Moonshots EP 181: AI Is Making More Millionaires Than Anything in History

Summary

A WTF episode with Peter Diamandis, Dave Blundin (Link Exponential Ventures), and Salim Ismail (Open Exo) covering the unprecedented AI startup boom and corporate innovation dynamics. The headline: 36 AI unicorns created in the first half of 2025, with median time to $1M ARR collapsing from 16 months (pre-2020) to 5 months, and time to $5M ARR from 41 months to 13 months — a 3-4x acceleration. Dave Blundin contextualizes this from personal experience: his first company took 9 years to reach a $1B valuation (1991-2000), which was considered fast. The panel highlights that headcount-to-value ratios have fundamentally changed — Cursor and similar teams reach multi-billion valuations with 30-50 people, compared to the 200+ needed a generation ago. Vibe coding startups (Cursor at $500M ARR, Lovable at $17M, Bolt at $20M) form the base layer enabling more startups on top. Dave describes Link XPV’s seed-stage approach: $0.5-2M opening checks, $7-10M per company over time, with a perfect record on MIT/Harvard teams that stick together (only failures come from team defections). The second half covers Apple’s desperation play — Bloomberg reports Apple is evaluating Anthropic or OpenAI to power Siri, with Anthropic demanding billions that double annually. Salim delivers a masterclass on corporate innovation: across 200+ Fortune 500 companies studied over 15 years, the only model that works for disruptive innovation is edge-separated teams with independent infrastructure, citing Walmart’s four failed attempts to compete with Amazon before finally succeeding on the fourth try by keeping the team on the edge until critical mass. The panel also covers Sam Altman confirming GPT-5 “sometime this summer” while Poly Market prices it at 26% by July 31 and 93% by year-end.

Key Segments

• [00:00-11:00] 36 AI unicorns in H1 2025, 4x revenue acceleration, headcount-to-value ratio collapse, company-of-one billion-dollar thesis
• [11:00-18:00] Vibe coding base layer (Cursor, Lovable, Bolt), vibe valuations problem, seed-stage investing strategy
• [18:00-25:00] GPT-5 summer timeline, Grok 4 benchmarks recap, AI as Jarvis (brilliant tool awaiting human direction)
• [25:00-35:00] Apple/Anthropic Siri deal, innovator’s dilemma masterclass, Walmart’s four failed Amazon-competition attempts

Notable Claims

• 36 AI unicorns in H1 2025; median time to $1M ARR collapsed from 16 months to 5 months
• Link XPV has never had a loss on MIT/Harvard/Northeastern founding teams except when a member defects
• Nvidia surpassed Apple in market cap; Apple had a third of TSMC capacity and was better positioned to become Nvidia but missed the opportunity entirely
• Anthropic wants billions from Apple that double annually to power Siri with Claude
• Salim: across 200+ Fortune 500 companies, edge-separated teams with independent infrastructure is the only model that ever works for disruptive innovation