Jensen Huang, Founder and CEO of NVIDIA

Stanford Graduate School of Business
5 Mar 202456:27

Summary

TLDRIn this inspiring interview, Jensen Huang, CEO of NVIDIA, shares his remarkable journey of transforming a startup into a pioneering tech giant. He reveals the core principles that have guided his leadership, emphasizing the importance of empowering employees, fostering a culture of innovation, and staying true to one's beliefs. Huang reflects on overcoming challenging times, such as the 80% market cap loss during the financial crisis, and highlights the role of 'early indicators of future success' in driving the company forward. He discusses the disruptive potential of generative AI and encourages a first-principles approach to harnessing its power responsibly. Throughout the interview, Huang's wisdom and passion for pushing the boundaries of computing shine through, inspiring viewers to redefine their own tomorrows.

Takeaways

  • 😀 Have a strong core belief and vision, but be willing to adapt and reinvent along the way based on new information and opportunities.
  • 🤖 Aggressively pursue new technologies and applications that can solve problems current general computing cannot address, even if the market doesn't exist yet.
  • 📚 Continuously learn and study from first principles - treat challenges as opportunities to gain new knowledge and insights.
  • 🏆 Seek out 'early indicators of future success' rather than getting fixated on short-term financial metrics.
  • 🧠 Empower employees by being transparent, informing them of the broader context, and showing how you reason through problems.
  • 🌳 Design the organizational structure from first principles based on what you are trying to build, not legacy hierarchical models.
  • 🔭 Look backwards from the future you envision for your company and work backwards to get there.
  • 🚀 Embrace generative AI as a paradigm shift that will fundamentally change computing and disrupts industries.
  • ⚖️ Support tailored regulation for AI products/services by existing bodies, not blanket regulation across domains.
  • 👥 Surround yourself with people you love and take them on the journey with your core beliefs as the guiding force.

Q & A

  • What motivated Jensen Huang to leave LSI Logic and become a founder?

    -Jensen was motivated by the opportunity to build a company focused on solving problems that normal computers powered by general-purpose computing couldn't, as suggested by his colleagues Chris and Curtis.

  • How did Nvidia choose its first killer app and what was it?

    -Nvidia's first killer app was decided to be 3D graphics, targeting the application of video games. This decision was based on the intersection of the challenge of making 3D graphics affordable and the non-existent video game market for PCs at the time.

  • What was Nvidia's approach to overcoming the challenge of their technology becoming incompatible with Microsoft's direct 3D?

    -Nvidia reset the company by adopting the OpenGL standard for computer graphics after their initial technology became incompatible with Microsoft's direct 3D, allowing them to innovate and create something the world had never seen before.

  • How did Jensen Huang ensure Nvidia stayed on track during challenging times, such as when its market cap dropped 80%?

    -Jensen focused on core beliefs and the fundamentals of the business, maintaining the same approach to work and leadership, and ensuring the company remained focused on its mission despite external market pressures.

  • What role does generative AI play in Nvidia's vision for the future?

    -Generative AI is seen as a revolutionary technology that will change the nature of computing, enabling software to understand and generate new content, which will have broad implications for various applications and industries.

  • What were the key principles behind the creation of Nvidia's organizational structure?

    -Nvidia's organizational structure is designed to be flat, prioritizing empowerment and information accessibility across all levels to ensure every employee can contribute effectively to the company's mission.

  • How does Jensen Huang view the potential challenges and regulations surrounding AI?

    -While acknowledging the need for technological advancements to address safety and ethical considerations, Jensen emphasizes the importance of context-specific regulations across different industries, rather than overarching regulations that might stifle innovation.

  • What was Nvidia's strategy for entering and growing the gaming market for PCs?

    -Nvidia focused on creating technology and markets simultaneously, recognizing the need to both innovate in 3D graphics technology and actively develop the gaming market for PCs to ensure the company's success.

  • How did Nvidia adapt when realizing their initial vision wasn't working as expected?

    -Nvidia was flexible in its approach, shifting focus from their initial vision to explore and capitalize on the emerging market of 3D graphics for gaming, demonstrating the company's ability to pivot based on market demands and technological possibilities.

  • What does Jensen Huang consider as the most important aspect of his leadership role at Nvidia?

    -Jensen Huang believes his most important role is to create the conditions that allow his employees to do their life's work, focusing on empowerment, information sharing, and fostering an environment where everyone can contribute to Nvidia's mission.

Outlines

00:00

👨‍💼 Founding Nvidia and Securing Initial Funding

Jensen describes how he and his co-founders Curtis Priem and Chris Malachowsky decided to start Nvidia during the microprocessor revolution in the early 1990s. Their goal was to build a company that could solve problems that normal computers couldn't handle. Jensen shares the story of how he secured funding from Don Valentine, a renowned Silicon Valley investor, with the help of his former boss at LSI Logic, Wilfred Corrigan.

05:00

🎮 Creating the 3D Gaming Market for PCs

Jensen explains how Nvidia's initial focus was on creating the technology and market for 3D graphics in video games on PCs. This involved developing affordable hardware and software solutions to enable 3D graphics, which were previously only available on expensive workstations. Nvidia had to create both the technology and the market, working with companies like Electronic Arts to drive adoption.

10:02

💡 Pivoting and Implementing the OpenGL Pipeline

Jensen discusses how Nvidia had to pivot and reset the company when their initial 3D graphics technology became incompatible with Microsoft's Direct3D standard. They implemented the OpenGL pipeline by studying a textbook on how Silicon Graphics did computer graphics. This pivotal moment gave the company confidence that they could succeed by learning and reinventing from first principles.

15:03

📈 Expanding into New Markets and Applications

Jensen talks about Nvidia's efforts to expand into new markets and applications beyond 3D graphics, such as computational drug design, weather simulation, and artificial intelligence. He highlights the importance of identifying 'early indicators of future success' to guide the company's investments in emerging technologies, even when the markets don't yet exist.

20:04

🤖 The Impact of Deep Learning and AI

Jensen discusses how Nvidia recognized the potential of deep learning early on and worked with researchers to develop tools and languages for expressing neural network algorithms on their processors. He emphasizes the importance of reasoning from first principles to understand the implications of new technologies like AI and how they could fundamentally change computing and software development.

25:05

💼 Leadership During Challenging Times

Jensen shares his approach to leading Nvidia through challenging times, such as when the company lost 80% of its market cap during the financial crisis. He emphasizes the importance of gut-checking core beliefs, focusing on the company's mission, and being transparent with employees to keep them motivated and empowered.

30:06

🗺️ Creating a Flat and Empowered Organization

Jensen explains his philosophy behind creating a flat and highly empowered organization at Nvidia. He believes in minimizing layers of information flow and trusting employees with important information to enable them to make smart decisions. Jensen's goal is to create conditions where employees can do their life's work and contribute to the company's mission.

35:06

🔮 The Future of Generative AI and Computing

Jensen discusses the implications of generative AI and its ability to understand and generate content across different modalities (text, images, speech, etc.). He explains how this technology could fundamentally change the way we process information, write software, and approach computing. Jensen encourages reasoning from first principles to understand how different industries and applications might be disrupted by generative AI.

40:08

🌍 Redefining Tomorrow and Nvidia's Contribution

When asked about redefining tomorrow, Jensen emphasizes the importance of making a unique contribution to advancing the future of computing, which he believes is the most important instrument for humanity. He envisions Nvidia being remembered as a company that changed everything through its work in creating new technologies and enabling new markets.

45:10

🤖 Addressing Concerns About AI Development Pace and Regulation

Jensen acknowledges the concerns about the pace of AI development and the need for regulation. He emphasizes the importance of advancing technologies for grounding, safety, and security to mitigate risks. Jensen advocates for product and service-specific regulations rather than an overarching 'super regulation' and highlights the need for social implications of AI to be addressed.

50:11

🤵 Rapid-Fire Questions and Advice

Jensen responds to a series of rapid-fire questions, including recalling his fondest memory from working at Denny's, what he would wear if there were a shortage of black leather jackets, and what he would title a hypothetical textbook. As a parting piece of advice, he encourages having a core belief, gut-checking it every day, pursuing it with all one's might, surrounding oneself with loved ones, and taking them on that journey.

