Good morning and welcome to NVIDIA's 2025 Annual Meeting of Stockholders. I'm Toshiya Hari, Vice President of Investor Relations and Strategic Finance. We are hosting a fully virtual annual meeting again this year. As a reminder, the rules of conduct for the meeting are posted on our virtual meeting website. We ask that you follow them to help the meeting run smoothly. Other members of NVIDIA's management who are on the line with me today are Jensen Huang, President, Chief Executive Officer, and Director; Colette Kress, Executive Vice President and Chief Financial Officer; and Timothy Teter, Executive Vice President, General Counsel, and Secretary. I would also like to introduce the Outside Members of our Board who are in attendance: Rob Burgess, Tench Coxe , John Dabiri, Persis Drell, Dawn Hutson, Harvey Jones, Melissa Lora, Steve Neal, Ellen Ochoa, Brooke Seawell, Aarti Shah, and Mark Stevens.
Finally, I would like to introduce Sarah Beelard from PricewaterhouseCoopers LLP, our independent registered public accounting firm. After the former portion of the meeting and Jensen's business update, we will set aside some time for questions. I will now turn the meeting over to Tim.
Thanks, Toshiya. Good morning and welcome to our 2025 Annual Meeting, which will now officially come to order. Jensen will serve as Chairman. I will serve as the Secretary and conduct the procedural portion of the meeting. First, a few housekeeping items. We have opened the online portal for stockholders to vote their shares and I will make an announcement when the polls are about to be closed. Stockholders may ask questions during the meeting. If you have a question, please enter it into the portal. The meeting rules of conduct have been posted to the virtual meeting website. As stated in our rules of conduct, the meeting is not to be used as a forum to present general, economic, political, or other views that are not directly related to our business, the business of the meeting, or for matters of individual concern.
In fairness to all attendees, we will limit each stockholder to one question. We may provide a single response to multiple questions generally asking the same or similar things. After the conclusion of the meeting, we will publish the questions, if pertinent to NVIDIA's business or the business of the meeting, along with our responses on our Investor Relations website. During this meeting, we may make forward-looking statements based on current expectations. These forward-looking statements 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 reports we may file from time to time with the Securities and Exchange Commission, including our annual report on Form 10-K and our quarterly reports on Form 10-Q.
All of our statements are made as of June 25, 2025, based on information available to us as of today, and except as required by law, we assume no obligation to update them. We will address the matters described in our proxy statement dated May 13, 2025. Following Jensen's business update, the polls will be closed, we'll announce preliminary voting results, and then the official portion of the meeting will be adjourned. I have a complete list of the stockholders of record of NVIDIA's common stock on the April 28, 2025, record date for this meeting. I also have an affidavit from Broadridge certifying that they commenced the mailing of the proxy materials on May 13, 2025. I'm appointing Chris Woods of American Election Services to act as the inspector of elections at this meeting. He will tally the final votes when balloting on all matters is completed.
Chris has taken the customary oath of office, and we will file this oath with the records of the meeting. The presence, in person or by proxy, of a majority of the shares entitled to vote at the meeting will constitute a quorum. There were approximately 24 billion shares outstanding on the record date, and Chris has informed me that proxies have been received for approximately 19 billion shares, or approximately 80% of the shares outstanding, which constitutes a quorum for today's meeting. Each share of common stock is entitled to one vote. If you are eligible to vote and have not submitted your proxy, or if you want to change your vote, please vote online now. You do not need to vote if you've already sent in your signed proxy or voted online or by telephone. Your votes will be counted automatically. The time is 9:05 A.M.
Pacific time, and the polls are currently open for voting. There are seven items of business for this meeting: four management proposals and three stockholder proposals. The management proposals are as follows: Proposal One: The election of Rob Burgess, Tench Coxe , John Dabiri, Persis Drell, Jensen Huang, Dawn Hutson, Harvey Jones, Melissa Lora, Ellen Ochoa, Steve Neal, Brooke Seawell, Aarti Shah, and Mark Stevens to serve as Directors until our 2026 Annual Meeting. Proposal Two: The approval of the compensation of our named executive officers for Fiscal Year Year 2025, as disclosed in our proxy statement. Proposal Three: The ratification of PricewaterhouseCoopers LLP as our independent registered public accounting firm for Fiscal Year Year 2026. Proposal Four: The approval of an amended and restated certificate of incorporation to remove all supermajority provisions. The stockholder proposals are as follows.
