Fewest deals in six years: why every startup pick now gets cross-examined.

TL;DR

This year, European investors put more money into fewer startups, with the total deal count at its lowest since 2020.

Fewer deals means more questions, and a pitch deck often doesn't answer the hardest ones: is the market real, has the revenue been checked, and who are the competitors that weren't mentioned?

In a recent evaluation of a pre-seed diagnostics startup, the investor's most useful insight came from the claims no source could back up, so those became the agenda for their next founder call.

A record that most funds did not feel

If you only read the headlines, 2026 looks like the best year venture capital has ever had. Investors put more than $560 billion into startups worldwide in the first half, and two rounds explain a large part of it: OpenAI raised $122 billion in the first quarter and Anthropic $65 billion in the second. Together they took about a third of all venture money raised in those six months.

Europe had its own version of the same story. Funding recovered to €44.1 billion in the first half, well above a year earlier, while the number of deals fell to just over 1,740, the fewest in six years. In the first half of 2024, €50.1 billion was spread over around 2,000 deals. In the first quarter of this year alone, the average round was about 17% larger than a year before. There is more money around, and it is landing on fewer companies.

If you run a small or mid-sized fund, you feel this in a very practical way. When a fund makes fewer investments, each one carries more of the fund's result, and everyone around the table knows it.

The hours stay the same, the questions multiply

A well-known survey of 885 venture capitalists, run by researchers at Harvard, Stanford, Chicago and UBC, puts numbers on how this work gets done. The average VC firm in that survey screened about 200 companies a year, invested in four, and spent 118 hours on due diligence for each deal that closed.

Making fewer deals does little to reduce those hours. Each remaining deal tends to be discussed for longer, and the partner championing it gets harder questions, first from colleagues and later from the fund's own investors. Is the market real? Who checked the revenue figure? Which competitors did the founder leave out, and why this company rather than the similar ones you passed on? All of these are really questions about how you reached your view, and you can only answer them well if the work behind the memo was done in a way you can show.

Where the memo comes from now

Something else changed over the same period. In a survey of more than 140 private market fund managers by Neuberger, presented in Singapore this September, about 90% said they were increasing their spending on AI in due diligence, and about 78% in the investment decision itself. Drafting a memo with a general assistant has become normal.

We understand why, because it saves a lot of writing time. What it cannot do is help with the questions above. A general model reads the deck it is given and summarizes it fluently, and the deck is the founder's best version of events, so you end up with the same blind spots in better prose. When someone on the committee asks where a number came from, "the deck" is a weak answer.

Three tests for every pick

We have spent a lot of time on this question, and it comes down to a few plain things.

The first is that every claim is traced to a source. If a deck says the company has 50,000 active users, the evaluation should show what the company's own website, its app listing or the press say, and if no source backs the claim up, it should say that too. In a recent evaluation of a pre-seed diagnostics startup, the most useful part of the report was the list of what no public source could confirm: revenue and burn, the regulatory route, and the team's track record. None of that rules the company out. It gave the investor the agenda for the next call with the founder.

The second is that the scorecard fits the company. A biotech in clinical validation and a Series A software company make progress in completely different ways, and if you score both on retention and payback, one of them will be misread.

The last is consistency. Two analysts will weigh the same company differently, and so will the same analyst on a different day. A committee comparing deals across a year needs a baseline that does not move with the reader.

How Kuanta approaches it

At Kuanta, specialist agents each take one part of an evaluation. They compare the founder's claims with outside sources, attach a citation to every finding and flag whatever no source could confirm. The company is scored on the framework for its industry, one of 24 with 835 criteria in total, and the report always has the same structure, ending with questions to put to the founder.

The judgment stays with the analyst. What changes is that the starting point is something you can put in front of a committee and walk through line by line.

Questions people ask

Why are there fewer venture deals in 2026? Investors are putting more money into a smaller number of large rounds, many of them in AI. In Europe, the first half of 2026 had the fewest deals in six years, while total funding rose.

What should a good startup evaluation check? Three things: that each claim is traced to a source, that the scoring framework matches the company's industry and stage, and that the process gives the same result whoever runs it.

Sources

KPMG Private Enterprise, Venture Pulse Q2 2026 · OpenAI, OpenAI raises $122 billion to accelerate the next phase of AI · Tech.eu, H1 2026 European Tech Ecosystem Report · Tech.eu, Fewer deals, bigger bets: Europe's venture market resets in Q1 2026 · Gompers, Gornall, Kaplan and Strebulaev, How Do Venture Capitalists Make Decisions? (survey of 885 VCs) · Neuberger Private Markets, survey of 140+ general partners, presented at the Asia PE-VC Summit, September 2026