All-In
Elon's Anthropic Deal, The Next AI Monopoly?, "FDA for AI" Panic, Trading the AI Boom
with Chamath Palihapitiya, David Sacks, and Brad Gerstner
9 May 2026
18 min read
2h 15m
TL;DR
Elon's deal to lease Colossus 1 to Anthropic transforms the AI landscape: it solves Anthropic's compute constraints while creating Elon Web Services (EWS) as a third hyperscaler competing with AWS and Azure. Anthropic is now on an exponential 10x revenue growth trajectory that could make it the most valuable company in history—but this raises hard monopoly questions about whether regulatory guardrails are needed before one AI lab dominates.
All-In is a weekly podcast featuring four prominent tech investors and entrepreneurs discussing the biggest stories in technology, business, and culture. The hosts bring deep expertise in venture capital, startups, and market trends, offering contrarian takes on everything from AI monopolies to regulatory capture.
Takeaways
1
Elon Web Services flips the AI infrastructure game By leasing Colossus 1 to Anthropic, Elon converts excess data center capacity into a $45B+ revenue stream while positioning SpaceX as a third hyperscaler. This sidesteps xAI's revenue problem (expensive training, no immediate products) and lets Elon subsidize Grok development with profits from Anthropic, Google, and others—a playbook that mirrors AWS's role in bankrolling Amazon's other bets.
2
Anthropic's 10x growth compounds into monopoly dynamics Growing from $10B to $30B to $44B ARR in four months, Anthropic is on track to hit ~$100B ARR by year-end and potentially $1T by 2027. At that scale and growth rate, network effects and first-mover advantage in coding AI create winner-take-most dynamics—but unlike traditional tech monopolies, this one controls the frontier model powering all downstream AI applications.
3
Regulatory framing risks hiding competitive consolidation Sacks warns that safety-focused regulation (an 'FDA for AI') can become political cover for monopoly formation, using the Rockefeller precedent: strict safety standards sound pro-consumer but entrench the leader who can afford compliance. The real risk isn't bad AI—it's one lab controlling 80%+ of the most important technology before antitrust has time to act.