55:13

🙏 Closing Remarks

The session concludes with Jensen expressing gratitude for the opportunity to share his insights and experiences.

Mindmap

Keywords

💡Founder

The term 'Founder' refers to an individual or group of individuals who establish a company or startup. In the video, Jensen discusses his journey from working at LSI Logic to becoming a founder of Nvidia alongside Chris and Curtis. This transition highlights the leap from being an employee to taking on the entrepreneurial challenge of building a new company, motivated by the desire to solve unique problems that conventional computing couldn't address at the time.

💡Accelerated Computing

Accelerated Computing refers to the use of specialized hardware to perform computational tasks more efficiently than traditional CPUs. The video narrative describes Nvidia's focus on accelerated computing as a cornerstone for its mission, aiming to build computers that solve problems normal computers can't. Jensen discusses the early days of the PC revolution and how the company decided to contribute to it by focusing on areas like computational drug design, weather simulation, and AI, highlighting the foundational goal of Nvidia.

💡3D Graphics

3D Graphics is used in the context of Nvidia's initial killer application, identified by Jensen and his team as the technology focus for their company. This choice was driven by the challenge and opportunity of making high-cost image generation technologies, like those used in workstations, more accessible for the video game market. The narrative emphasizes how this decision shaped Nvidia's early direction and success, underlining the importance of selecting impactful applications for emerging technologies.

💡Microprocessor Revolution

The Microprocessor Revolution refers to a pivotal period in computing history characterized by rapid advancements in microprocessor technology, leading to the widespread adoption of personal computers. Jensen recounts Nvidia's inception during this transformative era, highlighting the environment of innovation and the anticipation of significant change, such as the launch of Windows 95 and the Pentium processor, that Nvidia was born into and aimed to contribute to through accelerated computing and graphics.

💡Entrepreneurship

Entrepreneurship in this context refers to the process of starting and building a new business venture to innovate and solve problems. Jensen's narrative embodies the entrepreneurial spirit, from taking the risk of leaving a stable job to tackle the uncertainties of founding Nvidia, to navigating the challenges of identifying markets and technologies that could leverage the company's strengths. His story illustrates the essence of entrepreneurship - vision, risk-taking, and perseverance.

💡Computational Drug Design

Computational Drug Design is mentioned as one of the problems Nvidia set out to solve with its technology. This field involves using computer algorithms to discover, enhance, or study drugs and biological systems. Jensen's inclusion of this area underlines Nvidia's broader mission to impact sectors beyond traditional computing, demonstrating the company's ambition to contribute to advancing scientific research and healthcare through accelerated computing and AI.

💡AI (Artificial Intelligence)

AI is a central theme in Jensen's discussion, reflecting Nvidia's pivot and significant contribution to the field through GPU acceleration. Jensen discusses AI in the context of Nvidia driving down computational costs, enabling new software development paradigms where computers can generate code, leading to advancements in autonomous vehicles, robotics, and more. The mention of AI showcases Nvidia's role in the evolution of computing from graphics to general computational acceleration, facilitating groundbreaking advancements in technology.

💡Business Plan

The Business Plan segment offers insights into Jensen's approach to entrepreneurship and the practical challenges of starting a company. He humorously recounts his attempt to learn how to write a business plan from a book, only to realize the urgency and complexity of his endeavor required a more direct approach. This anecdote serves to underscore the hands-on, learn-as-you-go aspect of building a startup, emphasizing the importance of adaptability and vision over formal planning in the early stages.

💡Innovation

Innovation is a key concept throughout the video, reflecting Nvidia's commitment to pioneering new technologies and markets. Jensen details how Nvidia's willingness to tackle hard problems, like making 3D graphics accessible or venturing into AI and deep learning, showcases the company's innovation DNA. The story of Nvidia's adaptation and success in multiple technology revolutions exemplifies how continuous innovation is crucial for staying relevant and leading in the tech industry.

💡Ecosystem

Ecosystem in this context refers to the comprehensive range of Nvidia's technologies, software, and platforms that together enable various applications in computing, AI, gaming, and more. Jensen's discussion about Nvidia creating not just technology but markets emphasizes the importance of building an ecosystem where different elements complement and enhance each other, facilitating broader adoption and enabling new capabilities across industries.

Highlights

Jensen Huang explains his motivation for co-founding NVIDIA, driven by the belief that they could create a computer that solves problems regular CPUs couldn't handle and the aspiration to work on impactful projects that would make the world better.

Huang emphasizes the importance of finding early indicators of future success when investing in a long-term vision, even when the market doesn't exist yet. He cites examples like the first fuzzy cat image from deep learning research as an early sign of its potential impact.

During challenging times like NVIDIA's 80% market cap drop, Huang advocates for going back to core beliefs, gut-checking assumptions, and continuing the work if those foundational principles remain unchanged.

Huang attributes NVIDIA's success to creating a highly empowered organization where employees understand the context and can make smart decisions by being transparent with information and trusting them to handle complex situations.

Huang explains how generative AI represents a fundamental shift in computing, moving from retrieval-based models to generation-based models, and the implications this will have on industries, applications, and how we think about computing overall.

Huang advocates for going back to first principles and reasoning through the core implications of new technologies like generative AI, rather than being constrained by existing organizational structures or ways of thinking.

Huang believes that regulations for AI should be focused on products and services, leveraging existing regulatory bodies like the FDA and FAA, rather than creating overarching, cross-cutting regulations that may impede progress.

Huang advises having a core belief, gut-checking it every day, pursuing it with all your might for a very long time, and surrounding yourself with people you love on that journey.

Huang emphasizes the importance of creating the conditions for employees to do their life's work, which involves empowering them with information, transparency, and trust, rather than limiting information flow.

Huang highlights the value of learning from textbooks and archive papers, going back to first principles, and reinventing solutions based on current conditions and motivations, rather than incrementally building on past approaches.

Huang discusses the early decision to focus on 3D graphics and video games as the initial killer app for NVIDIA, despite the technology being expensive and the market being non-existent at the time, driven by the belief in the importance of the work.

Huang emphasizes the importance of making a unique contribution that advances the field of computing, which he sees as the most important instrument of humanity, and being remembered for doing something insanely hard that changed everything.

Huang advises against being constrained by existing organizational structures or hierarchies, and instead designing organizations based on first principles, considering the inputs, outputs, and environmental conditions the company operates in.

Huang shares his approach of looking forward in time and reasoning backwards, imagining the impact NVIDIA will have made and working backwards to achieve that desired end result, rather than looking forward from the present.

Huang highlights the importance of surrounding himself with genius-level talent across all domains, acknowledging that NVIDIA has the deepest technology management team the world has ever seen.

Transcripts

play00:00

[Music]