Proposal Five: The approval of a stockholder proposal to eliminate the holding period requirement to call a special stockholder meeting. Proposal Six: The approval of a stockholder proposal to adopt a new Director Election Resignation Governance Policy. Proposal Seven: The approval of a stockholder proposal to modify existing reporting on workforce data. The Board recommends a vote for each of the four management proposals and against the three stockholder proposals. The proxy statement provides the company's responses to the stockholder proposals and explains why the Board of Directors recommends a vote against each stockholder proposal. We'll now ask each stockholder proponent to introduce their proposal. At this time, we would like to introduce John Chevedden to present Proposal Five. Mr. Chevedden, you have three minutes. Please begin.
Hello, this is John Chevedden. Proposal Five: Special shareholder meeting improvement. Charles, ask the Board of Directors to take the steps necessary to remove the current provision that considers the voice of certain NVIDIA shareholders as non-shareholders. Currently, all NVIDIA shares held for less than one continuous year are considered non-shareholders if they seek to call for a special shareholder meeting on an important business matter. The current exclusion of all NVIDIA shares held for less than one continuous year makes the current so-called shareholder right to call for a special shareholder meeting useless. There's no point to have a useless right on the books of NVIDIA.
The reason to enable all shareholders to call for a special shareholder meeting is to allow one shareholder or a group of shareholders to quickly acquire NVIDIA shares to equal the challenging 15% share ownership requirement based on all shares outstanding, to call for a special shareholder meeting to incentivize a turnaround of NVIDIA should NVIDIA find itself in a slump in the future. The best strategies for turning around a company do not necessarily come from the company's existing shareholders. If NVIDIA finds itself in a future slump, NVIDIA shareholders and potential NVIDIA shareholders will not even consider acquiring more shares in order to call for a special shareholder meeting if they have to wait one year to call for a special shareholder meeting. A one-year holding period makes no sense if slumping stock price demands a quick response.
The fact that one shareholder or a group of shareholders can quickly acquire more shares to call for a special shareholder meeting is an incentive for NVIDIA Directors to avoid a slump in the first place, since the continued service of certain NVIDIA Directors could be terminated by a special shareholder meeting. This is a good incentive for the NVIDIA Directors to have for the benefit of all NVIDIA shareholders. At minimum, this proposal alerts shareholders to the severe limitation to the point of uselessness baked into the current NVIDIA rules for shareholders to call for a special shareholder meeting. Please vote yes, Special Shareholder Meeting Improvement, Proposal Five.
Thank you, Mr. Chevdon. At this time, we would like to introduce the New York City Carpenters Pension Fund, who has provided a prerecorded statement presenting Proposal Six. The New York City Carpenters Pension Fund has three minutes to present. Operator, please begin the New York City Carpenters Pension Fund statement.
My name is Michael Piccirillo, and I represent the New York City Carpenters Pension Fund, the proponent of the Director Election Resignation Guideline proposal. The fund is a long-term holder of NVIDIA common stock and supports the company's Board of Directors nominees. We appreciate the dialogue we have had with company representatives on the post-election director resignation issue. Shareholders possess several rights, with none more important than a right to vote for nominees. The company has in place a governance guideline that requires an incumbent director nominee to tender resignation for the Board consideration should the director nominee fail to win reelection under the majority vote standard. The Board has full discretion to determine whether to accept or reject a director's resignation. If the resignation is not accepted, the unelected director continues to serve as a holdover director under state corporate law.
