The number is staggering: $80 billion in annualized revenue across just 34 companies. Six months ago, that figure was around $38B. That's a doubling in less than a year—a pace of growth that makes even the most aggressive SaaS curves look modest. But the real story isn't the $80B. It's who captures it.
The 89% Problem
According to analysis from The Information, OpenAI and Anthropic now account for roughly 89% of all revenue generated by the leading AI startups. Not 89% of the market cap. 89% of actual revenue flowing through the door today.
$80B in annualized revenue sounds huge—until you realize two companies take nearly all of it.
This concentration isn't a fluke or a temporary quirk. It's a structural feature of how foundation model companies scale. Each new user costs almost nothing in marginal compute once the model exists. Enterprise contracts lock in at \$500K+ annually. Usage grows linearly while costs stay flat. The winner-take-most dynamics that defined consumer software have arrived in enterprise AI with a vengeance.
What's Left for Everyone Else?
The remaining 32 companies on that list—the well-funded others, the promising YC grads, the research-heavy upstarts—are fighting over roughly $9 billion combined. That's still a massive market, but it's a rounding error compared to the two giants. And the gap is widening, not narrowing.
Consider what's driving this split: Agentic workflows reward models that already have product-market fit and distribution. OpenAI has ChatGPT at 400+ million users. Anthropic has Claude, beloved by developers for its reasoning quality. Everyone else is still proving they belong at the table.
The real risk isn't that OpenAI and Anthropic are too big. It's that the funding math stops working for anyone else. When investors see 89% concentration in two players, the appetite to fund player #7 evaporates. The next fundraising cycle for mid-tier AI companies could be brutal.
TEXXR signals show the co-occurrence spike between Anthropic and OpenAI hitting a 90-day high. They're not just winning—they're the only two names in the conversation. Competitors exist, but they're not being talked about the same way.
The $80B question isn't whether AI revenue is real. It is. The question is whether the industry accepts this duopoly as the new baseline—or whether someone finds a way to crack it open.