Seed Round Benchmarks 2026: Size, Valuation, Dilution, Time

A seed round in 2026 is bigger, pricier and cheaper in equity than it was two years ago, but only at the median, and the median is now a software company that is probably AI-native. This page puts every seed benchmark that matters in one place: how much founders raise, at what valuation, how much of the company it costs, which instrument they sign, how long it takes, and what moved since 2024. Every figure comes from a report we opened while writing, labelled with its period.

This year's medians on Carta: a priced seed raises $4.1M at a $24.3M post-money and costs founders 18% of the company. Pre-seed is a stack of capped post-money SAFEs, the gap from seed to Series A runs close to two years, and in our work a well-run seed closes inside six months at a dilution we consider healthy. The tables below carry every figure with its source and period.

Two things to know before reading. First, the deepest data is US-only: Carta reports on companies that run their cap table on its platform, and PitchBook-NVCA on US venture deals, so Europe gets its own rows wherever a source exists. Second, the medians describe the middle of a market Carta itself calls bifurcated, with AI companies taking most of the money. Treat each number as the centre of a wide range, and our funding stages guide as the map of where seed sits in the sequence.

How big is a seed round in 2026?

Median seed rounds have stopped growing while the top of the market runs away. PitchBook's median US seed deal has been flat at $3.0M for three years, and Carta's software-only median sits a little higher. Pre-seed is smaller and fragmenting: most rounds stay under $1M and the middle band is shrinking. In our work, seed engagements typically run in the low single-digit millions.

The headline is the flat line. On PitchBook's count the typical seed cheque hasn't moved in three years, while the median Series A cheque has grown by about three-quarters since 2024 (PitchBook-NVCA). Carta's software-only sample reads higher because it leaves out the long tail of hardware, biotech and consumer rounds. Both agree on the shape: a crowded middle raising what it raised before, and a thin top tier raising sums that used to be called Series B.

Pre-seed is a different animal. It isn't one round but a stack of instruments signed over months, and Carta's data shows the stack getting more lopsided: cheques are larger, fewer companies get them, and the middle band is thinning out (Carta, State of Pre-Seed). If you're raising your first money, our pre-seed guide covers who writes those cheques; this page covers what the market pays for them.

Europe has no Carta-grade seed dataset, so we use what PitchBook and Dealroom publish. Dealroom's bands put a European seed at $1–4M, and PitchBook's half-year report notes early-stage deal sizes growing even as pre-seed valuations soften, with two seed rounds above $1B, both AI labs, pulling the average a long way from the median (PitchBook).

What is a seed valuation in 2026?

The median priced seed on Carta is a $24.3M post-money in the six months to July, a record, and PitchBook's US median pre-money has more than doubled since 2021. Pre-seed SAFE caps sit in the low teens of millions for typical round sizes. The spread matters more than the median: top-decile seed valuations run nearly 4x the middle, and AI and geography decide which side of it you're on.

Read the valuation rows together with the round-size rows. Carta's own explanation of the record is arithmetic: dilution stayed inside its historical band while the cheques for the companies investors wanted got larger, so the price followed (Carta). Valuation is the output, not the input. A bigger median valuation isn't free money; it arrived with a bigger round and the expectations that come attached.

Then read the spread. A software seed in the Bay Area is priced at more than two and a half times one in the rest of the country, New York sits in between, and outside those two hubs the median SaaS seed valuation hasn't moved in a year (Carta). An AI label added roughly half again to the pre-money at the start of last year. If your comps are the Carta median and you're a non-AI company in a secondary market, you're anchoring to a room you're not in. Our valuation methods guide covers how to build a range you can defend; in our work we see a 2–3x gap between what founders expect and what the market supports, and this split is where most of it comes from.

How much dilution does a seed round cost in 2026?

The median seed round on Carta sells 18% of the company this year, the lowest in several years, and founders typically keep a little over half once the round closes. Our healthy band is 15–20% for pre-seed and seed, measured after the option pool and the SAFE conversions land, because the headline round is rarely the number that hits your cap table.

The number investors quote and the number you feel are different. The 18% is what the new money buys. In the same closing, the lead usually requires an option pool inside the pre-money and every pre-seed SAFE converts at its cap, so the founders' line on the cap table drops by a lot more than the headline. We walked the full arithmetic, round by round on the same Carta medians, in startup dilution per round, and the dilution calculator runs it on your numbers; this page won't repeat it. The short version: quote your ownership after the pool and the conversions, or the associate will do it for you.

Two consequences for the raise. First, the size of the round sets the dilution more than the valuation does, because valuation follows the other two. Decide what the next 18–24 months need, then look up what that raise implies at a healthy dilution; the table in the last section does the conversion. Second, under-dilution is a signal too: a seed that sells well under our band usually means the round was too small to reach the Series A milestone, or the price was set by one enthusiastic angel rather than a lead. The same logic repeats at a higher price in our Series A benchmarks.

SAFE, priced round or convertible note: which instrument do seed rounds use in 2026?

Almost every pre-seed round is a SAFE, almost every SAFE is post-money, and almost every post-money SAFE carries a cap; Carta puts the first two above 90% in Q2 2026, with convertible notes at a record low. Priced equity takes over when there's a lead and the round is big enough to justify the legal work, which last year meant most rounds above $4M.

The instrument choice is mostly a round-size choice. Below a few million dollars with no lead, a post-money SAFE is faster, cheaper and standard, and investors will expect it; above that, a lead wants a priced round with preferred terms and a board seat. The pre-money SAFE is effectively gone, which matters: a post-money SAFE locks the investor's percentage at signing, so every later SAFE dilutes you and not them (Carta). Our SAFE note guide and convertible note explainer cover the mechanics; the practical rule is to model the whole stack after every signature, because a large pre-seed is now routinely ten or more instruments deep.