The fund proposal is straightforward and designed to strengthen shareholder voting rights. The Director Resignation Guideline, if implemented, would require the Board to accept the tendered resignation of an incumbent director following his or her failure to be reelected under a majority vote standard. The majority vote standard in elections was instituted for the explicit purpose of giving shareholders a meaningful right to determine who is elected to a Corporate Board. The election votes are not adversary. Rather, they are a legal effect in determining whether a director is elected. The company's current Director Resignation Policy, like that at most corporations, weakens shareholder voting rights through the grant of total discretion to the Board to determine an election outcome. The proposed resignation is designed to bolster shareholder voting rights in director elections.
The Board accountability created by the proposal guideline, combined with voting policies based on long-term corporate performance, will help focus-based Boards and management teams on the task of long-term value creation. We believe this is an overdue reform which serves the best interests of NVIDIA and its shareholders. Thank you.
Thank you. At this time, we would like to introduce the Trillium ESG Global Equity Fund, which has provided a prerecorded statement presenting Proposal Seven. Trillium has three minutes to present. Operator, please begin Trillium's statement.
Good morning. My name is Hyewon Han, and I am here to represent Trillium ESG Global Equity Fund. I hereby move Proposal Seven, requesting NVIDIA's workforce reporting include its EEO-1 report. Ensuring diverse, inclusive, and fair workplaces can benefit companies. Research indicates that companies with diverse workforces and fair and inclusive cultures have stronger growth prospects and are likelier to financially outperform their peers. Furthermore, the Black Lives Matter movement in 2020 highlighted the systemic barriers that Black people in the United States experience, including in employment. In response, companies committed significant resources to dismantle structural barriers and make workplaces more fair. NVIDIA was one of these companies. In its 2021 sustainability report, the company pledged to create more opportunities for Black and African-American employees. Yet, its workforce reporting since then is puzzling.
The representation of Black employees at NVIDIA has declined year -over -year, from 2.5% in 2021 to 2.1% in 2024, despite its pledge. Both numbers are well below the 8% average Black representation in the tech industry and 13% Black representation in the United States labor force. We believe the disparity raises important questions. In which job categories are the declines occurring? Are the resources invested having the intended impact? How can shareholders evaluate progress without better data? This is one of the reasons we believe NVIDIA should publicly disclose its EEO-1 report. The report provides a standardized breakdown of a company's workforce by race, gender, and job category and would support investors in assessing NVIDIA's human capital management performance at a granular level. Additionally, the standardized nature of the EEO-1 report allows for comparison across firms and sectors.
It is a valuable data set that, contrary to the Board's opposition statement, would provide additional helpful disclosures to investors. The broader market seems to agree. In 2021, just 33% of the S&P 500 companies disclosed their EEO-1 reports. Now, over 80% of the S&P 500 companies disclose EEO-1 reports following investor requests for more human capital management data and broader market trends. We urge NVIDIA to join its peers in reporting, help close the data gap, and demonstrate its commitment to transparency and fair workplaces. For these reasons, we urge you to vote for Proposal Seven. Thank you for your time.
Thank you. As the company has not received notice from any of its stockholders of any other matter to be considered at today's meeting, no other proposals will be addressed. If you have not voted and intend to vote, or if you want to change your vote, please do so online now. Proxies, votes, or any changes or revocations submitted after the closing of the polls will not be accepted. I'll turn it over to Jensen for a business update, following which we will announce the preliminary results of the meeting, if available, and then answer stockholder questions.
Thank you, Tim. This has been an extraordinary year for our company and industry. A new era of computing is in full steam. AI is transforming every sector, from cloud to enterprise to robotics, and the demand for NVIDIA computing has never been greater. At the center of this shift is a new kind of infrastructure: AI factories coming online around the world. NVIDIA is building the full-stack, data-center-scale platform to power them. Let me tell you how we got here. We started with a GPU chip, a processor for graphics with the potential of much more. In 2006, we introduced CUDA, opening the door to general-purpose accelerated computing. In 2016, we launched DGX-1, the first AI supercomputer. I delivered the first one to OpenAI. As AI model complexity grew exponentially, we turned to the entire data center. We built the full-stack: chips, systems, networking, and software.
