Venture Capital Is Concentrating Faster Than Ever. What Happens To Everyone Else?

1966woodenghost May 19, 2026 0

Capital concentration in the private markets is accelerating. Companies with breakout growth or experienced founders in compelling sectors are raising funding at a faster clip, while the rest of the market is increasingly left behind.

In 2025, 70% of U.S. funding — more than $200 billion — was invested in 389 companies that raised rounds of $100 million and over, Crunchbase data shows. Of that, $90 billion went to just six companies that each raised more than $5 billion last year.

Contrast that with around 6,000 companies that raised the remaining 30% of U.S. venture capital in 2025, a total of $88 billion in rounds starting anywhere from $1 million to less than $100 million.

Those numbers point to 2025 as the year of the most U.S. venture capital concentration on record, with investment clustering in top private companies even more so than it did in the previous peak year of 2021, when startup funding more than doubled year over year.

The capital concentration at the top did not come entirely at the expense of smaller startups, however. While private-market investment accumulated into the larger rounds, funding to the sub-$100 million rounds did not decline, but increased around $8 billion to roughly the same number of companies last year.

Looking back at 2021

In 2021, when capital concentrated into the largest rounds, 60% of capital went to companies that raised rounds of $100 million or more.

The difference between 2021 and 2025 was that in the earlier year, a greater portion of capital went to companies in the $100 million to $500 million range — around 770 companies — while in 2025 a greater portion of capital went to just 50 companies in rounds of $500 million and over.

Further concentration in 2026

Capital concentration shows no signs of abating this year.

Already just through April, U.S. venture capital totals in 2026 are on par with funding for all of 2025, and 80% of startup investment this year so far has gone to rounds of $500 million and more, across 29 companies.

A growing venture capital market

Crunchbase data indicates that the private markets are concentrating heavily into the largest companies, but also growing overall for startups across the size spectrum, albeit much more modestly for all but the largest companies.

While more funding and value accrued to the top in 2025, the rest of the market also increased year over year, with the exception of rounds between $1 million and $10 million. Even at the smaller round size, however, funding and deal counts were down less than 10% year over year.

Looking ahead

The fundamental question for the venture world now is: Does the rapid growth and capital concentration for the largest companies come at the expense of smaller startups, or does the success of OpenAI, Anthropic and the like expand the total addressable market for tech startups so drastically that it promises to grow and reshape opportunities across the entire ecosystem?

“While Anthropic and OpenAI are absolutely amazing companies, by virtue of the capital they’ve raised, they are going to have to go after incredibly large markets,” Daniel Docter, managing director at Dell Technologies Capital, said during a recent Crunchbase News-hosted virtual panel discussion about agentic AI. “There’s just so much white space around that, where really interesting founders and startups can play.”

Madison Faulkner, partner at NEA Ventures, concurred. “This is a moment where I’m extremely excited about betting on seed and Series A, especially in spaces that do compete with Anthropic and OpenAI,” she said during the event. “I think they’ve really struggled to focus at this point in time, and they’ve been spending so much time trying to win lots of different use cases.”

Nnamdi Okike, co-founder of 645 Ventures, said he looks for startups to invest in that can build a defensible position in the new market: “If you’re very deeply embedded into a company’s workflow, where you’re doing many different tasks, not to mention you’re assembling a proprietary corpus of data that you can train your models on, we think that can be a sustainable moat over time.”