One more thing the mix hides. A capped SAFE prices your company whether or not you call it a valuation, because the cap is the number the next lead anchors to. In our work, pre-seed founders who took a generous cap from a friendly angel often find the seed lead treats it as the ceiling, not the floor. Caps are rising fast at the top of the market, and a cap you can't grow into is a flat round waiting to happen.

How long does a seed round take in 2026, and how long until Series A?

Budget for two clocks. The raise itself: in our work a well-run seed goes from first meeting to signed term sheet in 4–6 months, and founders running the process solo take roughly twice as long. The gap to the next round: Carta's median wait from seed to Series A was close to two years at the end of 2025 and is now shrinking back toward the traditional window.

The two clocks interact. The seed has to fund the company to Series A metrics, and the gap between the rounds stretched for three years before it started to shrink this year (Carta), which is why we budget a seed for 18–24 months of runway rather than the shorter horizon a first model tends to assume. If the round closes late, the runway is already shorter than planned, and a bridge at a flat cap is the usual outcome. Our bridge round guide covers what that costs.

Where the time goes is mostly before the first meeting. In our work the raises that close inside the window had the deck, the model and the data room tied out before outreach, and a target list of the right size; the ones that drag are pitching Series A funds for a seed-sized cheque, or working a cold list with no warm paths. The fundraising plan lays out the 30-day prep sprint, and our investor database comparison covers where the target list comes from.

What changed between 2024–2025 and 2026, and why?

Prices moved; cheques didn't. The median seed valuation on Carta rose by half in two years while PitchBook's median seed deal stayed flat, dilution fell a couple of points, and priced rounds on Carta hit a six-year low. The reason is concentration: AI companies took about half of pre-seed dollars and most venture capital overall, and the top tenth of startups raised roughly half of all the money.

The valuation move is the part founders notice; the concentration move is the part that decides their round. Carta's data desk reads the price rise as arithmetic: dilution stayed near its historical band while the cheques for the companies investors wanted got larger, so the price followed (Carta). That happened for a narrow set of companies. For the rest, the cheque is the same as it was in 2024 and the bar is higher, which is why Carta's round count fell to a six-year low even as dollar totals set records, and why PitchBook sees a record number of first financings alongside a collapsing share of dollars for anything under $100M.

Three forces did it. AI absorbed most of the new money at every stage and roughly half of pre-seed dollars, in both 2025 and the first half of 2026 (Carta, State of Pre-Seed). Multi-stage funds moved down into seed: PitchBook counted Andreessen Horowitz and Y Combinator at 46 investments each in a single quarter, with General Catalyst and Sequoia close behind (PitchBook-NVCA). And AI tooling cut the cost of building, so the companies that do raise arrive with more product and revenue than the same round used to buy, which is one reason the seed-to-A clock started shortening. None of this is a bubble call. It's a description of who the median describes: a software company, probably AI-native, quite possibly in San Francisco or New York.

What these seed benchmarks mean for your raise

Use them to size the round, not to price it. Work out what the next 18–24 months need, then divide that raise by a healthy dilution for the valuation the market supports; for most non-AI companies outside the coastal hubs, the median is a ceiling. In our work across 600+ companies, the founders who close fastest arrive with that math, a tied-out model and a thesis-matched target list.

Here is the whole page in one paragraph. In 2026 the seed market reads differently depending on who is counting. Carta's benchmark of more than 1,000 priced rounds by US software companies, at least half of them AI-native, puts the median seed at $4.1M raised on a $24.3M post-money. PitchBook-NVCA's all-sector US median is a $3.0M seed deal, unchanged since 2024. Crunchbase counted $12B of global seed funding in the second quarter, more of it in small rounds than in the handful above $100M. In Europe, PitchBook's median pre-seed and seed pre-money is €6M. Pre-seed remains a SAFE market on Carta, overwhelmingly post-money and capped, with median caps in the low teens of millions. And the median wait from seed to Series A was 1.9 years at the end of 2025, trending back toward the traditional window.

Start with the raise. Our rule is the milestone that unlocks Series A plus a buffer, not the maximum an investor floats, because valuation follows raise and dilution, and an oversized round at an inflated price is the setup for a flat bridge. The table below turns a raise into the post-money our band implies; set it against the valuation table above and you can see where the median sits relative to a healthy seed.

Then the instrument. Under a few million with no lead, take the post-money SAFE the market expects, set a cap you can grow past inside a year, and keep a running pro forma of the stack. With a lead and a round in the upper half of the table, price it: the legal cost is real, but the cap-table clarity is worth more, and Series A investors price off a clean preferred round more readily than off ten SAFEs at four different caps.

Then the clock. Prep before outreach: the deck (seed decks in our work average 14 slides), a driver-based model, a data room indexed against a diligence checklist, and regular investor updates to the people already on the list. A $6M seed for an AI adtech client had stalled on a model that couldn't survive unit-economics scrutiny; we rebuilt it in a week and the round closed with a tier-1 lead a month later. The model was the gate, not the market.

The honest caveat: these benchmarks describe the middle of a market that is mostly deciding what to pay AI companies in two cities. Your round will be priced on your evidence, your sector and your investors' alternatives, and the benchmarks are the argument you bring when someone quotes a comp you don't resemble. If you'd like a second pair of eyes on the raise, our seed funding team runs the process on a retainer, scoped after a 30-minute diagnostic call. Founders raising solo typically take 9–12 months; the raises we run closed 70% faster on median.