In 2020, Mellanox Networking joined NVIDIA and brought it all together. The data center became a unified computer. Recognizing the importance of building complex software at data center scale, we built and operated our first supercomputer in June 2020. Selene is the first of a dozen NVIDIA AI supercomputers in operation today. Over the course of two decades, we completely transformed NVIDIA. We stopped thinking of ourselves as a chip company long ago. NVIDIA is an AI infrastructure and computing platform company today. Electricity powers every industry. The internet connects every system. Electricity and the internet are essential infrastructure for every society today. AI is a digital intelligence infrastructure and will drive the next wave of economic growth. AI is the next essential infrastructure for every society. Every company will use them. Every country will build them. Fiscal Year Year 2025 was a defining year. We transitioned from Hopper to Blackwell.
Our platform for real-time inference and AI at data center scale. Blackwell's rollout has been extraordinary. It debuted with $11 billion in sales in the fourth quarter and went on to more than double the contribution in our first quarter this year, easily the fastest ramp in our history. Today, nearly 100 NVIDIA AI-powered AI factories buildouts are in flight around the world. That's double what we saw a year ago, and they're getting larger. The average number of GPUs per factory has also doubled. These breakthroughs drove strong performance. Revenue more than doubled to $130 billion. Operating income and EPS grew 147%. We're at the beginning of a decade-long AI infrastructure buildout. Demand for sovereign AI is growing around the world. We're partnering with companies and governments across Europe, Japan, South Korea, Taiwan, Canada, Southeast Asia, the Middle East, Latin America, and Africa to build regional AI infrastructure.
Now, for the first time, we're building NVIDIA AI supercomputers in the United States. This milestone reflects President Trump's bold vision to reindustrialize America. His leadership, policies, and unwavering support for advanced manufacturing have accelerated the return of critical industries to the United States. That is what made it possible for NVIDIA to build the future of AI infrastructure right here at home. Blackwell chips are now in production in Arizona. Complete Blackwell systems are being assembled in Texas. Together with TSMC, Foxconn, Wistron, Amkor, and SPIL, we've secured over 1 million sq ft of advanced manufacturing. This investment is expected to create tens of thousands of jobs and bring critical AI infrastructure closer to the industries that depend on it. This is how we strengthen supply chains, support national priorities, and ensure the world builds on American technology. This year, we hosted GTC in San Jose, Taipei, and Paris.
Thousands joined us in person. Millions more participated online. GTC has become the launchpad of the AI ecosystem. AI is advancing at light speed. It started with perception, recognizing images, speech, and patterns. Then came generative, creating text, code, and meaning. Now we're entering the era of reasoning and problem-solving. Just ahead is physical AI systems that can understand and interact with the real world. State-of-the-art models reason, have short-term, long-term memory, plan, and use tools. They need orders of magnitude more compute to effectively think, solve novel problems, and produce higher-quality answers. Inference token generation is surging. Microsoft alone processed over 100 trillion tokens last quarter, five times more than a year ago. Grace Blackwell, essentially a thinking machine, was built for this moment. Grace Blackwell is a fully integrated rack system.
Seventy-two GPU modules are unified into a single massive computational entity through our revolutionary NVLink switch architecture. Each GPU module features two reticle-limited GPU dies, precisely stacked and interconnected via a silicon interposer. NVLink enables every GPU to communicate simultaneously with all others, delivering an aggregate bandwidth of 130 TB per second, comparable to the total peak bandwidth of the world's internet traffic. The NVLink switches in each rack are connected over 5,000 meticulously engineered impedance-controlled coaxial cables, spanning a total of two miles and weighing 60 lbs. Each rack consists of 1.2 million components, runs at 120 KW, and is fully liquid-cooled and weighs nearly two tons. Each Blackwell rack, Grace Blackwell rack, delivers nearly 1 1/2 exaFLOPS of performance, more than an entire supercomputer just a few years ago.
