The global AI startup sector is now generating close to $80 billion in annual revenue, a 112% increase over just six months, according to an analysis by The Information. This rapid growth is driven by a small group of 34 companies, but the benefits are heavily concentrated. Anthropic and OpenAI together account for 89% of the total revenue, with Anthropic recently overtaking OpenAI thanks to strong adoption of its AI coding tools. Despite the sector's expansion, most startups outside the top two remain far behind in revenue scale.

Beyond the leaders, only a few companies have crossed significant milestones: Perplexity, ElevenLabs, and Cognition have each surpassed $500 million in annual revenue. However, even Anthropic and OpenAI face immense cost pressures. Together, they spend over $30 billion per year, primarily on training advanced AI models. Anthropic shares its revenue with both Amazon and Google, while OpenAI is contractually obligated to return 20% of its revenue to Microsoft through 2030.

Investors like Sequoia point to these figures as evidence that the core economic value in AI is accruing to model developers rather than application-layer companies. The data underscores a widening gap between foundational model providers and the broader ecosystem of AI startups building on top of them.