Acquired: NVIDIA CEO Jensen Huang
Category: Acquired Podcast (Deep Dives) · Duration: 90 min · ▶ Watch
Speakers: Ben Gilbert and David Rosenthal · Jensen Huang
Segments (13)
- 00:00 · Introduction
- The hosts introduce the episode and their interview with NVIDIA CEO Jensen Huang.
- 02:54 · The RIVA 128 Story
- Jensen recounts the near-death experience of NVIDIA and betting the company on the RIVA 128 chip.
- 12:00 · The Origins of CUDA
- The development of CUDA and how it democratized supercomputing for researchers.
- 17:27 · The AI Revolution and OpenAI
- How deep learning emerged as the killer app for GPUs and NVIDIA’s early involvement with OpenAI.
- 24:57 · Statsig Ad Read
- An ad read for Statsig, an experimentation and feature flagging platform.
- 27:13 · NVIDIA’s Organizational Structure
- Jensen explains why NVIDIA is organized like a computing stack rather than a traditional hierarchy.
- 35:46 · Crusoe Ad Read
- An ad read for Crusoe, a clean energy cloud provider for AI workloads.
- 37:13 · Zero Billion Dollar Markets and Mellanox
- The strategy of targeting non-existent markets and the importance of the Mellanox acquisition.
- 45:46 · AI Safety and the Future of Jobs
- Jensen discusses AI safety mechanisms and predicts AI will create more jobs through increased productivity.
- 55:46 · Building Platforms and the Role of Luck
- The challenges of building a platform company and how luck played a role in NVIDIA’s success.
- 01:05:46 · Blinkist Ad Read
- An ad read for Blinkist, a book summary service.
- 01:08:57 · Lightning Round and Reflections
- Jensen answers rapid-fire questions about his favorite sci-fi, his car, and his early career.
- 01:19:45 · The Hardship of Entrepreneurship
- Jensen admits he wouldn’t start a company again knowing how hard it is, calling ignorance a superpower.
Predictions (2)
- [01:01:00, Future] Every car will be a software-defined car.
- [01:02:40, Near term] AI will create more jobs by increasing productivity and prosperity.
Key Technologies (4)
- RIVA 128: The first fully 3D accelerated graphics pipeline for a computer.
- CUDA: Compute Unified Device Architecture, enabling general-purpose computing on GPUs.
- Deep Learning: A universal function approximator that learns from large datasets.
- InfiniBand: High-performance networking technology used to connect GPUs in data centers.
Companies Mentioned (10)
NVIDIA · Microsoft · IKOS · OpenAI · Statsig · Crusoe · Mellanox · Blinkist · LSI Logic · Sequoia Capital
Notable Quotes (3)
Mission is the boss. — Jensen Huang @ 30:08
It’s better to be approximately right than exactly wrong. — Jensen Huang @ 33:34
Ignorance is a superpower. — Jensen Huang @ 01:20:56
Key Topics
NVIDIA History · RIVA 128 · CUDA · Deep Learning · OpenAI · Organizational Design · Mellanox Acquisition · Entrepreneurship
Takeaways
- NVIDIA’s early survival depended on a massive bet on the RIVA 128 and using emulation to ensure a perfect first tape-out.
- CUDA was developed to democratize supercomputing and found its killer app in deep learning.
- NVIDIA’s organizational structure is designed like a computing stack, not a traditional command-and-control hierarchy.
- Anticipating future needs (‘zero billion dollar markets’) is crucial for long-term success.
- Ignorance of how hard entrepreneurship is can be a superpower, as knowing the true difficulty might deter anyone from starting.