Venture capital flowing into artificial intelligence is pooling in a very small number of places.
According to MarketScale, OpenAI and Anthropic now account for 80% of the venture funding raised across the AI 50 — a share that leaves the remaining companies on the list splitting what's left. MarketScale frames this against a second trend running at the same time: enterprise-ready AI startups are multiplying, meaning more companies are building AI products aimed squarely at business customers rather than consumers.
Those two facts sit in tension. The number of credible AI startups selling into enterprises is growing, but the capital is concentrating rather than spreading out. More companies are competing for a smaller slice of the same pot.
The reporting available here is limited to that headline finding — MarketScale's item, surfaced through Google News, does not detail the time period measured, the dollar totals involved, or how the AI 50 cohort was assembled. Those caveats matter for interpreting the 80% figure, and this brief does not go beyond what the source states.
Still, the direction is clear enough to be useful. Concentration at this level suggests investors are treating frontier model development as a winner-takes-most race, where the cost of training and running large models rewards scale and punishes the merely competitive. Startups building on top of those models — the enterprise-ready wave MarketScale describes — increasingly compete for customers rather than for the capital to build foundational technology themselves.
It matters because when four-fifths of the funding goes to two companies, the shape of the AI industry that businesses eventually buy from is being decided by a very small group of investors and founders.