Summary
This episode of the All-In Podcast — featuring David Friedberg, Brad Gerstner, and David Sacks — leads with major upheaval at Google DeepMind. Demis Hassabis has been moved to chair of DeepMind and chief scientist at Google, with Axios reporting the shift is partly a consequence of low morale that has contributed to Gemini 3.5 Pro running months behind schedule. Adding to the pressure, Jeff Dean — employee number 30 at Google with 27 continuous years at the company — is departing alongside the Gemini co-lead and several other top researchers to found Discovery Loop, focused on deep scientific breakthroughs. Google’s share price dropped roughly 4% on the Dean news alone.
The hosts dig into the structural question underneath the personnel moves: whether it still makes sense for frontier labs to spend tens of billions competing at the model level when open-weights models are closing the capability gap rapidly. Gerstner and Sacks argue that capital is increasingly better deployed in compute infrastructure — where demand is predictable and returns are high — than in model research, where commoditization risk is growing.
The episode also covers SpaceX’s strong quarterly performance, Airtable’s reported 90% valuation collapse, and the geopolitical implications of US data being used to train Chinese AI models. For anyone tracking the competitive dynamics between major AI labs and the emerging bifurcation between closed and open-weights development strategies, this episode offers pointed analysis from investors with direct visibility into these companies.
📺 Source: All-In Podcast · Published August 08, 2026
🏷️ Format: Podcast
