Oct 16, 2023 · 1h 30m · acquired
NVIDIA CEO Jensen Huang · Acquired
Jensen Huang · 52m spoken Ben Gilbert · 15m spoken David Rosenthal · 12m spoken Acquired Theme Singer · 11s spoken
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this comprehensive interview on the Acquired podcast, NVIDIA founder and CEO Jensen Huang details the company's thirty-year journey from a struggling graphics startup to a trillion-dollar computing platform driving the generative AI revolution. He shares profound insights on high-stakes risk management, flat organizational design, early strategic bets on CUDA and data centers, and the intense emotional resilience required to build an enduring enterprise.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Ben and David hold 33% of the talking time here. How this is scored →
Ben and David as informed peer 4.7 Guest teaching 4.4 Guest disagreement 1.6 Ben and David pushing back 1.0
speaking balance: gold is Ben and David, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 11:09 Jensen rejects the 'betting the farm' framing
Jensen firmly pushes back on Ben's premise that NVIDIA simply gambled on the Riva 128, insisting that comprehensive simulation was an exercise in prefetching future certainty.
Hardest push from Ben and David ▶ 52:33 Ben presses Jensen on DirectNV's failure against MicrosoftBen directly synthesizes the hard lesson NVIDIA learned when Microsoft co-opted developer relations with DirectX, forcing NVIDIA to adapt.
Biggest teaching moment ▶ 50:54 Jensen corrects hosts on NVIDIA's platform originsJensen corrects David's assertion that NVIDIA began merely as a component company, detailing how UDA unified architecture was designed as a developer platform from day one.
Ben and David hold their own ▶ 1:05:30 David connects AI expansion to Sequoia's Moritz CorollaryDavid showcases deep institutional venture knowledge by linking Jensen's productivity arguments directly to Mike Moritz's corollary to Moore's law.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Ben and David as informed peer | Guest teaching | Guest disagreement | Ben and David pushing back | Why |
|---|---|---|---|---|---|---|
| Pre-Show Banter and Acquired Theme Song | 0 | 0 | 0 | 0 | Introductory housekeeping, banter, and teaser setup by the hosts before the interview begins. No substantive guest interaction takes place. | |
| The Riva 128 Gamble and Emulation Breakthrough | 7 | 5 | 3 | 2 | The hosts exhibit detailed historical research into the Riva 128 launch. Jensen reframes Ben's assertion about 'betting the farm,' explaining that simulation meant eliminating risk rather than gambling blindly. | |
| The Inception of CUDA and the Deep Learning Revolution | 6 | 6 | 2 | 1 | Hosts discuss CUDA's 10,000 person-year investment ahead of demand. Jensen educates them on first-principles reasoning about deep learning as a universal function approximator. | |
| Partnering with AI Pioneers and Scaling Language Models | 6 | 5 | 1 | 1 | Hosts probe Jensen's relationship with AI researchers and early delivery of DGX to OpenAI. Jensen explains the self-supervised intuition behind BERT and scaling LLMs. | |
| Sponsor: Statsig's Product Experimentation Platform | 0 | 0 | 0 | 0 | Mid-roll advertisement for Statsig. Monologue format. | |
| NVIDIA's Flat Org Architecture: Computing Stack as Company | 5 | 6 | 2 | 1 | Hosts ask about Jensen's 40+ direct reports and unconventional hierarchy. Jensen dismantles typical corporate command-and-control structures in favor of a flat computing stack model where 'mission is the boss.' | |
| Learning from Others Without Imitation | 5 | 5 | 2 | 1 | David asks which peer companies Jensen admires or models NVIDIA after. Jensen rejects the premise of imitating other companies, arguing that reading business books is for inspiration rather than adoption. | |
| Pivoting to the Data Center: Uncoupling Compute from Displays | 6 | 5 | 1 | 1 | Jensen shares the early genesis of data center computing via GeForce Now and uncoupling GPU compute from display monitors. The hosts follow along closely and connect the narrative to cloud gaming. | |
| Mellanox Acquisition and the Necessity of InfiniBand | 6 | 6 | 2 | 1 | Hosts ask why NVIDIA foresaw the necessity of InfiniBand and acquired Mellanox before anyone else. Jensen explains how AI distributed computing is the architectural opposite of commodity hyperscale virtualization. | |
| Sponsor: Crusoe Clean Energy Cloud Infrastructure | 0 | 0 | 0 | 0 | Mid-roll advertisement for Crusoe. Monologue format. | |
| Investing in Zero-Billion-Dollar Markets and Platform Networks | 5 | 6 | 2 | 1 | Hosts seek startup strategy advice. Jensen details NVIDIA's strategy of living in 'zero-billion-dollar markets' and building developer-enabling ecosystems rather than defensive moats. | |
| Architectural Compatibility: From DirectNV to CUDA | 7 | 6 | 2 | 2 | David suggests NVIDIA only became a platform with CUDA. Jensen corrects him, revealing the company was built as a platform from day one under UDA and DirectNV, with strict architectural backward compatibility. | |
| Navigating Startup Peril and the Role of Luck | 6 | 5 | 3 | 2 | Jensen demurs on giving standard founder advice, emphasizing that company building is brutally hard and depends on luck, such as John Carmack deciding to support accelerated 3D graphics. | |