play00:04

Jensen this is such an honor thank you

play00:06

for being here I'm delighted to be here

play00:08

thank you in honor of your return to

play00:10

Stanford I decided we'd start talking

play00:12

about the time when you first left you

play00:15

joined LSI logic and that was one of the

play00:18

most exciting companies at the time

play00:20

you're building a phenomenal reputation

play00:22

with some of the biggest names in Tech

play00:24

and yet you decide to leave to become a

play00:26

Founder what motivated

play00:28

you uh uh Chris and Curtis Chris and

play00:32

Curtis uh uh I was an engineer at LS

play00:35

logic and Chris and Curtis were at Sun

play00:38

and I was working with with uh some of

play00:41

the brightest Minds in computer science

play00:43

at the time of all time uh including

play00:46

andyto shim and others uh building

play00:48

building workstations and Graphics

play00:50

workstations and so on so forth and uh

play00:54

Chris and Curtis uh uh said one day that

play00:58

they like to leave some son and they

play01:01

like uh me to go figure out what they're

play01:05

going to go leave

play01:06

four and and um I had a great

play01:10

job but they they insisted that I uh

play01:13

figure out you know with them how to how

play01:16

to build a company and so so we hung out

play01:18

at Denny when whenever they Dro by and

play01:21

and uh uh which was which is by the way

play01:23

my alma marter my my first

play01:26

company uh you know my first job before

play01:29

for before CEO was a was a dishwasher

play01:32

and so and and I did that very

play01:35

well and and so anyways uh we got

play01:38

together and and we we DEC and it was

play01:40

during the the microprocessor Revolution

play01:42

this is 1993 and and 1992 when we were

play01:45

getting together the PC Revolution was

play01:48

just getting going you you know that

play01:50

Windows 95 obviously which is the

play01:52

Revolutionary version of Windows uh

play01:54

didn't even come to the market yet and

play01:56

Pentium wasn't even announced yet and so

play01:59

and this is this is all before the right

play02:01

before the PC Revolution and it was it

play02:03

was pretty clear that that uh the

play02:05

microprocessor was going to be very

play02:07

important and we we thought you know why

play02:10

don't we build a company uh to go solve

play02:13

problems that a normal computer that is

play02:16

powered by general purpose Computing

play02:18

can't and and so that that became the

play02:20

company's Mission uh to go to go build a

play02:24

computer uh the type of computers and

play02:26

solve problems that normal computers

play02:28

can't and to this day uh we're focusing

play02:30

on that and if you look at all the the

play02:32

problems that that um and the markets

play02:34

that we opened up as a result uh it's

play02:37

you know things like uh computational

play02:39

drug design um uh weather simulation

play02:42

materials design these are all things

play02:44

that we're really really proud of uh

play02:46

robotics uh self-driving cars uh

play02:49

autonomous autonomous uh software we

play02:52

call artificial intelligence and then

play02:54

all you know of course uh we uh we drove

play02:57

the the uh U the techn techology so hard

play03:01

that that eventually the computational

play03:03

cost uh uh went to approximately zero

play03:07

and then enabled enabled a whole new way

play03:09

of developing software where the

play03:10

computer wrote the software itself

play03:12

artificial intelligence as we know it

play03:13

today and so so I that was that was it

play03:16

that was the journey yeah thank you all

play03:18

for

play03:19

[Laughter]

