Startup Funding Stages Explained: From Pre-Seed to IPO (2025–2026 Data)

Dealroom’s European definition is $100K–$1M. That band currently holds 21,938 pre-seed companies, versus 16,818 at seed, a 22% pre-seed-to-seed conversion on Dealroom’s cohort tracking. Most ideas never raise a priced seed. That is normal, not a verdict on the founder.

PitchBook’s Q1 2025 European Venture Report put the median pre-seed deal at €0.7M, up from €0.6M at the end of 2024. The cheque grew; the count did not. Funds that once wrote €150K notes now write €400K–€800K and take fewer shots. Founders feel that as a higher bar for a first meeting: a deck, a product URL, and a reason this team can ship.

Who writes the cheque? Founders themselves, operators-turned-angels, scout programmes, and a thin layer of sub-€100M funds. Atomico warned in 2025 that Europe’s median VC fund has tripled since 2016 (from $32M to $105M) while the sub-€100M layer is thinning. That is why pre-seed feels both richer and harder: fewer specialised funds, larger tickets when they do engage.

Instruments are messy on purpose. SAFEs, ABSA/BSA AIR in France, and convertible notes delay valuation until seed. That is useful if you are pre-revenue. It is costly if you stack three SAFEs at different caps and walk into a seed with a messy cap table. Carta’s US data is the best public window on that damage: the priced seed looks like 19% dilution until the SAFE stack converts.

What investors underwrite at pre-seed is the team and the problem, not the model. Bring a one-page dataroom: incorporation, cap table, pitch, and any customer conversations. Onefive’s dataroom is built for exactly that share: granular access, not a public Google Drive link.

Dealroom places seed at $1–4M and counts 16,818 European seed companies. About 24% of that cohort later appear at Series A. The other 76% stall, sell, or raise quiet extensions that never show up as a new series. If you are raising seed, you are already in a minority of started companies, and still far from a default Series A.

PitchBook recorded a 34.6% jump in the European seed median to €2.2M in Q1 2025, with valuations up 15.4% to €5.6M. Atomico, using a slightly different cut as of August 2025, said European seed and Series A medians climbed 23% and 25% year on year, and that the European seed median is still only half the US figure. Both datasets agree on the direction: fewer rounds, larger cheques, a wider gap between a ‘standard’ seed and an AI outlier.

Crunchbase’s North America 2025 review is the US mirror. Seed-stage investors put about $20.4B into reported US and Canadian rounds, 9% below 2024, while deal counts drifted down to just over 1,300 financings in Q4. Globally, Crunchbase logged $9.9B of seed in Q4 2025 and $12B in Q2 2026, with a fat tail of $10M–$100M ‘seed’ labels that would be Series A anywhere else. Always ask which definition the person across the table is using.

Dilution is the number founders remember. Carta’s State of Private Markets Q1 2025 put median US seed dilution at 18.8%, down from 21.4% a year earlier, on a $16M median pre-money. Carta’s later software-only cut sits near 19.5% at seed, 18% at A, 14% at B, and 10% at C. Under 10% of 2025 US software seeds sold 30% or more. If a European term sheet asks for 30% on a €1.5M seed, you are not looking at the market median. You are looking at a small fund stretching a large ownership target.

What a seed dataroom must contain: a 12–18 month model, cohort or pipeline evidence, a clean cap table, and a use-of-proceeds slide that maps to hiring. DocSend-style link sharing is common; a permissioned dataroom is safer once more than three funds are in diligence. That is the job Onefive is designed for, alongside a verified profile investors can scan in the few minutes DocSend measured as the average visit to a seed deck.

Dealroom lists 10,881 European Series A companies. The seed-to-A conversion is 24%. Atomico described this jump as one of the hardest in Europe: you stop proving the product and start proving you can sell it without the founding team on every call. Around 15% of companies that raised a Series A in the first half of 2024 had raised their next round within 12 months, roughly double the prior cohort, but still a minority.

PitchBook’s early-stage European numbers are easy to misread because ‘early-stage VC’ mixes instruments. The cleaner Series A story in 2025 is qualitative: more capital per winner, fewer winners. Carta saw US Series A count fall 10% year on year in Q1 2025 even as the median pre-money rose 9% to $48M. KPMG’s Venture Pulse Q4 2025, built on PitchBook, put global VC above $500B for the year while deal counts fell, the same ‘fewer, larger’ pattern at every stage.

