Executive Summary
OpenAI missed growth targets while Anthropic secured $45 billion from Google and Amazon, revealing the first crack in the "compute equals revenue" thesis that has driven AI valuations. The critical insight: agents consume 50-100x more tokens than humans, but they select models based on performance, not human preference anchoring. This creates a fundamental shift where foundation models compete daily for agent selection rather than benefiting from user stickiness. Google emerges as the primary beneficiary through multiple vectors: owning Anthropic capacity via TPU bundling, maintaining search revenue independence, and controlling both sides of the AI infrastructure trade. The Anthropic deal validates Google's strategy of using custom silicon to capture NVIDIA's 70% gross margins while locking in foundation model partnerships. Meanwhile, private equity's first major AI-era casualty—Medallia's $5.1 billion equity wipeout—signals the death of traditional SaaS exit routes, forcing venture portfolios toward fewer, larger winners. The convergence of agent-driven model selection, infrastructure consolidation, and exit market compression creates a winner-take-most dynamic favoring hyperscalers with vertical integration capabilities.
Key Insights
what Rory O'Driscoll and Jason Lemkin said“More and more, the agent is going to choose what models and just what vendors we use. The agents are going to make the decision on everything.”
what Rory O'Driscoll and Jason Lemkin said“If you're going to add $90 billion in revenue capacity, someone between you and your partners has to find plus or minus $300 billion to buy chips, dig holes in the ground, build data centers”
what Rory O'Driscoll and Jason Lemkin said“They didn't weigh over leverage, they just weigh overpaid for it. You can't service 2 billion plus of debt on a 1 billion low growth company with a pre-AI story”