play03:21

coming well these applications are on

play03:24

all of our minds today but back then the

play03:28

CEO of LSI logic

play03:30

convinced his biggest investor Don

play03:32

Valentine to meet with you he is

play03:34

obviously the founder of seoa yeah now I

play03:36

can see a lot of Founders here edging

play03:38

forward in anticipation but how did you

play03:41

convince the most sought-after investor

play03:43

in Silicon Valley to invest in a team of

play03:46

firsttime Founders building a new

play03:48

product for a market that doesn't even

play03:50

exist I I didn't know how to write a

play03:53

business

play03:54

plan and and uh uh so I went to a went

play03:58

to a book

play04:00

bookstore and back then there were

play04:02

bookstores and and and um in the

play04:05

business book section there was this

play04:06

book and it was written by somebody I

play04:08

knew Gordon Bell and this book I should

play04:11

go find it again but it's a very large

play04:13

book and the book says how to write a

play04:16

business

play04:17

plan and and that was you know a highly

play04:20

specific title for a very niche market

play04:24

and it seems like he wrote it for like

play04:25

you know 14 people and I was one of them

play04:28

and and so I I bought the book I I

play04:31

should have known right away that that

play04:32

it was a bad idea because that you know

play04:35

Gordon is super super smart and super

play04:38

smart people have a lot to say and and

play04:41

they wanted you know and I I'm pretty

play04:43

sure Gordon wants to teach me how to

play04:45

write a business plan uh completely and

play04:48

so I I picked up this book it's like 450

play04:50

pages long well I never got through it

play04:54

not even close I I flipped through it a

play04:56

few pages and I go you know what by the

play04:58

time I'm done reading this this thing

play05:00

I'll be out of business I'll be out of

play05:02

money and and uh Lori and I only had

play05:05

about 6 months uh in the bank and we had

play05:08

already Spencer Madison and and uh and a

play05:10

dog and so the five of us had to live

play05:12

off of you know uh whatever money we had

play05:15

in the bank and and so I didn't have

play05:17

much time uh and so instead of writing

play05:20

the business plan uh I just went to talk

play05:22

to to W Coran he turn he called me one

play05:25

day and said hey you know you left the

play05:26

company you didn't even tell me what you

play05:27

were doing I want you to come back and

play05:29

explain it to me and so I went back and

play05:31

I explained it to Wi and wi wi at the

play05:33

end of it he he said I have no idea what

play05:36

you

play05:38

said and and um that's one of the worst

play05:42

elevator pitches I've ever

play05:45

heard um and then he picked up the phone

play05:48

and he called Don Valentine and he he

play05:51

called Don and he says Don I want you to

play05:53

give I'm going to send a kid over I want

play05:55

you to give him

play05:56

money he's one of the best employees l

play05:59

logic ever ever had and um I and and so

play06:04

the thing I learned is is

play06:07

uh uh you you can make up a great

play06:11

interview you could even have a bad

play06:13

interview but you can't run away from

play06:15

your past and so have a good past you

play06:19

know try to have a good past and and and

play06:21

in a lot of ways I was serious when I

play06:23

said I was a good dishwasher I was

play06:25

probably Denny's best

play06:27

dishwasher um I I planned my work I was

play06:31

organized you know I was Misan plus and

play06:34

then I washed The Living Daylights out

play06:35

of the dishes and then and then you know

play06:38

they promoted me to bus I was certain

play06:40

I'm the best bus boy Denny's ever had

play06:43

you know I was I never left a station

play06:45

with empty-handed I never came back

play06:47

empty-handed I was very efficient and

play06:49

then they and so anyways eventually I

play06:52

became you know a CEO I'm working I'm

play06:55

still working on being being a good CEO

play06:57

but you talk about being the bad you

play07:00

needed to be the best among 89 other

play07:03

companies that were funded after you to

play07:05

build the same thing and then with 6 to9

play07:08

months of Runway left you realized that

play07:11

the initial Vision was just not going to

play07:13

work MH how did you decide what to do

play07:16

next to save the company when the cards

play07:18

were so stacked against

play07:19

you well we started uh this company

play07:22

called for Accelerated Computing and the

play07:24

question is what is it for what's the

play07:26

killer app and and uh that was that that

play07:31

came our first great

play07:32

decision um and this is what sequa

play07:36

funded the first great decision was the

play07:39

first killer app was going to be 3D

play07:41

graphics and the the the technology was

play07:45

going to be 3D graphics and the

play07:46

application was going to be video games

play07:49

at the Time 3D Graphics was impossible

play07:51

to make cheap it was Million

play07:53

dooll image generators from Silicon

play07:57

graphics and the video and so it was a

play07:59

million dollars and and it's hard to

play08:01

make cheap um and the video game Market

play08:04

was0 billion doar so you have this

play08:07

incredible technology that's hard to uh

play08:11

commoditize and commercialize and then

play08:14

you have this Market that doesn't exist

play08:16

that was that intersection was the

play08:17

founding of our company and and I still

play08:20

remember uh when when Don at the end of

play08:23

my

play08:24

presentation uh you know Don was still

play08:27

kind of he he said you know know one of

play08:29

the things he said to me which made a

play08:31

lot of sense back then makes a lot of

play08:33

sense today he says startups don't

play08:36

invest in startups or startups don't

play08:38

partner with startups and his point is

play08:41

that in order for NVIDIA to succeed we

play08:44

needed another startup to succeed and

play08:47

that other startup was Electronic

play08:49

Arts and and then on the way out he he

play08:52

reminded me that electronic arts's

play08:55

CTO is 14 years old and had to be driven

play08:59

to work by his

play09:01

mom and he just wanted to remind me that

play09:03

that's who I'm relying on that that and

play09:08

then and uh and then after that he said

play09:11

if you lose my money I'll kill you and

play09:12

that that was that was kind of my

play09:14

memories of that first

play09:16

meeting uh but nonetheless uh we created

play09:19

we created something uh we went on uh

play09:22

the next several years to go create the

play09:24

market to create the gaming market for

play09:27

PCs and it took a long time to do so

play09:30

we're still doing it today uh we realize

play09:32

that not only do you have to create the

play09:35

technology and uh invent a new way of

play09:37

doing computer Graphics so that what was

play09:39

a million dollars is now you know three

play09:41

400 $500 um that fits in the computer

play09:45

and you have to go create this new

play09:47

market so we have to create technology

play09:48

create markets the idea that a company

play09:51

would create technology create markets

play09:53

defines Nvidia today almost everything

play09:56

we do we create technology we create

play09:58

markets that's that's the reason why

play09:59

people say we have a you know people

play10:01

call it a stack an ecosystem words like

play10:04

that um but that's basically it at the

play10:06

core for 30 years what Nvidia realized

play10:09

we had to do is in order to uh create

play10:13

the conditions by which somebody could

play10:14

buy our products we had to go invent

play10:16

this new market and uh it's the reason

play10:18

why we were early in autonomous driving

play10:20

it was the reason why we're early in

play10:22

deep learning it was the reason why

play10:23

we're early and just about all these

play10:25

things including uh computational drug

play10:27

disc drug design and and Discovery um

play10:30

all these different areas we're trying

play10:32

to create the market while we're

play10:33

creating the technology and so that

play10:35

that's um uh okay and then we got we got

play10:38

going and and then and then um Microsoft

play10:42

introduced uh a standard called direct

play10:45

3D and that spawned off hundreds of

play10:48

companies and we found ourselves a

play10:51

couple years later competing with just

play10:52

about everybody and and the thing that

play10:54

that we invented the company the

play10:56

technology we invented uh 3D graphics

play10:59

with the consumerized 3D with turns out

play11:01

to be incompatible with direct 3D so we

play11:04

started this company we had this 3D

play11:06

Graphics thing we million-dollar thing

play11:07

we're trying to make it consumerized and

play11:09

so we invented all this technology and

play11:12

then shortly after it became

play11:13

incompatible and um uh so we had to

play11:16

reset the company uh or go out of

play11:18

business but we didn't know how to we

play11:20

didn't know how to build it the way that

play11:22

Microsoft had defined it and um and I

play11:26

remember I remember a meeting at at you

play11:29

know on a weekend and the conversation

play11:31

was you know we now have 89

play11:34

competitors uh I understand that the way

play11:38

we do it is not not right but we don't

play11:40

know how to do it the right

play11:42

way and and um thankfully there was

play11:47

another bookstore and

play11:50

um and the bookstore is called fries

play11:53

Fries electronics I don't think I don't

play11:55

know if it's still here um and so I had

play11:58

I had I had um I I I think I drove madis

play12:02

and my daughter on a weekend to fries

play12:04

and and it was sitting right

play12:08

there the openg manual uh which would

play12:12

defined uh how silicon Graphics did

play12:14

computer graphics and so it was it was

play12:16

right there it was like $68 a book and

play12:18

so I had a couple hundred dollar I

play12:20

bought three books I took it back to the

play12:22

office and I said guys I found it our

play12:24

future and I handed out I had three

play12:26

versions of it handed out had a big nice

play12:29

centerfold you know the centerfold is

play12:32

the opengl pipeline which is the

play12:33

computer Graphics Pipeline and um uh and

play12:37

