Public Figure Quotes← All people
Jensen Huang

Jensen Huang

Photo: The White House, Public domain

Born1963-02-17

Occupationbusinessperson, chief executive officer, computer scientist, electrical engineer

EducationAloha High School, Oneida Baptist Institute, Oregon State University, Stanford University

AwardsIEEE Founders Medal, Queen Elizabeth Prize for Engineering, Time 100

Highlights

Tokens are now profitable, so model makers are in a race to produce more.
NVDA Q1 2027 earnings call2026-05-20Transcript
It is very clear Compute is revenues.
NVDA Q1 2027 earnings call2026-05-20Transcript
It's the most popular opensource project in the history of humanity and it did so in just a few weeks.
YouTube2026-03-16Watch1:47:32
It started the big bang of AI.
YouTube2026-03-16Watch12:56
I've said before, if you have the wrong architecture, even if it's free, it's not cheap enough.
YouTube2026-03-16Watch1:04:21
This is as big of a deal as HTML.
YouTube2026-03-16Watch2:05:28
When you accelerate data processing, when you accelerate computing, you get the benefit of speed, you get the benefit of scale, but most importantly, you also get the benefit of cost.
YouTube2026-03-16Watch22:49
The Chad GPT moment of self-driving cars has arrived.
YouTube2026-03-16Watch2:07:16
He says, "Jensen sandbagged.
YouTube2026-03-16Watch1:03:54
People have heard me say I believe that computing demand has increased by 1 million times in the last two years.
YouTube2026-03-16Watch51:48
The greatest infrastructure buildout in history is underway.
YouTube2026-03-16Watch1:42:56
We are an algorithm company.
YouTube2026-03-16Watch37:26
For any human, this would be a dream come true.
time.com2025-12-16Source
It didn’t matter to us whether people believed in us. We believed in ourselves. We had the courage to follow our own path,
carnegie.org2024-06-26Source
Back then, there wasn’t a counsellor to talk to … you just had to toughen up and move on,
carnegie.org2024-06-26Source
Browse all 347 quotes
Tokens are now profitable, so model makers are in a race to produce more.
NVDA Q1 2027 earnings call2026-05-20Transcript
It is very clear Compute is revenues.
NVDA Q1 2027 earnings call2026-05-20Transcript
It's the most popular opensource project in the history of humanity and it did so in just a few weeks.
YouTube2026-03-16Watch1:47:32
It started the big bang of AI.
YouTube2026-03-16Watch12:56
I've said before, if you have the wrong architecture, even if it's free, it's not cheap enough.
YouTube2026-03-16Watch1:04:21
This is as big of a deal as HTML.
YouTube2026-03-16Watch2:05:28
When you accelerate data processing, when you accelerate computing, you get the benefit of speed, you get the benefit of scale, but most importantly, you also get the benefit of cost.
YouTube2026-03-16Watch22:49
The Chad GPT moment of self-driving cars has arrived.
YouTube2026-03-16Watch2:07:16
He says, "Jensen sandbagged.
YouTube2026-03-16Watch1:03:54
People have heard me say I believe that computing demand has increased by 1 million times in the last two years.
YouTube2026-03-16Watch51:48
The greatest infrastructure buildout in history is underway.
YouTube2026-03-16Watch1:42:56
We are an algorithm company.
YouTube2026-03-16Watch37:26
For any human, this would be a dream come true.
time.com2025-12-16Source
It didn’t matter to us whether people believed in us. We believed in ourselves. We had the courage to follow our own path,
carnegie.org2024-06-26Source
Back then, there wasn’t a counsellor to talk to … you just had to toughen up and move on,
carnegie.org2024-06-26Source
This was an extraordinary quarter.
NVDA Q1 2027 earnings call2026-05-20Transcript
And there are trillion dollars this year I have every expectation it is gonna grow from here.
NVDA Q1 2027 earnings call2026-05-20Transcript
And then, of course, physical AI.
NVDA Q1 2027 earnings call2026-05-20Transcript
But we will have billions of agents.
NVDA Q1 2027 earnings call2026-05-20Transcript
So our share of inference is growing very quickly.
NVDA Q1 2027 earnings call2026-05-20Transcript
OpenAI's Codex is a harness around the GPT 5.5 model.
NVDA Q1 2027 earnings call2026-05-20Transcript
However, long term, if you look at look at industrial and enterprise, clearly, that is where future economics is going to be.
NVDA Q1 2027 earnings call2026-05-20Transcript
And as you know, NVIDIA has a large the largest suite of acceleration libraries in the world from computational lithography to fluid dynamics to particle physics to molecular dynamics to the list goes on.
NVDA Q1 2027 earnings call2026-05-20Transcript
But AI requires a tremendous amount of compute.
NVDA Q1 2027 earnings call2026-05-20Transcript
And we are growing share in inference very, very quickly.
NVDA Q1 2027 earnings call2026-05-20Transcript
In the future, it is gonna be about personal AI.
NVDA Q1 2027 earnings call2026-05-20Transcript
The second cluster is growing incredibly fast.
