Moonshots EP 205: OpenAI Going Public, China Catching Up, AI Reshaping S&P 500 and Jobs

Summary

Weekly roundup episode with Diamandis, Dave Blundin, and Alexander Wissner-Gross, recorded after returning from FII in Saudi Arabia. The panel covers OpenAI’s trajectory to $100B revenue (reached faster than any company in history — 2.5 years vs Nvidia’s 8, Google’s 10), with Wissner-Gross predicting OpenAI could hit $100B ARR by 2027 primarily through 24/7 agent workloads. A key chart shows S&P 500 and total US job openings decoupled in late 2023 — the panel debates whether this marks the historic beginning of labor-capital decoupling or is just post-COVID normalization. Blundin notes college graduates are already bifurcating sharply: AI-skilled grads get strong offers, everyone else struggles despite record S&P levels. Nvidia hits $5T market cap (equal to Saudi Arabia’s total asset value), prompting discussion on whether compute scarcity is sustainable or will diffuse across Broadcom, AMD, Qualcomm. Geoffrey Hinton’s shift toward AI optimism is discussed — he proposes building “maternal instinct” into superintelligence, which Wissner-Gross critiques as naive, preferring instrumental convergence arguments (James Miller’s “Reasons to Preserve Humanity” on LessWrong). The episode also covers US dominance in data centers (5,426 vs rest of world combined), deepfakes of Jensen Huang outperforming real content, and China’s competitive position in AI.

Key Segments

[00:01-08:00] FII Saudi Arabia recap — anecdotes with Eric Schmidt, Ray Dalio, Balaji, Bob Mumgard (Commonwealth Fusion — commercial reactor by 2032)
[08:00-11:00] OpenAI revenue trajectory — fastest to $100B ever, 800M subscribers, commerce monetization as growth driver
[12:00-14:00] US data center dominance (5,426 vs world), Nvidia at $5T (equal to Saudi Arabia’s total assets)
[16:00-21:00] S&P 500 vs job openings decoupling — COVID artifact or historic labor-capital split?
[22:00-26:00] Geoffrey Hinton’s optimism shift, “maternal instinct” for AI, orthogonality thesis critique

Notable Claims

OpenAI at $100B revenue in 2.5 years — fastest in corporate history
Nvidia $5T market cap equals Saudi Arabia’s total national asset value
US has 5,426 data centers — more than rest of world combined (though this counts all data centers, not just AI)
Commonwealth Fusion CEO Bob Mumgard: commercial fusion reactor by 2032, ~400 MW output
College grad job market already bifurcated: AI-skilled vs everyone else

Bias/Framing Notes

Heavily techno-optimist framing as usual. The Saudi trip recap occupies the first 8 minutes and reads as social proof / name-dropping. The S&P/jobs decoupling chart is presented provocatively but Wissner-Gross correctly notes it may just reflect Fed rate cycles and COVID normalization. Data center count comparison is misleading (raw count, not compute capacity).