I handed it to uh the same Geniuses that

play12:41

I founded the company with and we

play12:44

implemented the openg pipeline like

play12:47

nobody had ever implemented the opengl

play12:49

pipeline and we built something the

play12:51

world never seen and so uh a lot of

play12:54

lessons are right there that moment in

play12:57

time for our company

play12:59

uh gave us so much confidence and the

play13:02

reason for that is you can

play13:05

succeed in doing something inventing a

play13:08

future even if you were not informed

play13:11

about it at all and is kind of the my

play13:14

attitude about everything now you know

play13:16

when somebody tells me about something

play13:18

and I've never heard of it before or if

play13:21

I've heard of it never don't understand

play13:23

how it works at all my first thought is

play13:26

always you know how hard can it be

play13:30

and it's probably just a textbook away

play13:33

you know you're probably one archive

play13:35

paper away from figuring this out and so

play13:38

I spent a lot of time reading archive

play13:39

papers and um and it it's true it's true

play13:43

you can you can um now of course you

play13:45

can't learn how somebody else does

play13:47

something and do it exactly the same way

play13:49

and hope to have a different outcome but

play13:51

you could learn how something can be

play13:53

done and then go back to First

play13:56

principles and ask yourself um giving

play13:59

the conditions today given my motivation

play14:02

given the instruments the tools um given

play14:04

you know how things have changed how

play14:06

would I redo

play14:08

this how would I reinvent this whole

play14:10

thing how would I design a how would I

play14:12

build a car today would I build it

play14:14

incrementally from 1950s and 1900s how

play14:17

would I build a computer today how would

play14:19

I write software today does that make

play14:21

sense and so I go back to First

play14:23

principles all the time uh even in the

play14:25

company today and just reset ourselves

play14:29

you know because the world has changed

play14:31

and U the way we wrote software in the

play14:33

past was monolithic and it's designed

play14:35

for supercomputers but now it's

play14:37

disaggregated it's you know so on so

play14:39

forth and how we think about software

play14:41

today how we think about computers today

play14:43

how we think just always cause your

play14:45

company always cause yourself to go back

play14:46

to first first principles and it creates

play14:48

lots and lots of opportunities yeah the

play14:51

way you applied this technology turns to

play14:53

be revolutionary you get all the

play14:55

momentum that you need to IPO and then

play14:58

some more because you grow your Revenue

play15:00

nine times in the next four years but in

play15:02

the middle of all of this success you

play15:05

decide to Pivot a little bit the focus

play15:08

of innovation happening at Nvidia based

play15:11

on a phone call you have with this

play15:13

chemistry professor can you tell us

play15:15

about that phone call and how you

play15:17

connected the dots from what you heard

play15:19

to where you

play15:21

went uh remember at the core the company

play15:24

was uh pioneering a new way of doing

play15:26

Computing computer Graphics was the

play15:28

first application uh but we already

play15:30

always knew that there would be other

play15:32

applications and so image processing

play15:34

came particle physics came fluids came

play15:36

so on so forth all kinds of interesting

play15:38

things that we wanted to do uh we made

play15:41

the processor more programmable so that

play15:43

we could express more algorithms if you

play15:47

will and then one day we invented um uh

play15:51

programable shaders which made all forms

play15:54

of Imaging and computer Graphics

play15:56

programmable that was a great

play15:57

breakthrough so we invented Ed that on

play16:00

top of that we invented uh we we tried

play16:02

to look for ways to express um uh more

play16:07

comp more sophisticated algorithms uh

play16:09

that could be computation that could be

play16:11

computed on our processor which is very

play16:13

different than a CPU and so we we

play16:15

created this thing called CG this I

play16:17

think it was 2003 or so C for

play16:21

gpus it predated Cuda by about three

play16:25

years um the same person who wrote The

play16:27

textbook that saved the company Mark

play16:29

Hilgard wrote that

play16:31

textbook and um I and so CG was was

play16:36

super cool we wrote textbooks about it

play16:37

we started teaching people how to use it

play16:39

we developed tools and such um and then

play16:42

several several researchers discovered

play16:45

it uh many of the researchers here

play16:47

students here at Stanford was using it

play16:49

um many of the the engineers that that

play16:51

then became uh engineers at Nvidia were

play16:54

were uh playing with it uh

play16:57

uh a doctor a couple of doctors at at

play17:01

Mass General picked it up and used it

play17:04

for uh CT reconstruction so I flew out

play17:06

and saw them and said you know what are

play17:08

you guys doing with this thing and uh

play17:10

they told me about that and then and

play17:12

then uh a um uh a

play17:15

computational a Quantum chemist uh used

play17:19

it to um uh Express his his algorithms

play17:23

and so I I realized that that there's

play17:26

there's some evidence that people might

play17:27

want to use this

play17:29

uh and and it gave it gave us gave us

play17:31

you know incrementally more more

play17:34

confidence that that we ought to go do

play17:36

this that that this field this form of

play17:38

computing could solve problems that

play17:40

normal computers really can't and and um

play17:44

reinforced our belief and and kept us

play17:46

going every time you heard something new

play17:50

you really savored that surprise and

play17:53

that seems to be a theme throughout your

play17:54

leadership at Nvidia U it feels like you

play17:58

make the these bets so far in advance of

play18:01

Technology inflections that when the

play18:03

Apple finally falls from the tree you're

play18:05

standing right there in your black

play18:07

leather jacket waiting to catch

play18:10

it how do you find the conv always seems

play18:12

like a diving catch oh it does seem like

play18:15

a diving catch you do things based on

play18:18

core beliefs you know we we uh we we

play18:21

deeply believe that that we uh we

play18:25

could create a computer that solves

play18:27

problems Norm processing can't do that

play18:30

there are limits to what a CPU can do

play18:32

there are limits to what general purpose

play18:33

Computing can do and then there are

play18:35

interesting problems uh that we can go

play18:37

solve the question the question is

play18:39

always are those in interesting problems

play18:42

only or are they can they also be

play18:44

interesting markets because if they're

play18:46

not interesting markets it's not

play18:48

sustainable and Nvidia went through

play18:51

about a decade where we were investing

play18:54

in this future and the markets didn't

play18:57

exist there was only One Market at the

play18:59

time was computer Graphics uh for 10 15

play19:02

years the markets that fuels Nvidia

play19:05

today just didn't exist and so so how do

play19:08

you

play19:09

continue um uh with all of the people

play19:12

around you you know our company and you

play19:14

know nvidia's management team and all of

play19:16

the amazing Engineers that they're

play19:18

creating this future with me um all of

play19:21

your shareholders your board of

play19:22

directors all your partners you're

play19:24

you're taking everybody with you and

play19:27

there's no evidence uh of a market that

play19:30

is really really challenging you know

play19:33

the fact that the technology can solve

play19:34

problems and the fact that you have

play19:36

research papers that that are used that

play19:39

that are made possible because of it are

play19:41

interesting but you're always looking

play19:43

for that market but nonetheless before a

play19:46

market exists you still need early

play19:48

indicators of future

play19:50

success you know we we have this phrase

play19:52

in the company is is you know there's a

play19:55

phrase called key performance indicators

play19:58

unfortunately kpis are hard to

play20:01

understand I find kpis hard to

play20:03

understand what's a good

play20:05

kpi you know a lot of people you know

play20:08

when when we look for kpis we go gross

play20:10

margins that's not a kpi that's a

play20:14

result you know you're looking for

play20:16

something that's an early indicators of

play20:18

future positive results okay and as

play20:22

early as possible and the reason for

play20:24

that is because you want early indic

play20:26

that early sign that you're going in the

play20:27

right direction

play20:29

and so we have this phrase is called EO

play20:31

ifs FS you know early indicators e FS

play20:35

early indicators of future success and

play20:38

and um and it helps people uh uh because

play20:42

I was using it all the time to give the

play20:45

company hope that hey look we solved

play20:48

this problem we solved that problem we

play20:49

solved this problem the markets didn't

play20:51

exist but there were important problems

play20:53

and that's what the company's about to

play20:55

solve these problems uh we want to be

play20:57

sustainable

play20:59

and therefore the markets have to exist

play21:01

at some point but you you want you want

play21:04

to decouple the result from um uh from

play21:09

evidence that you're doing the right

play21:10

thing okay and so so so that's how you

play21:13

that's how you kind of solve this

play21:15

problem of investing into something

play21:17

that's very very far away um and having

play21:20

the the conviction uh to stay on the

play21:22

road is to find as early as possible the

play21:25

indicators that you're doing the right

play21:27

things and so uh start with a core

play21:30

belief unless something you know changes

play21:32

your mind you continue to believe in it

play21:34

and um look for early indicators of

play21:36

future success what are some of those

play21:38

early indicators that have been used by

play21:40

product teams at

play21:42

Nvidia uh all kinds

play21:44

um uh uh I saw I saw I saw a uh a paper

play21:49

uh long before I saw the paper I met

play21:51