NVDA Q1 2027 earnings call2026-05-20Transcript
And so VeraRubin is off to a tremendous start and it will it will surely be more successful than even Grace Blackwell.
NVDA Q1 2027 earnings call2026-05-20Transcript
And my sense is that we will be supply constrained throughout the entire life of VeraRubin.
NVDA Q1 2027 earnings call2026-05-20Transcript
And so what we need to do in the future is to generate tokens process tokens as fast as possible.
NVDA Q1 2027 earnings call2026-05-20Transcript
Edge including, of course, what is what most people see, self driving cars, robotics.
NVDA Q1 2027 earnings call2026-05-20Transcript
In the AI era, compute capacity is revenue, and profits.
NVDA Q1 2027 earnings call2026-05-20Transcript
And if they do not compute they will not have the revenues.
NVDA Q1 2027 earnings call2026-05-20Transcript
NVIDIA sits at the center of these transitions.
NVDA Q1 2027 earnings call2026-05-20Transcript
First way is that we help the hyperscale clouds accelerate their data processing and machine learning workloads.
NVDA Q1 2027 earnings call2026-05-20Transcript
NVIDIA is the platform of this era.
NVDA Q1 2027 earnings call2026-05-20Transcript
The LPX is designed for low latency and high token rate.
NVDA Q1 2027 earnings call2026-05-20Transcript
Well, agents do not rent cores.
NVDA Q1 2027 earnings call2026-05-20Transcript
That go to market is very complex, very diverse.
NVDA Q1 2027 earnings call2026-05-20Transcript
Economics of AI of the future is tokens per dollar.
NVDA Q1 2027 earnings call2026-05-20Transcript
Demand has gone parabolic.
NVDA Q1 2027 earnings call2026-05-20Transcript
Today, yesterday's computing was largely about personal computing.
NVDA Q1 2027 earnings call2026-05-20Transcript
So the amount of capacity that we are gonna bring online for Anthropic this year and next year is going to be quite significant.
NVDA Q1 2027 earnings call2026-05-20Transcript
They just want the work to be done fast.
NVDA Q1 2027 earnings call2026-05-20Transcript
And the reason for that is this year, the number of frontier model companies grew.
NVDA Q1 2027 earnings call2026-05-20Transcript
And so you expect the second category to develop slower than hyperscale.
NVDA Q1 2027 earnings call2026-05-20Transcript
And so we are gaining share tremendously fast in inference.
NVDA Q1 2027 earnings call2026-05-20Transcript
Second, we are in every hyperscale cloud.
NVDA Q1 2027 earnings call2026-05-20Transcript
AI can now do productive and valuable work.
NVDA Q1 2027 earnings call2026-05-20Transcript
Instead of 6 or 7 companies representing the revenues associated with our first category, The second category is hundreds, thousands of companies, and in the future, it will be hundreds of thousands of companies.
NVDA Q1 2027 earnings call2026-05-20Transcript
The fact that they can grow within 1 month what some of those SaaS companies would have taken a decade to grow tells you something.
NVDA Q1 2027 earnings call2026-05-20Transcript
We are fairly unique in our abilities to be able to serve this industry.
NVDA Q1 2027 earnings call2026-05-20Transcript
Vera opens a brand new $200 billion TAM for NVIDIA, a market we have never addressed before.
NVDA Q1 2027 earnings call2026-05-20Transcript
For fundamentally good reasons.
NVDA Q1 2027 earnings call2026-05-20Transcript
Many industrial companies there is no choice but to put the computer where the context is, where the action is, You cannot put that in the cloud.
NVDA Q1 2027 earnings call2026-05-20Transcript
First, NVIDIA is the only platform that runs every frontier AI model.
NVDA Q1 2027 earnings call2026-05-20Transcript
We built NVIDIA compute platform over 3 decades.
NVDA Q1 2027 earnings call2026-05-20Transcript
In a complete end to end way, in a full stack way, But then we, of course, open the platform so that it could be integrated into all the different environments.
NVDA Q1 2027 earnings call2026-05-20Transcript
So we are gonna need a lot more CPUs and Vera was designed to be an agentic CPU.
NVDA Q1 2027 earnings call2026-05-20Transcript
So first of all, we should be growing faster than hyperscale CapEx.
NVDA Q1 2027 earnings call2026-05-20Transcript
NVIDIA is the practically the only company serving physical AI today.
NVDA Q1 2027 earnings call2026-05-20Transcript
The world is rebuilding computing for agentic AI and robotic physical AI.
NVDA Q1 2027 earnings call2026-05-20Transcript
But some environments just require an enterprise, for example, require a company who has all of the technologies working together so that they do not have to build it.
NVDA Q1 2027 earnings call2026-05-20Transcript
And we added Anthropic to our partnership this year.
NVDA Q1 2027 earnings call2026-05-20Transcript
Because everybody needs AI, and we are gonna see AI being adopted by every industry, every country, every company.
NVDA Q1 2027 earnings call2026-05-20Transcript
But then we disassemble it so that people could build and buy it in the configure they want and assemble it the way they like.