| The Impact of AI: Safety, Job Creation, and Human Ambition | 6 | 5 | 2 | 1 | Hosts ask about job displacement from AI. Jensen argues that increased productivity leads to industry prosperity and higher net employment because human ambition is infinite, matching David's Moritz corollary thesis. | |
| Sponsor: Blinkist and Go1 Book Insights | 4 | 4 | 2 | 1 | Blinkist sponsor segment transitioning into rapid-fire personal questions about cars, favorite business books, and Don Valentine. | |
| Loyalty, LSI Logic, and Enduring Startup Trauma | 6 | 5 | 2 | 1 | Jensen reflects on his foundational tenure at LSI Logic, how high-level chip design abstraction shaped his philosophy, and getting introduced to Don Valentine without knowing how to write a business plan. | |
| The Superpower of Entrepreneurship and the Trillion-Dollar AI Future | 5 | 6 | 2 | 1 | Jensen reveals he would never start a company again if he knew the pain in advance, describing entrepreneurial ignorance as a superpower. He closes on NVIDIA transforming from a chip company into an intelligence manufacturing provider. |
Statements from this episode (36)
Huang: NVIDIA was the only consumer 3D graphics company in 1993
“Now remember, at the time that we started the company in 1993,
We were the only consumer, three D graphics company ever created”
Huang: 1997 was probably NVIDIA's best moment despite near bankruptcy
“And that time, 1997, was probably NVIDIA's best moment, and the reason for that was our backs were up against the wall, we were running out of time, we were running out of money, and for a lot of employees, running out of hope.”
Huang: Markets where technology is never good enough offer sustainable opportunity
“And for certain segments of the market where the technology is never good enough, like three D graphics, when we chose the right technology, three D graphics is never good enough. And we call it back then, three D gives us sustainable technology opportunity be…”
Huang: Betting the farm requires prefetching and simulating all future risk
“And so when you push your chips in, what you're really doing is you're, when you bet the farm, you're saying, I'm going to take everything in the future, all the risky things, and I'm going to pull it in advance. And that is probably the lesson. And to this da…”
NVIDIA's programmable shaders were the first massively parallel processors
“And if you just looked at the pipeline of a programmable shader, it is a processor and is highly parallel, and it is massively threaded, and it is the only processor in the world that does that.”
Huang: Causality matters less than outcome prediction for many problems
“And so causality for a lot of problems in the world doesn't matter. We just want to emulate the system and predict the outcome.”
Huang says almost all software will be programmed via machine learning
“Almost every piece of software in the world would eventually be programmed this way.”
Huang: CUDA is used across almost all fields of scientific research
“CUDA is used for almost all fields of science. Everything from molecular dynamics to imaging, CT reconstruction to seismic processing to, you know, weather simulations, quantum chemistry, the list goes on, right?”
Huang personally delivered the first DGX supercomputer to OpenAI
“DGX, the first version that we built was used internally, and I delivered the first one to OpenAI”
Huang: LLM reasoning and emergent capabilities stem naturally from reading text
“Within the world's languages and text, there's a fair amount of reasoning that's encoded in it, and we describe a lot of reasoning things, and so if you were To say that a few step reasoning is somehow learnable from just reading things. I wouldn't be surprise…”
Huang: Company org charts should mirror the architecture of their products
“Your organization should be the architecture of the machinery of building the product. Right. That's what a company is.”
Huang says traditional managers rely on information asymmetry, unlike at NVIDIA
“In a command and control system, the person who you reports to has more power than you. And the reason why they have more power than you is because they're closer to the source of information than you are. In our company, the information is disseminated fairly…”
Huang: Leaders should take inspiration from business books without adopting them
“Enjoy it, be inspired by it, but not to adopt it. That's not the whole point of these books. The whole point of these books is to share their experiences, and you, you're supposed to ask, you know, what does it mean to me in my world, and what does it mean to …”
Huang: GeForce NOW was NVIDIA's first data center product, preceding CUDA supercomputing
“GeForce Now was NVIDIA's first data center product. And our second data center product was remote graphics, putting our GPUs in, in the world's enterprise data centers, which then led us to our third product, which combined CUDA plus our GPU, which became a su…”
Huang: Separating compute location from display explodes hardware market opportunity
“The disconnection between where NVIDIA's computing is done versus where it's enjoyed, If you can separate that, your market opportunity explodes.”