some people that needed my help on on um

play21:55

uh on this thing called Deep learning at

play21:56

a time I didn't even know what deep

play21:57

learning Le was and um and they needed

play22:00

us to create a domain specific language

play22:04

so that um all of their algorithms could

play22:06

be expressed easily on our on our

play22:09

processors and we created this thing

play22:10

called

play22:11

cdnn and it's essentially the SQL um uh

play22:16

SQL is in in storage Computing this is

play22:19

um neuron network computing and uh we

play22:22

created a a language if you will domain

play22:24

specific language for that you know kind

play22:26

of like the openg GL of of uh deep

play22:29

learning and so we we uh they needed us

play22:31

to do that so that they they could

play22:32

express their mathematics and uh they

play22:35

didn't understand Cuda but they

play22:36

understood their deep learning and so we

play22:39

created this thing in the middle for

play22:40

them uh and the reason why we did it was

play22:42

because uh even though there were zero I

play22:44

mean this you know these researchers had

play22:46

no money uh and and this is kind of one

play22:50

of the the great skills of our company

play22:53

that that you're willing to do something

play22:55

even though the financial returns are

play22:57

complet completely non-existent or maybe

play23:00

very very far out even if you believed

play23:01

in it uh we we ask ourselves you know is

play23:05

this worthy work to do um does this

play23:07

Advance a field of science somewhere

play23:09

that matters notice this is something

play23:11

that I I've been talking about you know

play23:13

since the very beginning of time uh we

play23:16

ex we we find inspiration uh not from

play23:20

the size of a market from but from the

play23:22

importance of the

play23:23

work uh because the importance of the

play23:25

work is the early indicators of a future

play23:27

Market

play23:28

and nobody has to write a nobody has to

play23:31

do a a um a business case on it nobody

play23:34

has to show me a a pnl uh nobody has to

play23:37

show me a financial forecast the only

play23:39

question is is this important work and

play23:41

if we didn't do it uh would it happen

play23:43

without us now if we didn't do something

play23:46

and something could happen without us it

play23:48

gives me tremendous Joy actually and the

play23:51

reason for that is could you imagine the

play23:53

world got better you didn't have to lift

play23:55

a finger that's the definition of you

play23:58

know of of uh ultimate laziness and and

play24:01

and in a lot of ways in a lot of ways

play24:03

you want that habit and the reason for

play24:05

that is

play24:06

this uh you want the company to be lazy

play24:08

about doing things that other people

play24:10

always do can do if somebody else can do

play24:13

it let them do it we should go select

play24:15

the things that if we didn't do it the

play24:17

world the world would fall apart you

play24:18

have to convince yourself of that that

play24:21

if I don't do this it won't get

play24:23

done that is Inc and and if that work is

play24:27

hard and that work is impactful and

play24:29

important then it gives you a sense of

play24:31

purpose does that make sense and so our

play24:33

company has been selecting these

play24:35

projects deep learning was just one of

play24:37

them and the first indicator of of the

play24:39

success of that was this you know fuzzy

play24:42

cat that that Andrew an came up with and

play24:45

um then Alex

play24:47

kvki uh detected cats um you know not

play24:51

all the time but you know successfully

play24:54

enough that it was you know this might

play24:55

take us somewhere and then we reasoned

play24:57

about the structure of deep learning and

play24:59

you know we're computer scientists and

play25:01

we understand how things work and and so

play25:03

we we uh we convinced ourselves this

play25:05

could change everything and and um and

play25:09

anyhow that but that's an that's an

play25:10

example so these selections that you've

play25:12

made they've paid huge dividends both

play25:15

literally and figuratively um but you've

play25:18

had to steer the company through some

play25:20

very challenging times like when it lost

play25:22

80% of its market cap amid the financial

play25:25

crisis cuz what Wall Street didn't

play25:27

believe in your bet on ML um in times

play25:31

like these how do you steer the company

play25:33

and keep the employees motivated at the

play25:35

task at hand uh it's this is the my

play25:39

reaction during that time is the same

play25:41

reaction I had about this week uh

play25:43

earlier today you asked me about this

play25:45

week my pulse was exactly the

play25:49

same this week is no different than last

play25:51

week or the week before that um and so

play25:55

the opposite of that you know when you

play25:57

drop it

play25:58

80% um it don't get me

play26:02

wrong when when your share price drops

play26:05

80% it's a little embarrassing okay and

play26:09

and um you just want to you just want to

play26:12

wear a t-shirt that says wasn't my

play26:15

fault

play26:17

um but even more than that you just you

play26:19

just don't want to you you don't want to

play26:21

get out of your bed you don't want to

play26:22

leave the house um all of that is true

play26:26

all of that is true um but then you go

play26:28

back to go back to just doing your job I

play26:31

woke up at the same time I prioritize my

play26:33

day in the same way uh I go back to what

play26:36

do I believe uh you got to gut check

play26:39

always gut check back to the court you

play26:41

know what do you believe uh what are the

play26:43

most important things uh and uh just

play26:46

check them off you know sometimes

play26:48

sometimes it's helpful to you know

play26:51

family loves me okay check um you know

play26:53

double you know right and so you just

play26:55

got to check it off and and you go back

play26:57

to your core um and then go back to work

play27:00

and and then every conversations go back

play27:02

to the core uh keep the company focused

play27:04

back on the core do you believe in it

play27:06

did something change the stock price

play27:08

changed but did something else change

play27:10

the physics change the gravity

play27:13

change did did all of the things that

play27:16

that that we assumed uh that we believed

play27:19

that led to our decision did any of

play27:20

those things change because if those

play27:23

things change you got to change

play27:24

everything but if none of those things

play27:25

change you change nothing you keep on

play27:26

going yeah yeah that's how you do

play27:29

it in speaking with your employees they

play27:32

say that you try to avoid the

play27:35

public in speaking with your employees

play27:38

they've said that your leadership

play27:39

including the employees I'm just

play27:43

kidding no le lead leaders have to be

play27:45

seen unfortunately that's the hard

play27:47

that's the hard part you know I I I was

play27:49

I was I was at I was I was an electrical

play27:52

engineering student and I was quite

play27:54

Young when I went to school um when I

play27:56

went to went to College I was I was

play27:59

still 16 years old and so I was I was

play28:01

young when I did everything and and so I

play28:03

was a bit of an introvert kind of you

play28:06

know I'm shy I don't enjoy public

play28:08

speaking I'm delighted to be here I'm

play28:10

not suggesting um but but it's it's not

play28:13

something that I do naturally and and um

play28:16

I and so so when when things are

play28:20

challenging um uh it's not easy to be in

play28:23

front of precisely the people that you

play28:26

care most about

play28:28

you know and the reason for that is

play28:30

because could you imagine a company

play28:32

meeting we just our stock prices dropped

play28:33

by

play28:34

80% and the most important thing I have

play28:37

to do as the CEO is this to come and

play28:40

face you explain it and partly you're

play28:44

not sure why

play28:47

partly you're not sure how long uh how

play28:51

bad yeah you just don't know these

play28:52

things and and but you still got to

play28:54

explain it face face all these people

play28:58

and you know what they're thinking you

play29:00

know you you know some of them are

play29:02

probably thinking we're doomed uh some

play29:04

people are probably thinking you're an

play29:05

idiot and some people are probably

play29:07

thinking you know something else and so

play29:09

I um there are a lot of things that

play29:12

people are thinking and you know that

play29:13

they're thinking those things but you

play29:14

still have to get in front of them and

play29:16

and and deal you know do the hard work

play29:18

they may be thinking of those things but

play29:20

yet not a single person of your

play29:22

leadership team left during times like

play29:24

this and in fact

play29:26

unemployable

play29:28

that's what I keep reminding them I'm

play29:31

just kidding I'm surrounded by Geniuses

play29:33

I'm surrounded by Geniuses yeah other

play29:36

Geniuses un un unbelievable uh Nvidia is

play29:40

well known to have singularly the best

play29:43

management team on the planet this is

play29:45

the deepest technology management team

play29:48

the world's ever seen I'm surrounded by

play29:50

a whole bunch of them and they're just

play29:52

genius business teams marketing teams

play29:55

sales teams just incredible and

play29:57

engineering teams my research teams

play30:00

unbelievable yeah your employees say

play30:03

that your leadership style is very

play30:05

engaged you have 50 direct reports you

play30:08

encourage people across all parts of the

play30:11

organization to send you the top five

play30:13

things on their mind and you constantly

play30:15

remind people that no task is beneath

play30:18

you can you tell us why you've

play30:21

purposefully designed such a flat

play30:23

organization and how should we be

play30:25

thinking about our organizations that we

play30:26

designed in the

play30:28

future uh no task is is to me no task is

play30:32

beneath me because remember I used to be

play30:34

a dishwasher and I and I mean that I

play30:36

used to clean toilets I mean you know I

play30:38

cleaned a lot of toilets I've cleaned

play30:39

more toilets than all of you

play30:41

combined and and some of them just can't

play30:46

[Laughter]