NVDA Q1 2027 earnings call2026-05-20Transcript
And then, of course, in the future, every single base station, every single radio network would become an AI powered radio network.
NVDA Q1 2027 earnings call2026-05-20Transcript
First of all, you are correct.
NVDA Q1 2027 earnings call2026-05-20Transcript
The world's first CPU purpose built for agentic AI.
NVDA Q1 2027 earnings call2026-05-20Transcript
And every major hyperscaler and system maker is partnering with us to deploy it.
NVDA Q1 2027 earnings call2026-05-20Transcript
This is the way computing is gonna work in the future.
NVDA Q1 2027 earnings call2026-05-20Transcript
With the addition of Anthropic to our existing partners, OpenAI, xAI, Meta, Gemini, and many others, our share of Frontier AI is growing.
NVDA Q1 2027 earnings call2026-05-20Transcript
So that remains exactly consistent with what I have said before.
NVDA Q1 2027 earnings call2026-05-20Transcript
Third, our full stack complete AI factory solution and vast global ecosystem let us uniquely address new AI data center segments new AI cloud natives new AI native clouds and sovereign AI clouds and on premises enterprise and industrial infrastructure.
NVDA Q1 2027 earnings call2026-05-20Transcript
And that segment all of the inference, 100% of that, the vast majority of that is NVIDIA.
NVDA Q1 2027 earnings call2026-05-20Transcript
Paratas AI trains their operating room assistant robot in NVIDIA Isaac Lab, multiplying their data with NVIDIA Cosmos World models.
YouTube2026-03-16Watch2:11:22
And so our cost per token, yeah, our cost per token is the lowest in the world.
YouTube2026-03-16Watch1:04:06
For 20 years, we've been dedicated to this architecture.
YouTube2026-03-16Watch6:12
The data center needed to become a single unit of computing.
YouTube2026-03-16Watch1:08:39
This is the new computer.
YouTube2026-03-16Watch1:53:22
Well, today I'm going to show you something of the future.
YouTube2026-03-16Watch13:29
One of the biggest investments that we made and we couldn't afford it at the time. and it consumed the vast majority of our company's profits was to take CUDA on the backs of GeForce to every single computer.
YouTube2026-03-16Watch11:43
And so optimizing for high throughput and optimizing for low latency are in fact enemies of each other.
YouTube2026-03-16Watch1:27:35
Your factory is limited no matter what.
YouTube2026-03-16Watch1:07:01
Every company in the world today needs to have an open claw strategy and a gentic system strategy.
YouTube2026-03-16Watch1:53:17
Um but the importance is profound.
YouTube2026-03-16Watch1:47:26
For robots, compute is data.
YouTube2026-03-16Watch2:10:23
What used to take what used to take 2 days to install now takes two hours.
YouTube2026-03-16Watch1:14:12
We've also been able to use MVFP4 for training.
YouTube2026-03-16Watch1:00:08
Today, IBM and NVIDIA are reinventing data processing for the era of AI by accelerating IBM Watson X.
YouTube2026-03-16Watch20:40
We have to understand the applications.
YouTube2026-03-16Watch31:14
You could talk to any modality you want.
YouTube2026-03-16Watch1:51:55
There's no question in my mind Nvidia systems are the lowest cost infrastructure you could get for AI infrastructure in the world.
YouTube2026-03-16Watch54:52
170 teraflops in one computer.
YouTube2026-03-16Watch1:08:16
Of course, as you know, 2025 was NVIDIA's year of inference.
YouTube2026-03-16Watch54:24
This is the age of physical AI and robotics.
YouTube2026-03-16Watch2:10:02
What if we disagregated inference altogether with a piece of software called Dynamo?
YouTube2026-03-16Watch1:31:23
Moore's law was about getting performance doubling every couple of years.
YouTube2026-03-16Watch23:07
Our relationship with cloud service providers are essentially us bringing customers to them.
YouTube2026-03-16Watch25:16
An AI that was able to perceive became an AI that could generate.
YouTube2026-03-16Watch49:51
This allows us to help every country build their sovereign AI.
YouTube2026-03-16Watch2:01:16
Some of it was AI surrogates, AI physical models and some of it was physical AI robotics models.
YouTube2026-03-16Watch42:05
All of that is done statically in advance and scheduled completely in software.
YouTube2026-03-16Watch1:30:17
We've demonstrated now that we can inference NVFP4 without loss of precision but gigantic boost in performance and energy efficiency.
YouTube2026-03-16Watch59:57
So with power constraints, every unused watt is revenue lost.
YouTube2026-03-16Watch1:43:10
We now know we could successfully autonomously drive cars.
YouTube2026-03-16Watch2:07:18
I said last year at this time that Nvidia's Grace Blackwell NVLink 72 was 35 times perf per watt.
YouTube2026-03-16Watch1:03:32
We recently added tiles so that we could help people program tensor cores and the structures of mathematics that are so foundational to artificial intelligence today.
YouTube2026-03-16Watch6:29
A one gawatt factory will never become two.
YouTube2026-03-16Watch1:01:18
Well, we take we take all of our software as I as I told you, we vertically integrate, but we horizontally open.
YouTube2026-03-16Watch1:05:33
Just like we created RTX for 3D graphics, we created QDF for data frames, structured data.
YouTube2026-03-16Watch19:12
This is where AI wants to go.
YouTube2026-03-16Watch1:23:51
NVIDIA accelerates data processing in the cloud.
YouTube2026-03-16Watch21:52
We offer you the software.
YouTube2026-03-16Watch31:54
And we could stand up these platforms in any country in any airgapped region completely on prem, completely on site, completely in the field.
YouTube2026-03-16Watch29:10
No single model can serve every industry.
YouTube2026-03-16Watch1:58:09
And we're willing to nurture, willing to support every single one of these GPUs in the world because they're all architecturally compatible.
YouTube2026-03-16Watch9:42
The installed base of CUDA is the reason why the flywheel is accelerating.
YouTube2026-03-16Watch7:33
More people could use it.
YouTube2026-03-16Watch52:10
Now, I don't know if you guys feel the same way, but $500 billion is an enormous amount of revenue.
YouTube2026-03-16Watch53:03
But in order for us to activate those computing platforms, we need to have domain specific libraries that solve very important problems in each one of the verticals that we address.
YouTube2026-03-16Watch33:42
He accused me of sandbagging.
YouTube2026-03-16Watch1:03:54
There's so many applications that you can run on Nvidia CUDA.
YouTube2026-03-16Watch8:33
Until now, this data has been completely useless to the world.
YouTube2026-03-16Watch18:23
And the reason for that is this is the first time in history that every single one of these companies needs compute and lots and lots of it.
YouTube2026-03-16Watch44:28
It's going to expand the reach, expand the compute of open AI.
YouTube2026-03-16Watch26:09
We are now at the beginning of a new platform shift.
YouTube2026-03-16Watch45:59
The free tier allows me to attract more customers.
YouTube2026-03-16Watch1:26:26
I am certain computing demand will be much higher than that.
YouTube2026-03-16Watch54:11
We accelerate scientific principled solvers of all different kinds.
YouTube2026-03-16Watch8:45
Today's CPU data processing systems can't keep up.
YouTube2026-03-16Watch21:00
Finally, AI is able to do productive work and therefore the inflection point of inference has arrived.
YouTube2026-03-16Watch50:36
In the moments that matter, tokens are already there.
YouTube2026-03-16Watch1:16
Accelerated computing for the era of AI.
YouTube2026-03-16Watch21:39
Well, om Omniverse Omniverse was designed to hold the world's digital twin starting from the earth and it's going to hold digital twins of all sizes.
YouTube2026-03-16Watch1:45:31
It exceeded it exceeded what Linux did in 30 years.
YouTube2026-03-16Watch1:47:44
This chart basically describes 100% of Nvidia's strategies.
YouTube2026-03-16Watch6:59
We have now proven that it is the only infrastructure in the world that you could go anywhere in the world and build with complete confidence.
YouTube2026-03-16Watch56:41
It's completely revolutionary.
YouTube2026-03-16Watch1:16:02
Our systems is another platform and now we have a new platform called AI factories.
YouTube2026-03-16Watch4:01
But most importantly, it's because we are not going to give up working on it.
YouTube2026-03-16Watch1:59:56
In 2020, DGXA100 Super Pod became the first GPU supercomput combining scale up and scale out architecture.
YouTube2026-03-16Watch1:08:49
And one of the areas, one of the one of the things I'm super excited about this year is we're going to bring open AI to AWS.
YouTube2026-03-16Watch25:55
You ask it to use tools, take your context, read files.
YouTube2026-03-16Watch49:37
We're the only one in production with it today.
YouTube2026-03-16Watch1:15:58
This latency, this interactivity requires enormous amount of bandwidth.
YouTube2026-03-16Watch1:27:20
Optics comes directly onto this chip, interfaces directly to silicon.
YouTube2026-03-16Watch1:15:41
It completely revolutionized artificial intelligence, caused a big bang of modern AI.
YouTube2026-03-16Watch38:22
This is the structured data, the ground truth of business.
YouTube2026-03-16Watch17:20
We have a 150 year old company who are now part of Nvidia supply chain and partnering with us either upstream or downstream.
YouTube2026-03-16Watch33:13
There's not one software engineer today who is not assisted by one or many AI agents helping them code.
YouTube2026-03-16Watch49:07
This is where I torture all of you.
YouTube2026-03-16Watch1:02:27
We're in production with the Gro chip and uh you know we'll ship it in the second half probably about Q3 time frame.
YouTube2026-03-16Watch1:33:51
We saw$500 billion dollars of very high confidence demand and purchase orders for Blackwell and Reuben through 2026.
YouTube2026-03-16Watch52:44
Blackwell redefined AI supercomputing system architecture with NVLink 72.
YouTube2026-03-16Watch1:09:29
We have now a world-class open agentic framework that all of us could use to build our open claw strategy.
YouTube2026-03-16Watch2:05:31
Everybody's looking for land, power, and shell.
YouTube2026-03-16Watch1:07:03
And today we are announcing four new partners for Nvidia's robo taxi ready platform.
