Apr 25, 2023 · 52m · no-priors
No Priors Ep. 13 | With Jensen Huang, Founder & CEO of NVIDIA
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
NVIDIA Founder and CEO Jensen Huang joins hosts Sarah Guo and Elad Gil to discuss the evolution of accelerated computing, architectural breakthroughs in artificial intelligence, and his philosophy on organizational leadership and long-term enterprise conviction.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 15.2% of the talking time here. How this is scored →
The hosts as informed peer 4.4 Guest teaching 5.4 Guest disagreement 1.3 The hosts pushing back 1.4
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 19:31 Dismissal of sentience framing
Jensen immediately rejects Elad's question about machine sentience, bluntly stating he doesn't know what the word means technically before refocusing on concrete algorithmic reasoning.
Hardest push from the hosts ▶ 29:48 Sarah holds Jensen to her specific questionWhen Jensen jokes and pivots to whether he was the right person to lead the company, Sarah directly pushes back to restate her question about his conviction in accelerated computing.
Biggest teaching moment ▶ 6:20 Economic and engineering calculus of CUDA vs CPUsJensen masterfully breaks down the math of how a 150 million dollar R&D budget in a niche application had to outpace general CPU R&D through strict architectural compatibility and targeted acceleration.
The host holds their own ▶ 9:43 Elad highlights interconnect scaling advantagesElad demonstrates deep technical knowledge by highlighting that NVIDIA's true moat in AI workloads is not just the GPU core, but CUDA ecosystem lock-in and scalable fabric interconnects.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Jensen Huang's Early Career and the Genesis of NVIDIA | 3 | 4 | 1 | 1 | Sarah sets up open-ended foundational questions about Jensen's background at AMD, LSI, and the founding of NVIDIA. Jensen gives an expansive historical narrative explaining how early conviction in accelerated computing went against 99 percent of Silicon Valley consensus. | |
| The Architecture of Accelerated Computing and the Creation of CUDA | 5 | 6 | 1 | 1 | Jensen details the technical dilemma of keeping CUDA general enough for developers while maintaining acceleration advantage over multi-billion-dollar CPU R&D budgets. Elad contributes domain familiarity regarding interconnect scalability and parallelization. | |
| The 2012 Deep Learning Inflection Point and ImageNet Breakthrough | 5 | 6 | 1 | 1 | The hosts and Jensen discuss the 2012 deep learning convergence with AlexNet, Hinton, and Ng. Jensen educates on viewing neural networks not merely as computer vision algorithms but as universal function approximators that change computer science fundamentals. | |
| The Rise of Transformers and the Disruption of Computer Programming | 5 | 5 | 2 | 2 | Discussion centers on transformers, prompt-based programming, and ChatGPT. Jensen quickly brushes aside Elad's philosophical question about machine sentience to focus on concrete software reasoning and code generation capabilities. | |
| NVIDIA's Full-Stack Strategy and Long-Term Conviction | 5 | 6 | 2 | 3 | Sarah asks how Jensen balanced public market pressures and activist investors with NVIDIA's 30-year technical roadmap. When Jensen playfully pivots to whether he was right for the CEO job, Sarah firmly clarifies her premise regarding technical conviction. | |
| Organizational Architecture, Flat Management, and the H100 Hopper Breakthrough | 4 | 6 | 1 | 1 | Jensen explains his flat organizational architecture with over 40 direct reports and no one-on-ones, before detailing the engineering decisions behind the H100 Hopper chip, specifically 8-bit floating point quantization and the transformer engine. | |
| Emerging AI Architectures: Robotics and Generative Multimodality | 4 | 5 | 1 | 1 | Jensen outlines future frontiers including robotics foundation models learned from video structure, generative multimodality, and NVIDIA's foundational work starting from GANs through diffusion. |
Statements from this episode (17)
Huang: Founding NVIDIA was Chris Malachowsky and Curtis Priem's idea
“It wasn't my idea. It was theirs. Chris and Curtis wanted to leave Sun.”
Huang: For 25 years, the consensus favoring general-purpose computing was right
“At the time, the Valley was the way of designing computers was rather, rather split between general purpose computing versus using accelerators. And about 99% of the Valley was believed in general purpose computing. And about one percent believed in accelerati…”
Huang: NVIDIA made every chip CUDA-compatible despite few initial customers
“And so for the first five, 10 years, you know, we had very few customers for CUDA, but we made every chip CUDA compatible.”
Huang: Geoffrey Hinton's lab first trained deep neural networks on GeForce gaming cards
“The first GPU that Jeff Hinton got for his lab, you know Hila would tell you that, that Jeff came in and said, here's a couple of GPUs, it's called GeForce, and you guys should try to use that for DNN, and so it was a gaming card.”
Huang: Andrew Ng contacted NVIDIA around 2012 to train neural nets on GPUs
“Around 20 12, I guess, and it was because simultaneously Andrew Eng reached out to Bill Daly, our chief scientist to work on a way to get the neural network model that they were working on onto GPU so that they could, instead of using thousands of CPU servers,…”
Huang: Natural human language has disrupted traditional computer programming
“The observation that computer programming has now been completely disrupted. That for the very first time in the history of computing, the language of programming a computer is human. You know, any human language. And it doesn't even have to be grammatically c…”
Huang: NVIDIA is not trying to be an AI model company
“You know, we're not trying to be an AI model company. We're trying to help industries create AI models. Mostly we're trying to help developers.”
Huang: Making money is a learnable skill, not conviction
“You know, make me, making money is not a matter of conviction. Making money is a matter of skill. And it's a learnable skill.”
Huang: Someday all computing workloads will be accelerated
“We deeply believe that, that someday everything will be accelerated. And the reason for that is, is very clearly that, that the CPU will run its course. And there's a limit to how far you could scale general purpose computing and you'll always need it. You alw…”
Huang: Corporate org charts wrongly invert management spans of control
“The number of direct reports of CEOs are very few. And the direct reports of the people who are just learning how to manage first level managers are very large. It's exactly the opposite of how it should probably be architected.”
Huang has over 40 direct reports and does zero one-on-ones
“Yeah, I have 40 some odd direct reports. And no one-on-ones no career coaching, you know?”
Huang: NVIDIA maintains only one instruction set and computer architecture
“We have one instruction set. We have one computer architecture, and we're super disciplined about that.”
Huang: Language-driven robotic foundation models will be discovered
“I have every confidence that a robotic foundation model will be discovered, and that, that through, through expressing yourself using human language you could cause a mechatronic system of almost different types of limbs and, you know, agility to be able to fi…”
Huang predicts major robotics breakthroughs are five to ten years away
“But I think it's probably less than, my guess is going to be less than 10 years, probably about five years, and, you know, I think you're going to see some pretty amazing robots.”
Huang: Ignorance is a founder superpower that cannot be regained
“I think ignorance is one of the superpowers of an entrepreneur, and you'll never get it again.”