play30:49

unsee I don't know I I don't know what

play30:51

to tell you you know that's life and and

play30:55

so so uh uh you can't show me and you

play30:58

can't show me a task that is that's

play31:00

beneath me um now I'm not doing it I'm

play31:03

not doing it uh only because because of

play31:08

uh you know whether it's beneath me or

play31:09

not beneath me U if you send me

play31:11

something and you want my input on it

play31:14

and I can be of service to you and in my

play31:17

in my review of IT share with you how I

play31:20

reason through it uh I've made a

play31:22

contribution to you I've made I've made

play31:25

it possible for you to see how I reason

play31:26

through something and and by reasoning

play31:30

as you know how someone reasons through

play31:32

something empowers you you go oh my gosh

play31:35

that's how you reason through something

play31:36

like this it's not as complicated as it

play31:39

seems this is how you reason through

play31:41

something that's super ambiguous this is

play31:43

how you reason through something that's

play31:45

incalculable this is how you reason

play31:47

through something that you know seems to

play31:49

be very scary this is how you seem do

play31:51

you understand and so I show people how

play31:54

to reason through things all the

play31:56

time strategy things you know how to

play32:00

forecast something how to break a

play32:02

problem down uh and you're just you're

play32:05

empowering people all over the place and

play32:07

so that's how I see it if you send me

play32:09

something you want me to help review it

play32:11

uh I'll do my best and I'll show you how

play32:13

I would do it um I in the process of

play32:17

doing that of course I learned a lot

play32:18

from you is that right you gave me a

play32:21

seat of a lot of information I learned a

play32:23

lot and so I I feel rewarded by the

play32:26

process um it does take a lot of energy

play32:28

sometimes because you know you got in

play32:30

order to add value to somebody and

play32:31

they're incredibly smart as a starting

play32:33

point and I'm surrounded by incredibly

play32:34

smart people you have to at least get to

play32:37

their plane you know you have to get

play32:38

into their head space and that's really

play32:41

hard that's really hard um and that

play32:44

takes just an enormous amount of

play32:45

emotional and intellectual energy and so

play32:48

I feel exhausted after after I I work on

play32:51

things like that um I'm surrounded by by

play32:54

a lot of great people a CEO should have

play32:56

the most direct report rep s um uh by

play32:59

definition because the people that

play33:00

reports to the CEO requires the least

play33:02

amount of

play33:03

management it makes no sense to me that

play33:06

CEOs have so few people reporting to

play33:08

them except for one fact that I know to

play33:11

be true the the knowledge the

play33:14

information of a CEO is supposedly so so

play33:17

valuable so secretive you can only share

play33:20

with two other people or

play33:23

three and their information is so

play33:26

invaluable so incredibly secretive that

play33:29

they can only share with a couple

play33:31

more well

play33:34

um I don't believe in in in a culture an

play33:37

environment where the information that

play33:40

you possess is the reason why you have

play33:44

power I would like us all to to to

play33:48

contribute to the company and our

play33:50

position in the company should have

play33:52

something to do with our ability to

play33:54

reason through complicated things lead

play33:57

other people to um achieve greatness um

play34:00

Inspire Empower other people um support

play34:03

other people those are the reasons why

play34:05

the the management team exists in

play34:07

service of all of the other people that

play34:09

work in the company to create the

play34:11

conditions by which all of the all of

play34:13

these amazing people who volunteer to

play34:15

come work for you instead of all the

play34:17

other amazing high-tech companies around

play34:19

the world they elected they volunteer to

play34:22

work for you and so you should create

play34:24

the conditions by which they could do

play34:25

their life's work which is

play34:27

Mission you know you probably heard it

play34:30

i' I've said that you know pretty

play34:32

clearly and I and I believe that what my

play34:35

job is is very simply to create the

play34:37

conditions by which you could do your

play34:39

life's work and so how do I do that what

play34:42

does that condition look like well that

play34:44

condition should um result in great deal

play34:46

of empowerment you should you can only

play34:48

be empowered if you understand the

play34:49

circumstance isn't it right you have to

play34:52

understand the cont you have to

play34:53

understand the context of the situation

play34:54

you're in in order for you to come up

play34:55

with great ideas

play34:57

and so I have to create a circumstance

play34:59

where you understand the context which

play35:01

means you have to be

play35:03

informed and the best way to be informed

play35:06

is for there to be as little layers

play35:09

of information

play35:12

mutilation right between us and so

play35:15

that's the reason why it's very often

play35:18

that I'm reasoning through things like

play35:20

in an audience like this I say first of

play35:23

all this is the beginning facts these

play35:25

are the data that we have um this is how

play35:27

I would reason through it these are some

play35:28

of the assumptions these are some of the

play35:30

unknowns these are some of the

play35:33

knowns and so you reason through it and

play35:36

now you've created an organization

play35:37

that's highly empowered nvidia's 30,000

play35:40

people we're the smallest large company

play35:42

in the world we're tiny little company

play35:45

but every employee is so empowered and

play35:47

they're making smart decisions on my

play35:49

behalf every single day and the reason

play35:52

for that is because you know they

play35:54

understand that they understand my

play35:55

condition

play35:57

they understand my condition I'm very

play35:58

transparent with people um and uh and I

play36:03

believe that that I can trust you with

play36:04

the information often times the

play36:06

information is hard to hear and uh the

play36:09

the situations are complicated uh but I

play36:11

trust that you can handle it you're you

play36:14

know a lot of people hear me say you

play36:16

know these you're adults here you can

play36:18

handle this sometimes they're not really

play36:20

adults they just

play36:22

graduated I'm just kidding I know that

play36:25

when I first graduated was barely an

play36:27

adult and um I but I was I was fortunate

play36:30

that I was trusted with with uh with uh

play36:34

important information so I want to do

play36:36

that I want to create the conditions for

play36:37

people to do

play36:38

that I do want to now address the topic

play36:42

that is on everybody's mind AI last week

play36:45

you said that generative Ai and

play36:47

accelerated Computing have hit the

play36:49

Tipping Point so as this technology

play36:51

becomes more mainstream what are the

play36:53

applications that you personally are

play36:55

most excited about

play36:58

well you have to go back to First

play36:59

principles and ask yourself what is

play37:01

generative AI what happened um what

play37:04

happened was we have a we now have the

play37:06

ability to have software that can

play37:08

understand something they they can

play37:10

understand why you know what is first of

play37:12

all we digitized everything that was you

play37:15

know like for example Gene sequencing

play37:16

you digitized genes but what does it

play37:19

mean that sequence of genes what does it

play37:21

mean we've digitized amino acids um but

play37:25

what does it mean uh and so we now have

play37:27

the ability we dig digitize words we

play37:30

digitize sounds uh we digitize images

play37:33

and videos we digitize a lot of things

play37:35

but what does it mean we now have the

play37:37

ability through um a lot of study a lot

play37:40

of Da data and from their patterns and

play37:41

relationships we We Now understand what

play37:43

they mean not only do we understand what

play37:46

they mean we we can translate between

play37:47

them because we learned about the

play37:50

meaning of these things in the same

play37:52

world we didn't learn about them

play37:53

separately so we we learned about speech

play37:56

and and words and and paragraphs and

play37:59

vocabulary in the same context and so we

play38:01

found correlations between them and

play38:03

they're all you know registered if you

play38:05

will registered to each other and so now

play38:08

we uh not only do we understand uh the

play38:11

modality the meaning of each modality we

play38:13

can understand how to translate between

play38:15

them and so uh for obvious things you

play38:18

could caption video to text that's

play38:20

captioning uh text to uh images M

play38:23

Journey uh text to text chat GPT I

play38:26

amazing things and so so we now we now

play38:29

know that uh we understand meaning and

play38:31

we can translate uh the translation of

play38:34

something is generation of information

play38:36

and and um uh and all of a sudden you

play38:39

you have to take your you take a step

play38:40

back and ask yourself um uh what is the

play38:43

implication in every single layer of

play38:45

everything that we do and so I'm

play38:48

exercising in front of you I'm reasoning

play38:50

in front of you uh the same thing I did

play38:52

a quarter uh 15 years ago when I first

play38:55

saw um uh alexnet some 13 14 years ago I

play38:59

guess um I how I reasoned through it uh

play39:02

what did I

play39:04

see how interesting what can it

play39:08

do very cool but then most importantly

play39:11

what does it mean what does it mean what

play39:13

does it mean to every single layer of

play39:15

computing because you know we're in the

play39:16

world of computing and so what it means

play39:18

is that that the way that we um process

play39:21

information fundamentally will be

play39:22

different in the future that's what

play39:24

Nvidia builds you know chips and system

play39:27

the way we write software will be

play39:28

fundamentally different in the future

play39:30

the type of software we'll be able to

play39:31

write write in the future will be

play39:33

different new applications and then ALS

play39:36

also the processing of those

play39:39

applications will be different what was

play39:41

historically a retrieval based model

play39:44

where uh in uh information was pre

play39:48

pre-recorded if you will almost you know

play39:50

we wrote the text pre-recorded and we

play39:52

retrieved it based on uh some

play39:54

recommender system algorithm in the

play39:57

future uh some seed of information will

play39:59

be will be uh the starting point we call

play40:02

them prompts you as you guys know and

play40:04

then we generate the rest of it and so

play40:07

the future of computing will be highly

play40:09

generated well let me give you an

play40:10

example of what's

play40:12

happening for example uh we're having a

play40:14

conversation right now very little of

play40:17

the information I'm trans I'm conveying

play40:19

to you is Retreat most of it is

play40:23

generated it's called intelligence

play40:27

and so in the future we're going to have

play40:28

a lot more generative our computers will

play40:30

will perform in that way it's going to

play40:32

be highly generative instead of Highly

play40:34

retrieval based you go back and you got

play40:36

to ask yourself you know now for for you

play40:39

know entrepreneurs you got to ask

play40:40

yourself uh what industries will be

play40:43

disrupted therefore will we think about

play40:45

networking the same way will we think

play40:46

about storage the same way will we think

play40:47

about would we be as abusive of internet

play40:50

traffic as we are today probably not