YouTube2026-03-16Watch2:07:23
Together we take the next great leap into a bright new future built for all mankind.
YouTube2026-03-16Watch2:40
The higher the tier, the higher the quality, the higher the performance, the lower the volume, the lower the capacity.
YouTube2026-03-16Watch1:24:41
Most people would have thought we were their first supplier.
YouTube2026-03-16Watch28:08
NVIDIA is one of the largest contributors to open-source AI.
YouTube2026-03-16Watch1:58:29
But this is this is the house that GeForce made 25 years ago.
YouTube2026-03-16Watch11:09
And when all of that happens and we continuously update our software, the computing cost declines.
YouTube2026-03-16Watch9:17
So NVIDIA DSX is our new AI factory platform.
YouTube2026-03-16Watch1:46:05
Nvidia has three platforms.
YouTube2026-03-16Watch3:56
They work where human hands cannot.
YouTube2026-03-16Watch1:31
This is also the first time that the scale of the investments went from millions of dollars, tens of millions of dollars to hundreds of millions of dollars and billions of dollars.
YouTube2026-03-16Watch44:18
You couldn't find a GPU if you tried.
YouTube2026-03-16Watch50:19
In order to think, it has to inference.
YouTube2026-03-16Watch50:47
What you do this year will show up precisely next year as your revenues.
YouTube2026-03-16Watch1:21:12
We took some of the world's best security and computing experts and we worked with Peter to make open claw open claw enterprise secure and enterprise private capable.
YouTube2026-03-16Watch1:55:55
We've been working on CUDA for 20 years.
YouTube2026-03-16Watch6:04
16 GPUs connected with full alltoall bandwidth operating as one giant GPU.
YouTube2026-03-16Watch1:08:26
Every single software company of the future will be agentic and they will be token manufacturers.
YouTube2026-03-16Watch2:05:07
When the site goes live, the digital twin becomes the operator.
YouTube2026-03-16Watch1:44:42
450 companies sponsored this event.
YouTube2026-03-16Watch5:22
The install base is what attracts developers who then creates new algorithms that achieves a breakthrough.
YouTube2026-03-16Watch7:41
If most of your workload is high throughput, I would stick with just 100% Vera Rubin.
YouTube2026-03-16Watch1:29:19
We need a lot more capacity for CPO and that's the reason why we've been working with all of you to lay the foundation for this level of growth.
YouTube2026-03-16Watch1:40:13
With enough compute, developers everywhere are closing the physical AI data gap.
YouTube2026-03-16Watch2:11:15
It has taken us 20 years to now have built up hundreds of millions of GPUs and computing systems around the world that run CUDA.
YouTube2026-03-16Watch7:15
And at the tier that where your highest ASP and your most valuable segment, we increased it by 10x.
YouTube2026-03-16Watch1:25:38
We're vertical integration, horizontal open.
YouTube2026-03-16Watch1:05:37
It is through the connection of understanding of the algorithms with our computing platforms that we're able to open up to unlock these opportunities.
YouTube2026-03-16Watch42:26
Accelerated computing is not a systems problem.
YouTube2026-03-16Watch30:26
A thousand CUDA X libraries help developers make breakthroughs in every field of science and engineering.
YouTube2026-03-16Watch38:43
First of all, the adoption says something you know all in itself.
YouTube2026-03-16Watch1:52:45
And ultimately, the single hardest thing to achieve is the thing on the bottom, installed base.
YouTube2026-03-16Watch7:09
This is absolutely a new computing platform shift.
YouTube2026-03-16Watch58:45
This is the 20th anniversary of CUDA.
YouTube2026-03-16Watch6:00
We multiplied compute by 40 million.
YouTube2026-03-16Watch2:16:36
The longer you could use it, the lower the cost.
YouTube2026-03-16Watch54:50
AI needs rapid access to massive data sets.
YouTube2026-03-16Watch20:58
The way that it's built, the way it's manufactured, the way it's programmed completely changed.
YouTube2026-03-16Watch59:24
We serve just about every single industry.
YouTube2026-03-16Watch7:32
There are the three best models in the world.
YouTube2026-03-16Watch2:00:47
GeForce is Nvidia's greatest marketing campaign.
YouTube2026-03-16Watch10:37
Access sensitive information, execute code, communicate externally.
YouTube2026-03-16Watch1:55:35
There's no question in my mind there's a factor of two in here and a factor of two at the scale we're talking about is gigantic.
YouTube2026-03-16Watch1:42:30
And so the application reach is so great that once you install Nvidia GPUs, the useful life of it is incredibly high.
YouTube2026-03-16Watch8:50
It's going to pound on memory really hard.
YouTube2026-03-16Watch1:12:31
We're always backwards compatible.
YouTube2026-03-16Watch1:37:15
This is no different than any product that every company makes.
YouTube2026-03-16Watch1:24:39
We're in every computer company.
YouTube2026-03-16Watch7:26
It's a GPT moment for the bots from sim streets.
YouTube2026-03-16Watch2:18:05
This is how intelligence is made.
