The Blink Edition No. 9

The European private markets briefing you can't afford to miss. Deals, signals, and moves, decoded every Monday by SeedBlink.
‍

‍

When a founder tells me they lost money on a previous company, I hear something useful.

Losing money teaches a specific kind of attention. Once you have watched capital disappear, you look at the next decision with your eyes wide open. What I want to understand is whether they know what went wrong and why. If they do, they are unlikely to repeat it.

That is what experience actually consists of: a catalogue of mistakes already made and corrected. The difference between a founder at 25 and one at 50 is rarely intelligence. The older one has fixed the same problem three times, in three different ways, and knows which fix holds.

I have never asked anyone directly about their failures in a meeting. It shows anyway, in how precisely they describe what happened, in whether the explanation leans on bad luck or bad judgment, in the questions they have clearly already asked themselves.

Read correctly, scar tissue is one of the most reliable signals a founder can offer.
‍

‍
‍

The market pays for experience, and the premium is now measurable. It does not value every kind of experience equally, though.

According to PitchBook, the median deal size for European founders with at least one previous company is €4.3 million this year, 72% higher than the €2.5 million median for first-time founders. In AI, the median deal-size gap rises to 92%. Post-money valuations follow, at €17.2 million against €9.4 million, though step-ups round over round are identical at 1.5x for both groups. Investors are writing repeat founders bigger cheques rather than pricier ones.

That conviction is being institutionalised. Tapestry VC closed an $80 million fund in July focused on backing repeat founders, saying they have built around 60% of Europe's unicorns founded over the past eight years, and that companies led by second- and third-time European founders represent roughly $2.2 trillion in combined enterprise value.

The distinction matters here. Long-standing research by Gompers and colleagues, based on several thousand venture-backed companies, found that founders whose previous company went public had a 30% chance of repeating that outcome. Founders whose previous company failed sat at 22%, barely ahead of the 21% for first-time founders. The strongest premium goes to demonstrated success rather than to the scar itself.

So what is actually being bought? Partly track record, network and hiring speed, advantages repeat founders may retain even when the previous company did not succeed. Partly reputation, which compounds on its own. We would add a third factor that standard datasets struggle to capture: the difference between a founder who failed and one who understood why it happened.

That distinction rarely shows up in a spreadsheet, but it shows up in a meeting. A failure explained as bad luck reads as a risk still outstanding. A failure explained as a decision, with the reasoning attached, is the closest thing to evidence that the lesson was absorbed.

‍

What Q2 says about where European capital actually went

We went through the Q2 numbers from Crunchbase, PitchBook and KPMG, and the clearest signal is a market splitting in two. European startups raised $42B in H1, up 50% year over year, while North America pulled in $392B, up 158%. Deal volume fell 29% year over year here, with seed driving most of that decline, even as late-stage funding climbed 90%.

The pattern holds across the quarter: the market is hollowing out at the bottom and staying resilient at the top. Cleantech moved into third place among verticals, the Nordics had their strongest quarter since early 2021, and Ireland's investors kept backing profitability over frontier bets.

Read the full breakdown

‍

UK 🇬🇧

Callosum raises $100M seed

One of Europe's largest seed rounds ever, led by Atomico with the UK's Sovereign AI Fund making its first disclosed investment. Founded by two Cambridge neuroscientists, Callosum routes each AI task to the model and chip best suited to it, a bet against the assumption that progress comes from one model on identical hardware.

Read more

‍

Germany 🇩🇪

Oceanloop secures up to €38.5M

Munich-based Oceanloop combined equity from Hatch Blue and Stolt Ventures with a €32M venture-debt facility from the EIB to scale land-based fish farming, having become the first company in Europe to commercially farm Giant Grouper.

Read more

‍

Sweden 🇸🇪
‍
Pixelgen raises €13.26M Series B
‍
The Stockholm life sciences company closed an oversubscribed round led by Flat Capital to expand its Proxiome Kit, which maps the spatial architecture of proteins on individual cell surfaces at nanoscale resolution.

Read more

‍

Your startup can fail without making you a failed founder
‍
Nir Eyal on why failure is read as information inside Silicon Valley and as a verdict on the person nearly everywhere else, and what that difference does to whether founders try again.

Read more

‍

Marketplaces in the age of AI: threat or opportunity?
‍
FJ Labs' Fabrice Grinda argues the disruption is narrower than feared: browsing and intent-driven search are largely safe, thin commoditised transactions are not. Vinted's AI translation collapsed the language barrier that had siloed European e-commerce for years.

Read more

‍

When models learn

Today's models stop learning the day training ends. Test-time training would change that, and with it the unit economics: providers serving a separate model per user rather than one for everyone. Tunguz on where that cost pays for itself, and where personalisation turns into lock-in.

Read more

‍