of 1,000 compared to previous methodologies. The demand for computing resources will necessitate significantly more than $700 billion in investment for token generation. I am confident in our ability to keep generating tokens and expanding our compute capabilities. Every business now relies on software, which increasingly depends on AI. This makes every company a sort of AI factory, be it cloud data centers generating tokens for revenue or enterprise software companies creating tokens for their systems. Even robotics factories, exemplified by self-driving cars, utilize large supercomputers—essentially AI factories—to produce the necessary tokens for their operations. Moreover, these vehicles must be equipped with computers to consistently generate tokens. It is becoming clear that this is the future of computing. The shift from pre-recorded to real-time generative software is a significant change. Unlike earlier methods where everything was recorded in advance, today’s AI can create responses based on context, the user's situation, and intentions. This demands far more computational power than previous software approaches, akin to how a modern computer far surpasses a DVD recorder in capabilities. From an industrial perspective, all companies ultimately rely on software. As the demand for this new software grows, which requires token generation, it will drive an increase in data center construction, thereby boosting revenues. This understanding is becoming more prevalent. Additionally, the benefits that AI brings to the world have to translate into revenue creation. We're witnessing the emergence of agentic AI, which has reached a turning point recently. Agents are proving capable of tackling real-world problems, with our developers at NVIDIA making extensive use of systems like Claude Code and OpenAI Codex to enhance their work. We’re observing remarkable revenue growth, such as Anthropic's revenue increasing tenfold in a year, evidencing high demand and constraints in their capacity. The exponential growth in token generation rates is mirrored by companies like OpenAI, which also face incredible demand. The more computational power they can harness, the faster their revenues will rise, reinforcing the idea that in this new landscape, compute translates directly to revenue. This marks a new industrial revolution, with fresh factories and infrastructure being developed. The new computing paradigm is irreversible. If we believe in the future of token generation, as I do, and as a significant part of the industry does, we will continue to expand our capacities. The current wave is driven by agentic AI, and the next phase will involve physical AI applications in fields like manufacturing and robotics, presenting a tremendous opportunity ahead.

This concludes the question-and-answer session. I'll turn the call to Toshiya Hari. In closing, please note Jensen will be participating in a fireside chat at the Morgan Stanley TMT Conference in San Francisco on March 4. He'll also be giving a keynote at GTC in San Jose on March 16. Our earnings call to discuss the results of our first quarter of fiscal 2027 is scheduled for May 20. Thank you for joining us today. Operator, please go ahead and close the call.