YouTube2026-03-16Watch0:08
This is already for sure going to be a multi-billion dollar business for us.
YouTube2026-03-16Watch1:16:29
And we're in volume production now.
YouTube2026-03-16Watch1:15:27
Open Claw is the number one.
YouTube2026-03-16Watch1:47:28
But most importantly, it also enables these infrastructures to have extraordinarily useful life.
YouTube2026-03-16Watch8:23
Structured data is the foundation of trustworthy AI.
YouTube2026-03-16Watch16:02
I'm quite proud of the fact that I explained AI clouds to Oracle for the first time and we were their first customer.
YouTube2026-03-16Watch28:14
It's able to agentically break down a problem, reason about it, reflect on it.
YouTube2026-03-16Watch49:40
The high throughput low speed could be used for the free tier.
YouTube2026-03-16Watch1:22:22
And so it is no different than any other business in the world.
YouTube2026-03-16Watch1:24:49
We all needed to have a Kubernetes strategy which made it possible for mobile cloud to happen.
YouTube2026-03-16Watch1:53:12
Neotron 3 is going to be followed by Neotron 4.
YouTube2026-03-16Watch2:00:06
The fusion, the fusion of 3D graphics and artificial intelligence.
YouTube2026-03-16Watch13:39
The only way for us to accelerate applications going forward and continue to bring tremendous speed up, tremendous cost reduction is through application or domain specific acceleration.
YouTube2026-03-16Watch30:43
One of the capabilities that we offer is confidential computing.
YouTube2026-03-16Watch27:11
The libraries is the crown jewels of our company.
YouTube2026-03-16Watch38:01
Nvidia's token cost is world class. basically untouchable at the moment.
YouTube2026-03-16Watch1:04:51
Data is the ground truth that gives AI context and meaning.
YouTube2026-03-16Watch20:52
And what you see here on the left on on this side on this side is tokens per watt.
YouTube2026-03-16Watch1:01:04
About 90% of what's generated every single year is unstructured data.
YouTube2026-03-16Watch18:19
The more throughput, the more tokens you could produce.
YouTube2026-03-16Watch1:02:57
Art openclaw has open sourced essentially the operating system of agent computers.
YouTube2026-03-16Watch1:52:20
We work and integrate Nvidia's technology into whatever platform you would like us to integrate into.
YouTube2026-03-16Watch31:48
We created a brand new CPU.
YouTube2026-03-16Watch1:13:13
And as the world continues to increase the amount of high-speed tokens it wants to generate with super smart tokens it wants to generate, the value of this integration is going to get even higher.
YouTube2026-03-16Watch1:30:38
25 years ago, we invented the programmable shader.
YouTube2026-03-16Watch11:19
Real world data will never be enough to train for every scenario.
YouTube2026-03-16Watch2:10:14
The storage system is going to get pounded.
YouTube2026-03-16Watch1:35:40
If a lot of your workload wants to be coding and very high valued engineering token generation, I would add Grock to it.
YouTube2026-03-16Watch1:29:24
To accelerate developers, Nvidia built open-source Isaac lab for robot training and evaluation and simulation.
YouTube2026-03-16Watch2:10:51
And every CEO, every CEO in the world will be tracking it.
YouTube2026-03-16Watch1:20:41
Computing used to be retrieval based now it's generative.
YouTube2026-03-16Watch47:10
Well, you know, we never we never thought we would be selling CPUs standalone.
YouTube2026-03-16Watch1:16:20
They are token factories.
YouTube2026-03-16Watch1:06:02
The first global rollout of physical AI at scale is here.
YouTube2026-03-16Watch2:09:18
We integrate all of our software and all of our technology, however we could package it up and integrate it into the world's inference service providers.
YouTube2026-03-16Watch1:05:40
And with Oberon, we could also use optical scale out or excuse me, optical scale up to expand to MVLink 576.
YouTube2026-03-16Watch1:37:31
We run Dynamo, this incredible operating system for AI factories on top of it.
YouTube2026-03-16Watch1:33:08
Most of these components never meet each other.
YouTube2026-03-16Watch1:41:07
Agents used to wait and see, now act autonomously.
YouTube2026-03-16Watch2:17:35
You want to put it in any of the clouds, we're delighted by that.
YouTube2026-03-16Watch56:48
And as a result, the content is beautiful, amazing, as well as controllable.
YouTube2026-03-16Watch15:45
But you will be studying the throughput and this token speed of your AI factories.
YouTube2026-03-16Watch1:20:51
Together, 35 times more throughput per megawatt.
YouTube2026-03-16Watch1:11:08
AI wants the tools to be as fast as possible.
YouTube2026-03-16Watch1:12:59
Today, we've reinvented computing.
YouTube2026-03-16Watch38:40
Tokens per watt is important because every data center every single factory by definition is power constrained.
YouTube2026-03-16Watch1:01:13
We have now set up a supply chain that could manufacture thousands a week of these systems essentially multi- gigawatts of AI factories per month inside our supply chain.
YouTube2026-03-16Watch1:34:42
Notice the smarter the AI, the lower your throughput.
YouTube2026-03-16Watch1:02:13
Nobody would have expected 35 times higher.
