The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch

20VC: "Anti-Data Centres is a Chinese Psyop" | How Many Planned Data Centers Will Actually Get Built? | Is Energy AI's Biggest Bottleneck? With Thomas Sohmers, Co-Founder @ Positron

September 19, 2026

AI Summary

5 min read

“A single golf course uses more water than the largest data centers in the United States.” That is the kind of factual reframe Thomas Sohmers, co-founder and chairman of Positron AI, deploys to push back against what he sees as a dangerous political consensus taking shape on both sides of the Atlantic. Sohmers joined Harry Stebbings on 20VC to argue that the anti-data-center sentiment sweeping Western politics is, in his view, “almost entirely a Chinese psyop”—a strategic gift to a competitor that faces no such regulatory friction. But the conversation ranged far beyond geopolitics. Sohmers offered a deep technical walkthrough of the inference infrastructure that powers generative AI, explained why the biggest model providers are actually wildly profitable despite the narrative of cash incineration, and made the case that the most important bottleneck for AI progress is not energy or chips, but human alignment—and the political will to build.

The Memory Wall and the Shift from Training to Inference

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

What you'll learn

  • 1 (04:20) **Positron's Place in the AI Stack** - Thomas Sohmers defines Positron as a semiconductor company building chips, software, and full rack-scale systems for generative AI inference.
  • 2 (04:58) **Training vs. Inference: The Fundamental Shift** - Sohmers explains that training is a compute-bound problem, while inference is heavily memory-bound due to its autoregressive nature.
  • 3 (07:12) **The Memory Wall: Why It Exists and Why It Matters** - Sohmers breaks down the divergence between compute and memory bandwidth improvements over the past decade.
  • 4 (10:56) **Token Economics: The Hidden Margins and Misconceptions** - Sohmers reveals how cache tokens create enormous margins for model providers, and why the narrative of unprofitable AI companies is wrong.
  • 5 (13:30) **Analyzing "Pacing the Frontier"** - Sohmers offers a skeptical take on the push to slow down AI development, arguing it poses two major risks.
  • 6 (17:26) **The China Question: Why Pacing is Naive if Competitors Don't** - Sohmers argues that Western leaders are naive to think they can safely slow down while China accelerates.
  • 7 (20:42) **The Anti-Data Center Movement: A Chinese Psyop** - Sohmers makes his most provocative claim, arguing that opposition to data centers is a strategic disadvantage for the West.

+ Full timestamped outline available in the app

Show Notes

Thomas Sohmers is the co-founder and chairman of Positron AI, building chips to make running AI dramatically cheaper and more energy-efficient. The company recently announced an $875 million Series C at a $5 billion valuation, backed by investors including Gavin Baker's Atreides Management, NEA, Valor Equity Partners and Netscape co-founder Jim Clark.

AGENDA:

04:30 Why Does AI Inference Need Different Hardware from Training?

10:25 What Is Nobody Telling You About AI's Token Economics?

13:10 Should We Really Slow Down the AI Frontier?

16:55 What Happens If America Slows Down—and China Doesn't?

19:30 Should We Restrict China's Access to AI Chips?

20:15 Why Aren't Zuckerberg and Jensen Backing an AI Slowdown?

21:25 Are We Being Misled About Data Centers?

23:25 Can the West Beat China While Drowning in Regulation?

26:20 How Many Planned Data Centers Will Actually Get Built?

27:15 Data Centers in Space: Real Opportunity or Elon Hype?

28:10 Is Energy AI's Biggest Bottleneck?

29:55 Could the Debt Boom Derail AI?

31:25 What Is KV Caching—and Why Does It Change AI Economics?

35:25 Does Compressing AI's Memory Make It Less Intelligent?

41:05 How Do We Solve AI's Exploding Memory Demands?

43:45 Will Every Company Own Its Own AI Model?

47:10 Why Bet on Ever-Bigger Models?

48:20 If We Already Have AGI, What Comes Next?

53:25 What Happens When Every AI Lab Builds Its Own Chips?

57:30 If DeepSeek Can Slash Costs with Software, Why Build New Hardware?

59:50 How Cheap Will AI Tokens Be by 2028?

1:03:15 What Would Be the First Warning Sign of an AI Bust?

1:03:55 Could Open Models Break OpenAI and Anthropic's Growth?

1:05:55 Could Mercor and Surge Become $200 Billion Companies?

 

The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch

More from this podcast

The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch →