Seed Funding Records Fall as Callosum Lands Atomico

A company founded barely a year ago just raised nine figures before most startups finish their first sales hire, and Callosum closed a $100 million seed round on August 20, 2026. Atomico led the deal, with Plural, DCVC and the UK Sovereign AI Fund joining. It stands as one of the largest seed rounds ever raised by a British startup.

That headline invites a reasonable question from anyone building a normal company: has seed funding changed, or is this simply a category exception? The honest answer is both, and the distinction matters when you set your own raise target. Founders who misread rounds like this one tend to ask for money the market will not give them.

What Callosum Raised and Why

Callosum was started in 2025 by Danyal Akarca and Jascha Achterberg. The company works on what it calls heterogeneous intelligence, meaning it orchestrates AI workloads across different chips and different models to improve cost, energy use and performance.

The round also marks the first disclosed investment from Britain’s sovereign AI vehicle, which was set up with roughly £500 million to back domestic AI capability. Government money entering at seed stage is unusual and worth noting on its own.

Why Capital-Heavy Categories Break the Rules

Think of seed size as a function of what the first eighteen months physically require. Software companies can reach a demo with a small team and a cloud bill, so a $2 million to $5 million seed remains normal.

Infrastructure is different. Silicon access, compute contracts and specialized engineering talent all cost real money before a single customer signs, which is why these rounds look nothing like a typical seed. The same dynamic showed up in the concentration of AI startup funding across Europe this year.

So read the round as a statement about capital intensity, not about founder quality. Raising more than you need in a low-capital business mostly buys you a harder Series A.

How Founder Credibility Substitutes for Traction

Pre-product rounds still get underwritten on something. Usually it is a founding team with rare, verifiable expertise in the exact problem, plus a thesis an investor already believes.

You can build a version of that without a research background. Document a specific insight you hold that competitors do not, show the work behind it, and let a credible reference vouch for your judgment. That is the practical substitute for a track record, and it aligns with the early-stage fundraising rules that experienced investors rarely state out loud.

Be realistic about the odds. Rounds like this are rare, and planning your company around one is closer to a lottery ticket than a strategy.

Setting a Seed Number You Can Defend

Work backwards from milestones instead of forwards from ambition. Decide what has to be true for a Series A investor to say yes, price the eighteen to twenty-four months required to get there, then add a buffer.

Dilution is the other constraint. Selling too much of the company at seed limits what you can offer employees later and weakens your position in every subsequent negotiation. Many founders should also weigh money that costs no equity at all, since grants and competitions offer real non dilutive funding for early teams.

A defensible number sounds boring in a pitch and reads well in diligence. That tradeoff is usually worth making.

Signals to Track in European Early-Stage Deals

Watch whether other sovereign or state-backed funds start appearing on seed cap tables. If public capital keeps entering at this stage, round sizes in strategic categories will stay elevated, and private investors will price accordingly.

Also watch what Callosum ships. A large seed buys time, not proof, and the next milestone is working technology in customer hands. Investors across the category will treat that outcome as a reference point for future deals.

Seed Round Questions Founders Ask

How big is a normal seed round?

For most software startups it falls in the low single-digit millions. Very large seeds cluster in capital-intensive fields like chips, energy and biotech.

Can I raise a seed with no product?

Sometimes, though it usually requires deep domain expertise, a strong reference network, or a category where building anything demands capital first.

How much equity should a seed round cost me?

Many founders target roughly 15% to 25% for a seed, but the right answer depends on the amount raised and your realistic path to the next round.

Use this deal as calibration rather than aspiration. Raise the amount your specific milestones require, protect your ownership, and let capital intensity, not headlines, set your target. Details on the lead investor’s portfolio and approach are published by Atomico.