YouTube2026-03-16Watch1:03:30
Accelerated computing has a missing word.
YouTube2026-03-16Watch30:28
Accelerated computing is not a chip problem.
YouTube2026-03-16Watch30:20
Every single SAS company will become a gas company, an agentic as a service company.
YouTube2026-03-16Watch1:54:45
And maybe one day there'll be a premium model that allows you a premium service that allows you to generate token speeds that are incredibly high because you're in a critical path or maybe you're doing really long research and $150 per million tokens is just not a thing.
YouTube2026-03-16Watch1:23:17
And that that analysis is going to lead directly to your revenues.
YouTube2026-03-16Watch1:21:06
The diversity of AI is also its resilience.
YouTube2026-03-16Watch58:30
AI factories coming alive.
YouTube2026-03-16Watch2:16:12
It's now a factory to generate tokens.
YouTube2026-03-16Watch1:06:57
Nvidia is in full production with Spectrum X.
YouTube2026-03-16Watch1:16:04
On April 6th, 2016, a decade ago, we introduced DGX1, the world's first computer designed for deep learning.
YouTube2026-03-16Watch1:07:59
These data frames are giant spreadsheets and they hold all of life's information.
YouTube2026-03-16Watch17:15
20 years ago, we built CUDA, a single architecture for accelerated computing.
YouTube2026-03-16Watch38:36
This is the ground truth of enterprise computing.
YouTube2026-03-16Watch17:24
Just as GeForce brought AI to the world, AI is now going to go back and revolutionize how computer graphics is done all together.
YouTube2026-03-16Watch13:21
That infrastructure is going to get completely reinvented.
YouTube2026-03-16Watch36:39
And so that gives you a sense of how you would add Grock to Vera Rubin and extend its performance and extend its value even more.
YouTube2026-03-16Watch1:29:41
But the real world is massively diverse, unpredictable, full of edge cases.
YouTube2026-03-16Watch2:10:09
You give an AI agent a task, go to sleep, it runs 100 experiments overnight, keeping what works and killing what doesn't.
YouTube2026-03-16Watch1:49:58
This flywheel, this flywheel is now accelerating.
YouTube2026-03-16Watch8:01
If we release a new optimization, it benefits millions.
YouTube2026-03-16Watch9:53
Meanwhile, as we continue to nurture and continue to update software over its life, not only do you get the first time pop, you get the continuous cost reduction of accelerated computing over time.
YouTube2026-03-16Watch9:31
Well, Moore's law has run out of steam.
YouTube2026-03-16Watch23:23
Just as Linux gave the industry exactly what it needed at exactly the time just as Kubernetes showed up at exactly the right time just as HTML showed up it made it possible for the entire industry to grab onto this open-source stack and go do something with it.
YouTube2026-03-16Watch1:55:01
The other 40% is just everywhere.
YouTube2026-03-16Watch58:09
And so this is the incredible power of extreme code design.
YouTube2026-03-16Watch1:06:41
Nearly 3 million open models across language, vision, biology, physics, and autonomous systems enable AI builds for specialized domains.
YouTube2026-03-16Watch1:58:19
We also operated, connected to the grid so that we could interact with each other, send each other information so that we could adjust grid power and data center power accordingly, saving energy.
YouTube2026-03-16Watch1:41:54
And um this is our moment.
YouTube2026-03-16Watch2:03:46
We need data generated from AI and simulation.
YouTube2026-03-16Watch2:10:21
I know why you're not impressed.
YouTube2026-03-16Watch53:22
Computers don't like extreme amount of flops, extreme amount of bandwidth because there's only so much surface area for chips that any systems has.
YouTube2026-03-16Watch1:27:26
This is the brand new Gro system.
YouTube2026-03-16Watch1:15:09
Nvidia is a vertically integrated computing company with open horizontal integration with the world.
YouTube2026-03-16Watch42:37
These are some of the largest companies in the world.
YouTube2026-03-16Watch43:06
Now open claw has made it possible for us to create personal agents.
YouTube2026-03-16Watch1:52:36
You could wave at it and it understands you.
YouTube2026-03-16Watch1:51:53
Open Models is one of the largest and most diverse AI ecosystems in the world.
YouTube2026-03-16Watch1:58:15
Agentic systems in the corporate network can have access to sensitive information.
YouTube2026-03-16Watch1:55:21
The next computing platform has arrived.
YouTube2026-03-16Watch21:38
We're willing to do so because the install base is so large.
YouTube2026-03-16Watch9:51
Tokens are the new commodity and like all commodities once it reaches an inflection once it becomes mature or becomes maturing it will segment into different parts.
YouTube2026-03-16Watch1:22:12
The span of reach of AI is its resilience.
YouTube2026-03-16Watch58:36
You want to put it on prem, we're happy about that.
YouTube2026-03-16Watch56:52
It could spawn off and call upon other sub aents.
YouTube2026-03-16Watch1:51:45
Everything was completely simulated.
YouTube2026-03-16Watch42:20
I showed you the fusion of generative AI with computer graphics and it brought computer graphics to life.