In the latest MLPerf, Blackwell has achieved world records in training as well as inference, showing broad adoption. There are 56 B200 air-cooled system submissions from partners and 51 GB200 NVL 72 liquid-cooled system submissions. Grace Blackwell is built across 150 factories. We ship thousands of racks each month. Blackwell is the most ambitious and expansive architecture in our history. Whereas Hopper was largely designed for AI clouds, Blackwell architecture spans every segment of computing, from desktops to data centers, cloud to edge, x86 to Grace, Windows to Linux, Hypervisor to bare metal, AI to supercomputing to graphics. It is one platform. Same software, same models, same tools. NVIDIA GGX for developers, NVIDIA RTX Pro for enterprise, NVL 72 for AI supercomputing at scale. Train anywhere, deploy anywhere, it all works the same. Just beyond AI agents, the next wave is already coming. The next wave is physical AI.
Robots that see, reason, and interact with the physical world. Robots and all the AI infrastructure to train them will be the next multi-trillion-dollar industry. NVIDIA builds the AI factories to train them, the digital twin worlds to simulate them, and the robotic computers deploy them. We give them intelligence in an AI factory, fine-tune their skills in a virtual gym, and eventually deploy them into factories, warehouses, and homes to increase the world's labor force. We're at the beginning of a powerful new wave of growth. Grace Blackwell, the first thinking machine, is built to power the full lifecycle of AI, from training giant models to running reasoning agents at light speed. Our rack-scale AI supercomputers are in full production, ramping across every market segment: cloud, enterprise, and edge. AI is no longer one-shot.
Reasoning inference demands orders of magnitude more compute, and token generation has exploded up 50-100 times in just the past year. To meet this exponential surge, giant AI factories are rising everywhere. Some still think inference is easy. It's not. Inference at scale is extreme computer science and systems engineering. Algorithms are evolving constantly. Architecture must be flexible to adapt, performant to deliver extraordinary experiences, and efficient to sustain the economics of generating trillions upon trillions of tokens. CUDA, in our end-to-end data center scale expertise, let us drive advancements like no other. The next wave is physical AI: robots and autonomous machines that can do physical work and amplify the world's labor force. With $50 trillion of global GDP in industrial sectors, the opportunity to rewire the world's factories, warehouses, and supply chain with intelligence is massive. The age of AI is here.
Every country will build it. Every company will be built on it. Like electricity and internet before it, AI is the next essential infrastructure, and NVIDIA is leading the world to build it. Thank you.
Thanks, Jensen. The time is 9:29 A.M. Pacific Time, and the polls are now closed. Based on preliminary voting results, the Inspector of Elections has confirmed that all 13 nominees have received the requisite number of votes to be re-elected as Directors of the company, and proposals two and three have been approved. Proposals four, five, six, and seven have not been approved. The final vote results are not yet available to us, but will be published in a Form 8K, which we expect to file with the U.S. Securities and Exchange Commission within four business days. That concludes the formal portion of today's annual meeting, and I now declare the official portion of the meeting adjourned.
I'll turn it over to Toshiya for stockholder questions. Thanks, Tim.
We will now move to Q&A. We grouped together duplicative questions that we received to streamline the discussion. For our first question, how can NVIDIA improve sales given competition?
We have a multi-trillion-dollar long-term growth opportunity with AI and robotics. Our markets are rapidly expanding, and we're confident in our ability to grow. While we have a lot of competition, we have leadership positions in the markets in which we play. There are a number of elements that will help us maintain our lead. We're an AI infrastructure company offering the full stack of computing, networking, software, and system-level innovations. A large and growing footprint of NVIDIA data center infrastructure and thousands of ecosystem partners, combined with over 6 million developers and continued software optimization, ensure a high pace of innovation on our platform.
Finally, we're shipping new data center products every year, raising the bar on performance and our competition with each turn of the calendar.
For our next question, where do you see growth and profit opportunities for NVIDIA?
We have many growth opportunities across our company with AI and robotics, the two largest, representing a multi-trillion-dollar growth opportunity. In AI compute, demand is surging due to increased use of innovative applications like ChatGPT, continued proliferation of generative and agentic AI applications. The next generation reasoning models consume orders of magnitude more inference compute versus single-shot models. NVIDIA's Blackwell platform was created to run reasoning AI inference applications. Not surprisingly, the vast majority of the world's inference runs on our infrastructure. AI factories, a new type of data center, are being stood up all over the world to manufacture AI and power these next-generation applications.
