20VC: NVIDIA vs Groq: The Future of Training vs Inference | Meta, Google, and Microsoft's Data Center Investments: Who Wins | Data, Compute, Models: The Core Bottlenecks in AI & Where Value Will Distribute with Jonathan Ross, Founder @ Groq
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch
The Twenty Minute VC
4.4 • 693 Ratings
🗓️ 17 February 2025
⏱️ 81 minutes
🧾️ Download transcript
Summary
Jonathan Ross is the Founder & CEO of Groq, the creator of the world's first Language Processing Unit (LPUTM). Prior to Groq, Jonathan began what became Google's Tensor Processing Unit (TPU) as a 20% project where he designed and implemented the core elements of the first-generation TPU chip. Jonathan next joined Google X's Rapid Eval Team, the initial stage of the famed "Moonshots Factory", where he devised and incubated new Bets (Units) for Google's parent company, Alphabet.
In Today's Episode We Discuss:
04:20 Interview with Jonathan Ross Begins
04:59 Scaling Laws and AI Model Training
06:22 Synthetic Data and Model Efficiency
12:01 Inference vs. Training Costs: Why NVIDIA Loses Inference
17:06 The Future of AI Inference: Efficiency and Cost
18:15 Chip Supply and Scaling Concerns
20:57 Energy Efficiency in AI Computation
25:40 Why Most Dollars Into Datacenters Will Be Lost
31:05 Meta, Google, and Microsoft's Data Center Investments
41:11 Distribution of Value in the AI Economy
42:10 Stages of Startup Success
43:17 The AI Investment Bubble
45:00 The Keynesian Beauty Contest in VC
48:40 NVIDIA's Role in the AI Ecosystem
53:39 China's AI Strategy and Global Implications
57:51 Europe's Potential in the AI Revolution
01:10:14 Future Predictions and AI's Impact on Society
Transcript
Click on a timestamp to play from that location
| 0:00.0 | We did not raise $1.5 billion. That's revenue. That's actually about 30% of the revenue of Open AI. |
| 0:05.0 | Your job is not to follow the wave. Your job is to get positioned for the wave. |
| 0:10.0 | You can almost say we're one of the best things that's ever happened to in the video because they can make every single GPU that they were going to make |
| 0:15.0 | and they can sell it for training, high margin, gets amortized across deployment, and we'll take the low margin high volume |
| 0:21.1 | inference business off their hands and they won't have to sell either margin. |
| 0:24.2 | We're growing faster than exponential. |
| 0:27.1 | And when you are growing faster than exponential, there is no amount of profit that you can |
| 0:31.1 | make that matters. |
| 0:32.4 | What matters is getting a toehold in the market and becoming relevant. |
| 0:42.9 | This is 20VC with me, Harry Stebbings, and today we feature a company that has just booked $1.5 billion in revenue. They say very simply, Nvidia should own the training market for AI, |
| 0:50.0 | and they will own the inference market. Simple. This is an exceptional discussion recorded in Paris last week with Jonathan Ross, |
| 0:58.0 | founder and CEO of GROC, the creator of the world's first language processing unit, LPU. |
| 1:04.0 | And prior to GROC, Jonathan began Google's tensor processing unit, TPU, |
| 1:08.0 | and implemented the core elements of the first generation TPU chip at Google. |
| 1:13.4 | But before we dive in today, turning your back of a napkin idea into a billion dollar startup |
| 1:18.8 | requires countless hours of collaboration and teamwork. It can be really difficult to build a team |
| 1:24.6 | that's aligned on everything from values to workflow, but that's exactly |
| 1:28.2 | what Coda was made to do. Coda is an all-in-one collaborative workspace that started as a napkin |
| 1:33.8 | sketch. Now, just five years since launching in beta, Coda has helped 50,000 teams all over the world |
| 1:41.0 | get on the same page. Now, at 20VC, we've used Coder to bring structure to our |
| 1:45.6 | content planning and episode prep, and it's made a huge difference. Instead of bouncing between |
| 1:50.5 | different tools, we can keep everything from guest research to scheduling and notes all in one |
... |
Please login to see the full transcript.
Disclaimer: The podcast and artwork embedded on this page are from The Twenty Minute VC, and are the property of its owner and not affiliated with or endorsed by Tapesearch.
Generated transcripts are the property of The Twenty Minute VC and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.
Copyright © Tapesearch 2026.