YouTube2026-03-16Watch46:37
We completely rearchitected the system, disagregated the computing system alto together and created MVLink 72.
YouTube2026-03-16Watch59:16
This is the largest most comprehensive sweep of AI that has AI inference that has ever been done.
YouTube2026-03-16Watch1:00:58
I would add Grock to maybe 25% of my total data center.
YouTube2026-03-16Watch1:29:34
There's so many humanoid robots here, but one of my favorites, one of my favorites is a Disney robot.
YouTube2026-03-16Watch2:08:55
But we also have been working on physically embodied agents for a long time.
YouTube2026-03-16Watch2:06:20
This is your factory in the future.
YouTube2026-03-16Watch1:07:42
We integrate with your technology so that we can bring accelerated computing to everybody in the world.
YouTube2026-03-16Watch31:57
01 made generative AI trustworthy and grounded on truth.
YouTube2026-03-16Watch47:59
Emerald AI agents interpret live grid demand and stress signals and adjust power dynamically.
YouTube2026-03-16Watch1:45:05
The inflection, the inference inflection has arrived.
YouTube2026-03-16Watch52:20
With accelerated Watson X data running on Nvidia GPUs, Nestle can run the same workload five times faster at 83% lower cost.
YouTube2026-03-16Watch21:24
We even built a supercomputer to help us optimize kernels and help us optimize our complete stack.
YouTube2026-03-16Watch1:00:19
You want to put it in any country, anywhere, we're delighted to support you.
YouTube2026-03-16Watch56:48
Well, now we're going to have AI use structured data and we better accelerate the living daylights out of it.
YouTube2026-03-16Watch17:26
We just don't have enough bandwidth.
YouTube2026-03-16Watch1:28:48
As we continue to optimize the algorithms and because our our reach is so large and our install base is so large we can reduce the computing cost increasing the scale increasing the speed for everybody continuously.
YouTube2026-03-16Watch23:38
With DSX, Nvidia and our ecosystem of partners are racing to build AI infrastructure around the world, ensuring extreme resiliency, efficiency, and throughput.
YouTube2026-03-16Watch1:45:15
This allows me to serve my most valuable customers.
YouTube2026-03-16Watch1:26:30
The faster you can inference, the faster you could of course respond.
YouTube2026-03-16Watch1:01:49
Very quickly, Nvidia went from a chip company to a AI factory company or AI infrastructure company, AI computing company.
YouTube2026-03-16Watch1:40:43
And so we're going to crank out these these Vera Rubin racks while we're cranking out the GB300 racks.
YouTube2026-03-16Watch1:34:55
We've landed a whole bunch of our partners there.
YouTube2026-03-16Watch28:24
This is the beginning of something very, very big.
YouTube2026-03-16Watch33:35
The number of robo taxi ready cars in the future are going to be incredible.
YouTube2026-03-16Watch2:07:49
We are integrated with working with literally every single company that we know of building robots.
YouTube2026-03-16Watch36:14
I could totally imagine in the future every single engineer in our company will need an annual token budget.
YouTube2026-03-16Watch2:04:15
NVIDIA open models give researchers and developers the foundation to build and deploy AI for their own specialized domains.
YouTube2026-03-16Watch1:59:37
I really love what my stuff enables that person to do.
YouTube2026-03-16Watch1:50:16
And with NVIDIA Alpamo, vehicles now have reasoning, helping them operate safely and intelligently across scenarios.
YouTube2026-03-16Watch2:09:25
It stands to reason there's going to be a whole new crop of really important companies, consequential companies for the future of the world.
YouTube2026-03-16Watch45:42
There's so much power that is squandered in these AI factories.
YouTube2026-03-16Watch1:40:56
Speaking of agents, agents as you know perceive, reason and act.
YouTube2026-03-16Watch2:06:04
Unlike Reuben that slides in horizontally, Ruben Ultra goes into a whole new rack.
YouTube2026-03-16Watch1:18:03
MVLink 72 is so gamechanging.
YouTube2026-03-16Watch1:28:22
Tokens are harnessing a new wave of clean energy and unlocking the secrets of the stars.
YouTube2026-03-16Watch0:35
We want to make sure that these AI factories come together designed in the best possible way.
YouTube2026-03-16Watch1:41:03
We are now a computing platform that runs all of AI.
YouTube2026-03-16Watch56:57
We need a lot more capacity for optics.
YouTube2026-03-16Watch1:40:13
But once upon AI time training was the paradigm.
YouTube2026-03-16Watch2:16:51
Most of the agents in the world today that I've spoken about are digital agents.
YouTube2026-03-16Watch2:06:07
We're on to something here.
YouTube2026-03-16Watch33:35
We’re building the most impactful technology the world has ever known,
time.com2025-12-16Source
grateful for everything that we've done to help him
time.com2025-12-16Source
I have a pretty great life,
time.com2025-12-16Source
He was just very, very, happy.
time.com2025-12-16Source

In the News

Every quote on this page passed an independent check against its original source before publication. These are short excerpts, bibliographic citations, and headline links only — follow each link to read the full article, watch the full video, or find the full work.