What Series A investors underwrite: net revenue retention or a credible path to it, a CAC payback you can defend, a hiring plan that does not assume a frictionless market, and a cap table that still has room for a B. AI companies cleared that bar faster in 2025. Atomico found AI names carried about a 20% valuation premium at seed and Series A in Europe, widening later.

Process length stretches. A European A often runs longer than a seed because more partners vote and more data rooms get opened. Founders who arrive with versioned financials, a security review, and a single investor CRM close faster. Onefive’s workflow (profile, dataroom, matched intros, view analytics) is the Series A version of that hygiene, not only the seed version.

Only 4,384 European companies sit in Dealroom’s Series B band, a 30% conversion from Series A. Atomico reported that about one in four companies that raised a Series B in H1 2024 secured their next round within a year. The B is where Europe’s growth-capital gap becomes visible: the product works, the market is there, and the local fund that led the A cannot write the whole B.

Valuations moved faster than cheques. Atomico said median European Series B and C valuations rose more than 40% from the end of 2024, while US Series C valuations jumped as much as 69%. AI names in Europe carried a 50% higher median post-money at Series B than non-AI peers. A ‘standard’ B and an AI B are no longer the same asset class.

Governance tightens. Information rights, board composition, and secondary selling for early angels show up in the term sheet. Venture debt also becomes rational. Atomico estimated European venture debt at a record $5.6B in 2025, after debt reached 12% of total capital invested in 2024 (versus 20–25% in the US). Debt does not replace a B; it extends the B you already earned.

Dealroom’s $40–100M Series C band holds 2,093 European companies, a 33% conversion from Series B. The absolute numbers are small. The concentration is not. Atomico showed that in France and Sweden the top five 2025 deals accounted for more than half of national funding; in Finland the share reached 80%. A European C is often one of a handful of cheques that move a country’s year.

US exceptionalism is the backdrop. Atomico reported $177B of private US tech investment in the first nine months of 2025, almost double the same period in 2024, with the US taking two-thirds of global private tech capital. Crunchbase’s H1 2026 global total of $510B already exceeded all of 2025. European C founders are not competing with the European median. They are competing with US growth funds that can write $100M+ primary.

That is why so many European C rounds now include a US or sovereign co-lead, and why Atomico’s founder survey keeps flagging late-stage capital, not idea quality, as the constraint. If you are heading for a C, your dataroom starts to look like an IPO binder: audited or audit-ready financials, a hiring and geographic plan, and a narrative that still works if AI multiples compress.

Atomico said European growth-stage capital was on track in 2025 for its strongest year since 2022, after a brutal reset from the 2021 triple. Early-stage funding, by contrast, had been roughly stable since 2018. The recovery is real and uneven. The UK led 2025 country totals at $14B (+22% year on year). Germany, Sweden, and Finland also rose. Seventy-six percent of countries outside the top ten saw funding fall.

Crunchbase’s North America 2025 number ($280B into seed-through-growth, +46% year on year) shows how wide the Atlantic still is. Close to $69B of that went to Series A and B companies. Growth in the US is an asset class with pensions and sovereigns. Growth in Europe is still a thin market that a few megarounds can swing.

For founders, a growth round is a choice to stay private. The diligence looks like a public-market process without the prospectus: cohort quality, gross margin trajectory, litigation, and a path to an IPO window. Onefive is built for the earlier stages, but the same discipline (one profile, one permissioned dataroom, one audit trail) is what growth investors expect when they arrive.

PitchBook’s European venture reports through mid-2025 described an exit mix still dominated by M&A, on the order of three-quarters of exits in early 2025, with only a handful of VC-backed IPOs. Atomico’s LP and GP survey named the lack of M&A routes and the shallow European public markets as the two biggest barriers to deploying more capital. The bottleneck is not a shortage of seed companies.

The US reopened its IPO window faster. That liquidity loop is why US funds can raise larger vehicles and write larger growth cheques. Europe ended 2025 with 413 unicorns (Atomico, from 127 at the end of 2016) and 28 new billion-dollar companies already minted by late October 2025. Dealroom’s live Europe page lists 673 companies in the $1–10B unicorn band and 50 decacorns. The companies exist. The listings have not kept pace.

Founders planning an IPO start the work years early: IFRS or US GAAP, a CFO who has sat on an earnings call, and a cap table that can survive lock-ups. Secondary programmes and tender offers are the private-market substitute while the window stays narrow. None of that replaces the earlier craft of a clean seed and Series A. The companies that list well are usually the ones that ran a disciplined dataroom from the first priced round.