E132: SEC goes after crypto giants, Sequoia splits, LIV/PGA, Messi's deal + LIVE Q&A!

The SEC’s twin cases against Binance and Coinbase marked a shift from post-FTX cleanup toward a fight over whether crypto platforms could keep operating as they had in the United States. Chamath Palihapitiya expected fines, litigation and offshore relocation, while the panel sharply disagreed over whether enforcement or suppression was the real goal.

Sequoia’s breakup with its China and India/Southeast Asia arms was read less as portfolio housekeeping than as a structural retreat from an increasingly unwieldy global model. The panel tied China’s separation to geopolitical scrutiny around AI and chips, while Chamath considered the India split harder to justify because of India’s growth potential.

The PGA-LIV merger showed the leverage Saudi capital created by spending roughly $2 billion to recruit golfers, including a reported $200 million guarantee accepted by Phil Mickelson. The PGA had condemned LIV and sidelined defectors, yet its commissioner was then slated to run the combined entity, exposing how completely the economics changed the outcome.

Messi’s Inter Miami package was built like an ownership deal: Apple and Adidas reportedly gave him revenue participation tied to subscriptions and merchandise, not just salary. The hosts argued stars should similarly receive phantom equity in teams, citing Golden State’s rise from about $480 million to $5.3 billion during Stephen Curry’s era.

Making bank executives personally liable for depositor losses could reduce risk-taking, but Friedberg argued it would also change banking’s economics rather than simply make banks safer. If banks cannot earn spreads by investing deposits, they would shift toward charging customers custody and service fees—an outcome he said was already beginning after bank failures.

For AI investing, the clearest warning was that the middle of the stack may have almost no durable moat: a company valued at $500 million could be replicated weeks later by open-source software costing roughly $100,000. Chamath’s preferred barbell was silicon at the bottom and proprietary data providers at the top.

Friedberg argued that the deeper AI opportunity is not language generation itself but connecting probabilistic models to structured data and deterministic computation. That combination—LLMs for inference, software and mathematical systems for exact calculation—could make AI substantially more reliable and commercially useful across biology, chemistry, materials and other specialized domains.