Accel is a venture capital firm. It gives money to startups early, usually before anyone else has, and takes a stake in return. It has done this since 1983, and it wrote the first institutional cheque into Atlassian, CrowdStrike, Flipkart and Slack.
On Tuesday it said it had raised $3.5bn for four new funds, Bloomberg reported. All four were raised at the same time, which the firm has not done before in 43 years.
The split matters more than the total. A $1.35bn global expansion fund will write bigger early cheques and follow on quickly. Then $800m for the United States, $800m for Europe and Israel, and $550m for India.
Europe and America now get the same money
Both the US and the Europe funds have grown from $650m to $800m. A Silicon Valley firm founded in California is now raising the same amount for European and Israeli startups as for American ones.
That is the ninth Europe fund Accel has raised, Sifted reported, and the firm has been active in European defence and AI companies. Its European practice has been run since 2004 by Harry Nelis, a partner based in London.
The portfolio explains the confidence. Accel backs Anthropic, Cursor, Perplexity and Vercel, and it co-led the round that valued Stockholm’s Lovable at $13.3bn.
One caveat on the parity. The $800m covers Europe and Israel together, so it is not a purely European fund. Accel has said a partner’s proximity to the region lets it reach Israeli founders within hours.
India got less, and that is the honest bit
The India fund is $550m, which is $100m smaller than the one before it. It closed within weeks and was oversubscribed, TechCrunch reported.
It is also smaller because Accel has not spent the last one. More than 55% of the previous $650m India fund is still available, and the firm does not expect to start deploying the new money until 2027.
Raising a fund you will not touch for a year, while sitting on half the last one, is a decision about access rather than need. Partner Shekhar Kirani said investors preferred to assess the whole global platform in one process rather than several.
The thesis is narrower than it looks. Partner Prayank Swaroop said the early movers went to large language models, and the bigger opening for Indian startups is the application layer. Accel points to RapidClaims, which automates medical coding for US healthcare providers at about 95% accuracy.
India is not short of competition for that access. Peak XV raised $1.3bn, General Catalyst committed $5bn over five years, and OpenAI and Anthropic both call India their largest market outside the US. Cursor now runs localised India pricing.
The seed round is not a seed round any more
Here is what the fund structure is actually responding to. Last year Accel joined a $300m seed round in Periodic Labs, an AI scientific discovery startup, at a $1.3bn valuation.
Rounds like that have a name now. The trade calls them neolabs: companies that take enormous sums at inception to fund research rather than a product.
Thinking Machines Lab, founded by former OpenAI chief technology officer Mira Murati, is another Accel company. It is approaching a $50bn valuation after a $2bn seed.
Steve Loughlin, a partner in the San Francisco Bay Area, was blunt about what that does to a portfolio. “You can’t really construct a fund of just those types of deals,” he told Bloomberg.
So the $1.35bn expansion vehicle exists to let a firm with deliberately modest early-stage funds split a large cheque between two pockets. It can keep the discipline of small funds and still turn up to rounds those funds could not fund alone.
Nelis put the cost of that plainly. “Companies raise more money, more quickly, earlier in their company life than ever before,” he said. The opportunity is bigger, and “the risk is still pretty much the same”.
Two companies are absorbing the market
The backdrop is a venture market that has stopped behaving like one. Global funding hit a record $510bn in the first half of 2026, according to Crunchbase figures cited by Tech Funding News.
OpenAI and Anthropic together took more than 40% of it. The desk has tracked the same concentration in the US, where megadeals took 87.5% of venture dollars in the second quarter.
Accel is on both sides of that. It is an Anthropic backer, so it benefits from the concentration, and it is raising early-stage funds in a market the concentration is squeezing.
It is not alone in trying. Sequoia is putting $10bn behind AI and reindustrialisation, and Yann LeCun has left Meta to run a new AI fund.
The money is moving into physical things
Loughlin said Accel has spent two years widening from AI software into deep tech, including material science and manufacturing. Other firms have made the same turn, betting AI can rebuild complex industries in ways ordinary software could not.
The newer names in the portfolio show it. Pulse 2.0 lists RadixArk, which builds open-source inference and training systems, and Fractile, which designs chips for inference workloads.
Not everything is expansion. Daniel Levine, one of the firm’s better-known early-stage investors and a backer of Scale AI and Vercel, will not make investments from this fund or future ones. He remains a partner and keeps his board seats.
What this actually tells you
Add the April raise and the picture sharpens. Accel closed $5bn for late-stage bets three months ago, split between a $4bn Leaders fund and a $650m sidecar. That is roughly $8.5bn of new capital since the spring.
A firm that writes the first cheque in about 80% of its investments has now built the machinery to keep writing cheques all the way to an IPO. That is either admirable persistence or an admission that staying early is no longer possible on its own.
The test is not the fundraise. It is whether the Europe fund, now equal in size to the American one, produces companies that need the expansion vehicle. If it does, the parity was real. If the big cheques keep going to California and Israel, it was a marketing document with a European label on it.
Get the TNW newsletter
Get the most important tech news in your inbox each week.
