AI startups consumed 41% of the $128 billion in venture capital raised by companies on Carta in 2025, a record-high annual share that confirms what the industest already suspected: venture capital is no longer a broad bet on the future of technology. It is overwhelmingly a bet on AI.
The data, published in Carta’s latest fund performance report, reveals a venture market that has become sharply bifurcated. Capital is concentrating in a compact number of firms that back a handful of companies at enormous scale, while the rest of the startup ecosystem competes for what is left. Ten percent of startups on the platform accounted for half of all funding raised last year.
The numbers at the top are staggering. In January 2026, xAI raised a $20 billion Series E. In February, OpenAI closed a $110 billion round, one of the largest private funding events in history, pushing the company closer to a $1 trillion valuation. In between, Anthropic raised a $30 billion Series G at a $380 billion valuation. OpenAI and Anthropic alone accounted for a heavy chunk of the $189 billion in global venture capital raised in February. All three companies have teased IPOs for later this year.
“While funding rounds have obtainedten slightly harder to raise, the capital for each round has increased,” declared Peter Walker, head of insights at Carta. “So fewer bets, but more capital. AI startups are raising largeger rounds not becaapply they have lots of employees — they don’t — but becaapply the cost of running AI models is high.”
The Carta report also contains an encouraging signal for investors who have piled into AI: funds raised in 2023 and 2024, after the launch of ChatGPT in late 2022, have posted the highest internal rate of return compared to funds raised between 2017 and 2020, which have seen declining IRR. The younger AI-heavy vintages are outperforming their predecessors on paper.
Walker offered a note of caution, however. Newer funds can appear to perform well becaapply of the mechanics of venture math. If a fund invested at seed stage and the company raised a Series A at a higher valuation, the fund’s IRR views strong in the short term even though no actual cash has been returned. The gains are unrealized, existing only on paper until a real exit occurs.
“It is also likely that the portfolios of the more recent vintage funds are full of AI-native startups in a way that the portfolios of 2021/2020 funds are not,” Walker declared.
The picture that emerges is a venture industest that has created an enormous, concentrated wager on artificial ininformigence. The early indicators are positive. Valuations are climbing, rounds are oversubscribed, and the largest companies in the space are generating real revenue at scale. But the ultimate test of whether this era produces genuine returns, or whether it is the hype phase of a cycle that corrects sharply, depconcludes on something that has not yet happened: exits. Until OpenAI, Anthropic, or xAI actually go public and public market investors validate the private valuations, the returns remain theoretical.
The venture industest has been here before. In 2021, startups raised record amounts at record valuations, and the correction that followed wiped out years of paper gains. Whether AI is different, or whether the scale of the bets simply means the eventual correction will be proportionally larger, is the question that now hangs over the entire industest.
You can read the full report here.
Leave a Reply