Huang: CEOs must position companies near future opportunities before they appear
“You want to pave the way to future opportunities. You can't wait until the opportunity is sitting in front of you for you to reach out for it, and so you have to anticipate. You know, our job as CEO is to look around corners and anticipate where will opportuni…”
Huang: Distributed networking and infrastructure define data centers, not chips
“A data center is distinguished from a desktop computer versus a cell phone, not by the processor in it. A desktop computer in a data center uses the same CPUs, uses the same GPUs, apparently, right? Very close. And so it's not the chip, it's not the processing…”
Huang: AI computing is the inverse of traditional hyperscale cloud computing
“Observation that whereas cloud computing started in hyperscale, which is about taking commodity components, a lot of users, and virtualizing many users on top of one computer, AI is really about distributed computing where one job, one training job is orchestr…”
Huang: Someday every car will be software-defined
“And I said, I don't think anybody can tolerate that, but it doesn't change the fact that someday every car will be a software defined car.”
Huang spends significant time on Omniverse, a 'zero-billion-dollar' business
“And so we spent about a decade in zero billion dollar markets, and today I spent a lot of time on Omniverse, and Omniverse is a, you know, classic example of a zero billion dollar business.”
Huang: Enabling an ecosystem in an early market builds a network moat
“Oftentimes, if you created the market, you ended up having, you know, what people describe as moats. Because if you build your product right, and it's enabled an entire ecosystem around you, To help serve that end market, you've essentially created a platform.…”
Huang: NVIDIA is only accelerator maker with universal architectural compatibility
“We are the only accelerator on the planet where every single accelerator is architecturally compatible with the others. None has ever existed.”
Huang: NVIDIA has 250M to 300M active compatible CUDA GPUs globally
“There are literally a couple of hundred million, right? 250,000,300 million installed base of active CUDA GPUs being used in the world today, and they're all architecturally compatible.”
Architectural compatibility is NVIDIA's only non-negotiable operating rule
“And so, that's the only unnegotiable rule in our company. Everything else is negotiable.”
Huang: Founders can do everything right and fail, or make mistakes and succeed
“You could do all the right smart things and still fail. You could do a whole bunch of dumb things, and I did many of them, and still succeed.”
Huang: NVIDIA's RIVA 128 chip design methodology became the industry standard
“The number of smart decisions we made, which are smart to this day, how we design chips is exactly the same to this day. Because, gosh, you know, nobody's ever done it back then, and we pulled every trick in the book in a desperation because we had no other ch…”
Huang: Quake and Unreal were the first two killer consumer 3D apps
“If he hadn't decided to go to OpenGL and accelerate, accelerate for Quake frankly, you know, what would be the killer app that put us here? And so Carmack and Sweeney, both between Unreal and Quake, created the first two killer applications for consumer three …”
Huang: The primary benefit of building for developers is seeing external ideas
“As many good ideas as we have you don't have all the world's good ideas. And the benefit of having developers is you get to see a lot of good ideas.”
NVIDIA partnered with Getty and Shutterstock instead of scraping data
“Instead of scraping the internet we partnered with Getty and Shutterstock to create commercially fair way of applying artificial intelligence sharing to the AI.”
Huang says workers will lose jobs to AI users, not AI
“It's more likely that someone, ah, will lose a job to someone else Some other human that uses an AI, you know, and not, not likely to an AI, but to some other human that uses an AI.”
Jensen Huang: There is plenty of time if you don't let Outlook control your schedule
“There's plenty of time. If you prioritize yourself properly and you make sure that you don't let Outlook be the controller of your time, there's plenty of time.”
Jensen Huang: NVIDIA is his final job unless he is fired
“Until, until I'm fired, This is my last job.”
Huang: Startup financial forecasts matter far less than core defensibility
“It turns out that making a financial forecast that nobody knows is going to be right or wrong turns out not to be that important. But the important things that a business plan probably could have teased out, I think that the art of writing a business plan ough…”
Huang says he wouldn't have started NVIDIA if he knew the suffering
“The reason why I wouldn't do it, and it goes back to why it's so hard, is building a company and building a video turned out to have been a million times harder than I expected it to be, any of us expected it to be. And at that time, if we realized the pain an…”
Huang: An entrepreneur's superpower is asking 'how hard can it be?'
“And I think that that's kind of the superpower of a entrepreneur. They don't know how hard it is, and they only ask themselves, how hard can it be? And to this day, I trick my brain into thinking, how hard can it be? Because you have to.”