Bison Ventures says billion-dollar AI seed rounds are still the exception

Billion-dollar seed rounds in AI are getting the market’s attention, but Bison Ventures says the basic math of venture investing has not changed. For investors watching private companies become future IPO candidates, the key issue is entry price: the more early backers pay upfront, the less room there may be for returns to multiply.

Ellie McDonald, a principal at Bison Ventures, pointed to recent AI financings as examples of the new headline cycle. Yann LeCun raised $1 billion for a new company, Project Prometheus launched with $6.2 billion, and Unconventional AI raised $475 million two months after being founded, according to Crunchbase-linked company data cited by McDonald.

McDonald argues those deals should not be read as proof that seed investing has been permanently reset. A seed round is usually a startup’s first major outside financing. In venture capital, the goal is often to buy ownership early enough that a later sale or initial public offering can turn a small check into a much larger outcome.

The biotech comparison

Bison Ventures has studied biotech, a sector where very large first rounds have been around for longer. McDonald wrote that biotech companies often need substantial money early because clinical trials are expensive. A Phase 1 trial, the first stage of testing a drug in humans, cannot usually be run on a small software-style seed budget.

That need for capital does not automatically lead to stronger investor returns, according to McDonald. Bison Ventures compiled publicly available data on roughly 200 first financing rounds of $100 million or more over the past 15 years. The firm found that 20% of those companies had recorded exits, meaning a sale or public-market listing.

Only a small subset of those exits produced what McDonald described as venture-like performance: at least 10 times MOIC. MOIC, or multiple on invested capital, measures how many dollars an investor gets back for each dollar invested. By Bison’s count, about 1% of companies that raised at least $100 million in their first round produced returns high enough to support the venture model.

AI may add winners, but price still matters

McDonald said the dataset could improve if leading AI companies deliver strong exits. She wrote that OpenAI and Anthropic would about double the number of outlier returns in the dataset when they exit. Citing reports, she said OpenAI’s projected IPO valuation could mean 30 to 40 times returns for first-round investors.

That would be a strong result, but McDonald compared it with earlier venture wins where entry prices were much lower. She cited reported returns for Sequoia Capital and Kleiner Perkins on Google of more than 300 times, after each firm turned about $12.5 million into roughly $4 billion. She also cited reports that First Round Capital turned about $500,000 in Uber into $2.5 billion, or nearly 5,000 times.

Her point is that company quality is only one part of the outcome. The valuation paid in the first round can determine how much upside remains for early investors if the company later becomes very large.

Mega-seeds are still a small slice

McDonald wrote that seed rounds of $50 million or more have increased sharply since 2018, while traditionally sized first rounds have also grown. She said the largest deals remain a small share of funded startups and an even smaller share of the companies likely to return venture-scale capital.

Several current AI companies began with smaller first rounds, according to figures cited by McDonald. Cursor raised less than $10 million, ElevenLabs raised $2 million, Legora raised $11 million, Sierra raised $25 million, and Cohere raised $5 million. McDonald said each is now valued above $5 billion and generating hundreds of millions in revenue.

For retail investors, the takeaway is less about predicting which private AI startup wins and more about understanding the financing math before those companies reach public markets. Large early rounds can fund ambitious plans, but Bison Ventures’ analysis suggests capital intensity alone has not historically produced the highest venture returns.

This story draws on original reporting from Crunchbase News.