play40:54

notice we're having a conversation right

play40:55

now and and I to get in my car every

play40:57

every

play40:58

question so we don't have to be as

play41:01

abusive of of transformation information

play41:05

transporting as we used to um uh what's

play41:08

going to be more what's going to be less

play41:09

uh what kind of applications you know

play41:11

etc etc so you can go through the entire

play41:13

industrial spread and ask yourself

play41:15

what's going to get disrupted what's

play41:16

going to get be different what's going

play41:17

to get NED you know so on so forth and

play41:20

and that reasoning starts from what is

play41:22

happening what is generative

play41:24

AI Foundation Al what is happening go

play41:28

back to First principles with all things

play41:30

there was something I was going to tell

play41:31

you about organization you asked the

play41:32

question and I forgot to answer it the

play41:34

way you create an organization by the

play41:36

way someday um don't worry about how

play41:39

other companies or charts look you start

play41:42

from first principles remember what an

play41:44

organization is designed to

play41:46

do the organizations of the past where

play41:49

there's a king you know

play41:53

CE and then then you have all all these

play41:56

you know the Royal subjects you know the

play41:58

Royal Court and then eaff and then you

play42:01

keep working your way down eventually

play42:03

they're employees well the reason why it

play42:05

was designed that way is because they

play42:07

they wanted the employees to have as low

play42:09

information as possible because their

play42:10

fundamental purpose of the soldiers is

play42:13

to die in the field of

play42:15

battle to die without asking questions

play42:18

you guys know

play42:19

this I don't I only have 30,000

play42:22

employees I would like them none of them

play42:24

to die

play42:27

I would like them to question everything

play42:29

does that make sense and so the way you

play42:31

organize in the past and the way you

play42:32

organize today is very different to

play42:34

Second the question is what is nid what

play42:36

does Nvidia build an organization is

play42:38

designed so that we could build what it

play42:40

whatever it is we build

play42:43

better and so if we all build different

play42:45

things why why are we organized the same

play42:48

way why would why would this

play42:51

organizational Machinery be exactly the

play42:54

same irrespective of what you build it

play42:55

doesn't make make any sense you build

play42:58

computers you organize this way you

play42:59

build healthare Services you build

play43:01

exactly the same way it makes no sense

play43:04

whatsoever and so you had to go back to

play43:06

First principles just ask yourself what

play43:07

kind of Machinery what what is the input

play43:09

what is the output what are the

play43:11

properties of this environment you know

play43:14

what what is the what is the what is the

play43:16

forest that this animal has to live in

play43:19

what is this characteristics is it

play43:21

stable most of the time you're trying to

play43:22

squeeze out the last drop of water or is

play43:25

it changing all the time being attacked

play43:28

by everybody and so you got to

play43:30

understand you know you're the CEO your

play43:33

job is to architect this company that's

play43:34

my first job to create the conditions by

play43:37

which you can do your life's work and

play43:38

the architecture has to be right and so

play43:41

you have to go back to First principles

play43:42

and think about those things and I was

play43:44

fortunate that that when I was 29 years

play43:46

old you know I had the benefit of of of

play43:49

taking a step back and asking myself you

play43:51

know how would I build this company for

play43:52

the future and what would it look like

play43:54

and you know what's the operating system

play43:55

which is called culture what do we what

play43:57

kind of behavior do we en encourage

play44:00

enhance and what what do we discourage

play44:02

and not enhance you know so on so forth

play44:04

and anyways I want to save time for

play44:07

audience questions but um this year's

play44:10

theme for view from the top is

play44:11

redefining tomorrow and one question

play44:13

we've asked all of our guests is Jensen

play44:16

as the co-founder and CEO of Nvidia if

play44:19

you were to close your eyes and

play44:21

magically change one thing about

play44:22

tomorrow what would it

play44:25

be

play44:29

were we supposed to think about this in

play44:35

advance I I'm going to give you a

play44:38

horrible answer

play44:42

um I I don't know that it's one thing

play44:45

look there are a lot of things we don't

play44:48

control you know there are a lot of

play44:50

things we don't control um your job is

play44:53

to make a unique contribution live a

play44:56

life of

play44:57

purpose to do something that nobody else

play45:00

in the world would do or can do to make

play45:03

a unique contribution so that in the

play45:05

event that after you done

play45:09

um everybody says you know the world was

play45:13

better because you were

play45:15

here and so I think that that to me um I

play45:20

live I live my life kind of like this I

play45:22

go forward in time and I Look Backwards

play45:26

so you asked me a question that's

play45:27

exactly from a from a computer vision

play45:30

pose perspective exactly the opposite of

play45:32

how I think I never look forward from

play45:35

where I am I go forward in time and look

play45:38

backwards and the reason for that is

play45:40

it's

play45:42

easier I would look backwards and kind

play45:44

of read my

play45:45

history we did this and we did that way

play45:47

and we broke that prom down does that

play45:49

make sense and so it's a little bit like

play45:52

um how you guys solve problems you

play45:55

figure figure out what is the end result

play45:57

that you're looking for and you work

play45:59

backwards to achieve it and so I imagine

play46:02

Nvidia uh making a unique contribution

play46:04

to advancing the the future of of uh of

play46:07

computing which is the single most

play46:09

important instrument of all

play46:11

Humanity now it's not about our self

play46:14

self-importance but this is just what

play46:16

we're good at and it's incredibly hard

play46:18

to do and we believe we can make an

play46:20

absolute unique contribution it's taken

play46:22

US 31 years to be here and we're still

play46:24

just beginning our journey and so this

play46:26

is insanely hard to

play46:28

do and uh uh When I Look Backwards I

play46:31

believe that we made I believe that that

play46:34

we're going to be remembered as a

play46:36

company that kind of changed everything

play46:38

not because we went out and changed

play46:39

everything through all the things that

play46:41

we said but because we did this one

play46:43

thing that was insanely hard to do that

play46:45

we're incredibly good at doing that we

play46:47

loved doing we did for a long time I'm

play46:49

part of the GSP lead I graduated in 2023

play46:53

so my question is how do you see see

play46:56

your company in the next decade as what

play46:59

challenges do you see your company would

play47:01

face and how you are positioned for that

play47:03

first of all can I just tell you what

play47:04

was going on through my head as you say

play47:07

what

play47:08

challenges the list that flew by my

play47:12

head was so so large uh that that I was

play47:17

trying to figure out what to

play47:20

select um now the honest truth is is

play47:23

that when you ask that question

play47:26

most of the challenges that showed up

play47:27

for me were technical

play47:30

challenges and the reason for that is

play47:32

because that was my

play47:34

morning if you were to you know chosen

play47:37

yesterday um it might have been Market

play47:39

creation

play47:40

challenges there are some markets that I

play47:42

gosh I just desperately would love to

play47:44

create I just can't we just do it

play47:47

already you know but we can't do it

play47:49

alone Nvidia is a technology platform

play47:52

company we're here in service of a whole

play47:55

bunch of other the companies so that

play47:57

they could realize if you will our hopes

play48:01

and dreams through

play48:02

them and and so some of the things that

play48:05

I would love I would love for the world

play48:08

of biology to to be at a point where

play48:11

it's kind of like the world of Chip

play48:13

design 40 years ago computer AED and

play48:16

design um Eda that entire industry

play48:20

really made possible for us today and I

play48:23

believe we're going to make possible for

play48:24

them tomorrow

play48:26

computer AED drug design because we're

play48:29

able to now represent genes and proteins

play48:32

and even cells now very very close to be

play48:35

able to represent and understand the

play48:37

meaning of a cell a combination of a

play48:39

whole bunch of genes um what is a cell

play48:42

mean it's kind of like what does that

play48:44

paragraph mean well if we could

play48:47

understand a a cell like we can

play48:49

understand a paragraph imagine what we

play48:51

could do and so uh so so I'm I'm anxious

play48:55

for that to happen you know I'm kind of

play48:57

excited about that uh there's some that

play48:59

I'm just excited about that I know we

play49:02

around the corner on for example uh

play49:04

humanoid

play49:05

robotics very very close around the

play49:07

corner and the reason for that is

play49:08

because if you can tokenize and

play49:09

understand speech why can't you tokenize

play49:11

and understand uh

play49:13

manipulation and so so these kind of

play49:16

computer science techniques you once you

play49:18

figure something out you ask yourself

play49:19

well if got do that why can't I do that

play49:21

and so I'm excited about those kind of

play49:23

things um and so that challenge is kind

play49:25

of a happy

play49:26

challenge uh some of the some of the

play49:29

other challenges some of the other

play49:30

challenges of course are industrial and

play49:33

geopolitical and they're social and and

play49:36

but you've heard all that stuff before

play49:38

these are all true you know the social

play49:40

issues in in the world uh the

play49:42

geopolitical issues in the world uh why

play49:44

can't we just get along uh things in the

play49:47

world why do I have to say those kind of

play49:48

things in the world um why do we have to

play49:50

say those things and then amplify them

play49:52

in the world uh why do we have to judge

play49:53

people so much in the world uh you you

play49:55

know all those things you guys all know

play49:57

that I don't have to say those things

play49:58

over again my name is Jose I'm a class

play50:00

of the 2023 uh from the GSB my question

play50:04

is uh are you worried at all about the

play50:06

pace at which we're developing AI um and

play50:09

do you believe that any sort of

play50:11

Regulation might be needed thank you uh

play50:14

yeah that's uh the answer is yes and no

play50:17

um we need uh you you know that the the

play50:20

the greatest breakthrough in uh modern

play50:23

AI of course deep learning and it

play50:25

enabled great progress but another

play50:28

incredible breakthrough is something

play50:30

that that humans know and we practice

play50:32

all the time uh and we just invented it

play50:34

for uh for language models called uh

play50:37

grounding reinforcement learning human

play50:39

feedback um I provide reinforcement

play50:42

learning human feedback every day that's

play50:44

my job um and their for their parents in

play50:47

the room uh you're providing

play50:49

reinforcement learning human feedback

play50:50

all the time okay now we just figured

play50:53

out how to do that um at a system

play50:55

systematic level for artificial

play50:57

intelligence there are a whole bunch of

play50:59

other technology necessary to uh

play51:02

guardrail uh

play51:04