AI factory projects require tens of GW of AI infrastructure. It's set to be built in the coming years, and NVIDIA is uniquely positioned to capture this opportunity with our full-stack approach, system engineering expertise, and large ecosystem of technology partners. In robotics, NVIDIA is poised for significant growth, with autonomous vehicles serving as the first major commercial application. NVIDIA's Drive platform is powering the transition to software-defined vehicles and providing the immense computational horsepower required for self-driving capabilities. It has seen adoption across the transportation industry, from robo-taxis to delivery vehicles to Mercedes-Benz, which is deploying DRIVE across its fleet of millions of vehicles. We recently released NVIDIA Cosmos, a new family of reasoning AI models for humanoid robotics. We're working with many of the leading companies in the space, including Boston Dynamics, KUKA, and Universal Robots.
We're working towards a day where there will be billions of robots, hundreds of millions of autonomous vehicles, and hundreds of thousands of robotic factories that can be powered by NVIDIA technology.
What is NVIDIA's plan if there is a sudden loss of interest in artificial intelligence and it cannot do all that is promised? Could we be doing more to develop sales in gaming, which seems to be the other obvious market?
First, AI is the future of computing. Every industry will be affected, including gaming, and will be an essential infrastructure that transforms every industry. The platforms that power growth across our markets are developed from a single architecture. Investments in one area can be leveraged across our products and services in other areas, enhancing their features, functionality, and capability.
For example, to power AI algorithms, our new Blackwell gaming GPUs include Tensor Cores, the same technology found in our data center GPUs. In Fiscal Year 2025, we invested nearly $13 billion in research and development. These investments speed the velocity of innovation and product introductions and will help grow our markets and increase sales. The shift to accelerated computing and AI is a major long-term growth opportunity for our business. The end of Moore's Law is necessitating a new computing stack comprised of GPUs, algorithms, and systems advances that will ensure the economic and technological benefits that have propelled human progress for over decades and will continue in the future.
Next question is about the company's strategic direction beyond Blackwell. Does management believe they have applied enough resources to launch a parallel quantum computing effort beyond the Blackwell chip's advancement in classical semiconductor computing technology?
As we know now, quantum computing should potentially revolutionize computing in the same space massive GPU arrays do today. What is NVIDIA's strategy for parallel quantum computing beyond its Blackwell architecture? Beyond Blackwell, we have Vera Rubin and Feynman. Once a year, we're going to have breakthroughs in computing. Quantum computing will serve a niche of computing that is quite special. Commercial quantum applications will run on hybrid GPU quantum computers, consisting of GPUs and quantum processing units, or QPUs. Far from being in competition with QPUs, useful quantum computing will rely upon partnering with AI supercomputing and GPUs. Blackwell is already helping researchers work towards full quantum computing. Through our CUDA-Q software platform, GB200 has demonstrated world-record simulation, world-record results in quantum hardware and algorithm simulations. Researchers are connecting their QPUs to our GPUs in supercomputing centers around the world to do quantum error correction.
Our strategy is to advance quantum computing with our CUDA-Q GPU quantum platform. Our platform innovations are complemented by our work to nurture and grow a vibrant and innovative ecosystem of developers, research labs, system builders, and Fortune 500 companies that are leveraging our CUDA-Q platform to move the industry forward.
Will NVIDIA increase its dividend? Will there be another stock split?
We return capital to our stockholders through share repurchases as well as quarterly cash dividends. For Fiscal Year 2025, we utilized $33.7 billion of cash for share repurchases and paid an additional $834 million in dividends. A year ago, we increased our quarterly cash dividend by 150%. In Fiscal Year 2025, we also completed a 10-for-1 split of our stock. Our Board reviews our capital return program regularly and would consider a stock split if it is in the best interest of our stockholders.
Our program has now concluded. A copy of this webcast will be available online on our website through June 24, 2026. We look forward to another great year at NVIDIA. Thank you for attending and for your continued support of NVIDIA. Our 2025 annual meeting is now closed. This now concludes the meeting. Thank you for joining, and have a pleasant day. The host has ended this call. Goodbye.