Which Company Is Going To Win The AI Race? - David Friedberg

The panel’s core thesis was that the five-year AI race may favor vertically integrated companies rather than model-only labs. Nvidia, Tesla and Google dominated their selections because each controls several layers of the stack—chips, models, distribution, data or physical products—rather than relying on a single AI product.

Tesla drew the strongest upside case because AI could extend from its own computing stack into cars, robotaxis and Optimus humanoid robots. Chamath argued that combining Tesla’s vision systems and hardware with xAI’s language and reasoning models could create an unusually complete physical-AI platform.

One speaker argued Tesla and xAI should ultimately be merged so Elon Musk is not splitting leadership and engineering talent between them. He estimated Tesla at roughly $1 trillion and xAI at $100 billion, saying a combination would unite Tesla hardware, X’s real-time data and xAI’s models under one direction.

Google’s strongest defense against declining traditional search is that it can shift monetization from “price per click” toward AI-driven value across its enormous user base. Chamath pointed to Gmail, YouTube, Workspace and search as distribution surfaces where Google could monetize AI while preserving or improving advertising effectiveness.

The speakers argued Google could remain economically powerful even if it loses search share because better AI could make each advertisement more valuable. They cited Google’s knowledge from Gmail, Chrome, Android, YouTube and conversational searches as data that could improve targeting and potentially increase ad-network performance.

Nvidia was treated as the most durable AI infrastructure business, but China was identified as its most serious structural threat. Freeberg cited reported Chinese advances in advanced semiconductor manufacturing and roughly $40 billion of investment, arguing that export restrictions are simultaneously creating stronger incentives for China to build alternatives to Nvidia’s ecosystem.

Google was also valued as a portfolio of independent AI-era bets rather than merely a search company. The discussion highlighted Waymo, quantum computing, Isomorphic’s biology work and Gemini, arguing that one major success outside search could offset deterioration in the legacy business.

The conversation’s broader technical conclusion was that AI products are moving away from a single dominant model toward systems combining multiple specialized models and agents. Google’s work in areas such as weather forecasting and graph-based models was presented as evidence of research depth that could later improve search, Gmail, advertising and YouTube.