A seed round is supposed to be the small one. In 2026 the definition has stopped holding: more than 40 percent of all seed and Series A investment globally has gone into financings of $100m or more, and among American startups that share has passed half. The largest single seed round of the year, raised by a company called Humans&, was $480m.
That is not a rounding error in a dataset. It is a structural change in how the earliest stage of venture capital now works, and it sits directly on top of a seed market that is doing fewer deals than it was two years ago. For a founder raising a first or second round, those two facts pull in opposite directions.
When a Seed Round Buys a Data Centre
Crunchbase data on early-stage megadeals puts the 2026 year-to-date figure above 40 percent of seed and Series A dollars going to rounds of $100m and up. The individual deals explain the arithmetic. Ricursive Intelligence raised a $300m Series A at a $4bn valuation. Merge Labs, founded by Sam Altman and backed by OpenAI, took $252m at seed. Mal, based in Abu Dhabi, raised $230m in its first funding round. Upscale AI closed a $200m Series A.
These are sums that, a decade ago, would have been growth rounds for companies with revenue, customers and a repeatable sales motion. They are now being committed to businesses with none of those things, on the basis of a team and a technical thesis.
The pattern is not new so much as returning. Early-stage megadeals grew around the 2021 market peak, fell away over the two years that followed, picked up again in 2024, and have come back hard this year. What is different in 2026 is the share: the giant round has moved from an outlier to a meaningful slice of the whole category.
Britain Took 39% of European Venture Capital
The UK figures show the same shape at national scale. British startups raised roughly $17bn in the first half of 2026, the strongest half since 2022 and close to double the same period in 2025. The first quarter alone accounted for $7.8bn, up 60 percent year on year.
That performance gave the UK 39 percent of all European venture capital raised in the half, which is a substantial share for one market. It is also a figure worth reading carefully, because a large national total assembled from a small number of very large cheques behaves differently from the same total spread across hundreds of companies.
Four rounds crossed $1bn between January and June: Isomorphic Labs at $2.1bn, Nscale at $2bn, Wayve at $1.2bn and Ineffable Intelligence at $1.1bn. Between them those four deals came to about $6.4bn, or well over a third of everything British startups raised in the half. Remove them and the picture is a considerably more ordinary year.
This is the reading problem with a national funding total. It is an aggregate, and an aggregate built from a heavily skewed distribution tells you very little about the median company inside it. A regional investment agency quoting the $17bn figure and a founder in Leeds trying to close £800,000 are both describing the same market accurately, and their experiences of it have almost nothing in common. The same caution applies to the European comparison: the UK’s 39 percent share reflects where the largest cheques were written, not how many British companies were funded relative to French or German ones.
The Seed Market Underneath Is Getting Smaller
The other half of the story rarely makes a headline, because a falling number is harder to promote than a record one. Across the 2025 calendar year, UK seed deals numbered 704, a fall of 27 percent. Seed value held roughly flat at £2.1bn. Fewer companies split a similar amount of money, which means the average cheque grew while the number of founders who got one shrank.
The trend carried into this year. UK early-stage funding in the first quarter of 2026 came to £1.5bn across 418 deals, down 43 percent on value. Meanwhile the median gap between rounds stretched from 12.4 months to 14.4 months, so the cash a company raised had to last roughly two months longer than the previous cohort planned for.
Two extra months of runway is a real operating constraint. It shows up as delayed hiring, a slower product roadmap, and in some cases a bridge round on terms the founder would not have accepted with more time. None of that appears in a headline funding total.
Why Artificial Intelligence Is Doing Almost All of It
The concentration has a single dominant cause. The preponderance of early-stage megadeals are going to AI companies, and the UK breakdown makes the scale of that plain: around $12.6bn of the $17bn raised in the first half went to AI businesses, roughly 75 percent of the total, and more than four times the equivalent figure in 2025.
There is a straightforward capital-intensity argument behind it. Training and serving large models requires compute bought up front, which is closer to a manufacturing capital expenditure than to a traditional software cost base. A company whose first eighteen months involve buying or reserving substantial compute genuinely cannot run on £1.5m, whatever the stage label on the round says. Nscale, one of the four UK billion-dollar raisers, builds AI data centre capacity, which is about as capital-hungry as a startup gets.
Whether the returns justify the cheques is a question nobody can answer from here, and it is not one a news desk should pretend to settle. What can be said is that the structure of the financing follows the structure of the cost base, and that the cost base in this sector is unlike the one venture capital spent twenty years optimising for.
What Changes for a Founder Raising Now
The practical consequence is that “seed round” has stopped describing a size. It describes a position in a company’s life, and the amount attached to it now varies by two orders of magnitude depending on the sector.
For a founder outside AI infrastructure, the relevant numbers are the deal-count figures rather than the record totals. Fewer rounds are being done, they are taking longer to arrive, and the money that makes the aggregate look strong is largely committed elsewhere before the process starts. Planning against the headline total is how a company ends up short at month fourteen.
There is a second-order effect on everyone else. Venture funds raised on a given fund size have finite partner attention, and a fund writing a $200m cheque is running a different diligence process, on a different clock, from one writing twenty $10m cheques. Concentration at the top of the market does not only move money; it moves the time of the people allocating it.
That shift also changes what a first meeting is for. When a fund’s return is expected to come from a small number of very large positions, the screening question becomes whether a company could plausibly absorb a nine-figure cheque later, rather than whether it can reach profitability on a modest one. Businesses with sound unit economics and a realistic ceiling are not failing that test on merit; they are answering a question nobody asked them. Founders in that position have generally had more success with revenue-based lenders, regional funds and angel syndicates, whose mandates have not moved in the same direction as the headline seed and Series A figures.
The Numbers Worth Watching Next
Three measures will show whether this is a durable change or a phase of the cycle. The first is the deal count at seed, which is the cleanest read on whether new companies are being funded at all. The second is the gap between rounds, which turns directly into runway pressure and was already stretching before this year’s totals were recorded. The third is the share of early-stage dollars going to a single sector, currently around three-quarters in the UK.
If deal counts recover while the giant rounds continue, the market has simply added a new tier at the top. If they keep falling, the aggregate totals are measuring something narrower than the health of the startup economy they are usually quoted to describe. That distinction matters to anyone reading a funding figure as a signal about their own sector, in much the same way that strong hiring numbers sat alongside slowing output in UK manufacturing earlier this year. Two true figures can describe two different markets.
For now the split is the finding. Record money is going into early-stage companies, and fewer early-stage companies are getting any of it. Both sentences are supported by the same datasets, and neither one is the whole picture on its own.
About The Author
More Stories
PizzaExpress Closed 73 Sites and Now Wants a Thousand
Employer Health Cover Climbs 10% While Europe Averages 8%
John Ternus Begins at $4tn Where Cook Began at $350bn