The Missing Middle: Why $1M to $2.5M Pre-Seed Rounds Are Vanishing

Carta's State of Pre-Seed report for the first quarter of 2026 counts roughly 3,000 US startups closing a pre-seed round, raising more than $2.3 billion, with the final figure expected near $2.9 billion once late filings land. Total dollars look stable. The distribution does not.

Here is the number worth staring at. Rounds between $1 million and $2.5 million made up 24 percent of all pre-seed rounds in the first quarter of 2023. In the first quarter of 2026 they made up 18 percent. Rounds under $1 million grew their share. A small number of very large first rounds grew too. The middle thinned.

The median pre-seed round in the quarter landed near $1 million on a post-money SAFE cap in the $4 million to $6 million range. If you have been anchoring on a $2 million round at a $12 million cap because a friend raised that in 2022, the market you are about to pitch does not look like that.

A barbell means capital pools at both ends and drains from the center. At one end you have small rounds, often under $250,000, assembled from angels and small checks with no lead. At the other end you have $2.5 million and up, led by an institutional fund that sets terms and usually takes a board observer seat.

These are not two sizes of the same thing. They are two different processes with different timelines, different diligence, and different expectations about what happens next.

The common failure is running the wrong process for your size. Founders who want $1.5 million often run a lead-led process, spend four months meeting institutional funds who write $3 million or nothing, and end up assembling a party round anyway after burning a third of their runway. Choosing your end of the barbell early is worth more than another week polishing the deck.

Three reasons, and none of them are about your company.

Fund math. As seed funds raised larger vehicles, the check size that moves their ownership needle went up. A fund with $150 million under management cannot deploy it in $1.5 million checks without hiring more partners than the fee stream supports. So they moved up.

AI concentration. AI startups took roughly 30 percent of pre-seed dollars a few years ago. By the first quarter of 2026 they were near 50 percent. That concentration pulls averages up and leaves less patient capital for everything else.

Angel infrastructure. Rolling funds, syndicates, and SPVs made it easy to assemble many small checks quickly. That grew the bottom end. A founder who needs $600,000 can now put it together in three weeks with no lead, which was much harder in 2021.

Answer three questions and the choice usually makes itself.

How many months does the round need to buy, and what proof point sits at the end of them? If $700,000 gets you to a signed pilot with revenue, the small end is fine and you keep more ownership.

Do you need a lead's brand to hire? Some markets, especially deeptech and regulated categories, need an institutional name on the cap table to recruit the first senior person. If yes, the longer lead-led process is worth it.

Can you tell a story that justifies $2.5 million with no product? TechCrunch ran a Disrupt session this July on exactly that: raising pre-seed on conviction and storytelling rather than traction. It is possible. It is also the harder path for most first-time founders, and the fallback costs you months.

At the small end, investors are buying you and a wedge. They want evidence you understand a specific customer's problem in more detail than they do, and that you can ship. A working prototype plus three customer conversations with real quotes usually clears the bar.

At the lead-led end, the bar moved. AI founders often raise on a working prototype plus signed letters of intent. Non-AI founders are more often asked for a paying pilot or something comparably concrete. That gap is real, and it is worth planning around rather than arguing with.

Getting the story tight matters more than deck design. Foundra is one of the tools first-time founders use to turn a rough idea into a defined problem, a named customer, and a sequence of proof points, which is the raw material any pre-seed conversation runs on.

The instrument matters less than founders expect. Post-money SAFEs remain the default at both ends.

Median caps have been clustering near $10 million for rounds in the $250,000 to $1 million range and near $15 million for rounds between $1 million and $2.5 million. Those are medians, not entitlements.

Two errors show up constantly. The first is pricing to a peer's headline. AI companies price roughly 42 percent above non-AI peers right now. If you are not in that category, benchmarking against an AI friend's cap sets an expectation you cannot fill, and a stalled round costs more than a lower cap.

The second is optimizing the cap at the expense of closing. Dilution at pre-seed is a rounding error next to the cost of running out of money before your next proof point. Take the closable round.

Watch stacked SAFEs. Three high-cap or uncapped notes can create an unpleasant conversion surprise at your priced round.

Assume ten to fourteen weeks from first meeting to money in the bank at the small end, and sixteen to twenty at the lead-led end. Build runway math backward from that, and start when you have six months left rather than three.

Build a list of forty to sixty investors sized to your end of the barbell. Sort by whether they have written a check your size in the last twelve months. Ignore fund brand at this stage, and ignore anyone whose most recent pre-seed check was in 2023.

Batch your meetings so momentum is visible. Twenty meetings across three weeks creates urgency. The same twenty across three months creates the impression of a round that is not happening.

Send a short written update every two weeks to everyone who passed. A meaningful share of pre-seed rounds close on a second look.

Quarterly venture data lags. Pre-seed rounds are often filed late or never filed at all, so the current quarter always looks weaker than it eventually reads. Use these numbers for shape, not precision.

Be careful about treating the barbell as a verdict on your company. The middle shrinking is a fund structure story more than a quality story. Plenty of good businesses raise $800,000 and never need another round.

And the roughly ninety new unicorns minted so far this year should not be your reference class. Those outcomes are visible because they are rare. The median pre-seed company is raising about a million dollars and trying to reach a paying customer, which is a perfectly reasonable place to be standing in August.

Mid-sized pre-seed rounds are shrinking as a share of the market, from 24 percent in early 2023 to 18 percent in early 2026.

The market is a barbell: sub-$250,000 angel rounds at one end, $2.5 million and up lead-led rounds at the other.

Pick your end before you start. Running a lead-led process for a $1.5 million round wastes months.

Median pre-seed is near $1 million on a $4 million to $6 million post-money SAFE cap.

AI companies price roughly 42 percent above non-AI peers. Do not benchmark against them if you are not one.

Closing beats cap optimization. Dilution is cheaper than running out of runway.

Start the raise with six months of cash left, batch meetings into three weeks, and keep updating the investors who passed.

Is a party round with no lead a bad signal? Not by itself in 2026. Syndicate infrastructure normalized it. What reads badly is a round that stays open for six months and keeps changing size.

Should I raise more than I need because the market is tight? Raise what buys a clear proof point plus a buffer of about three months. Raising well beyond that sets a valuation you then have to grow into.

What if I get a term sheet below the median cap? Compare it to your alternative, not to the median. A closed round at a $6 million cap beats an open process at $12 million.

Do I need revenue to raise pre-seed? Not always, but the substitute has to be concrete. A paying pilot, signed letters of intent, or a prototype with real usage all work. A pitch deck alone rarely does.

How many investors should I expect to pass? At pre-seed, a 5 to 10 percent conversion rate from first meeting to check is normal. That is why list size matters more than any single meeting.

1. State of Pre-Seed: Q1 2026 (Carta)
2. State of Private Markets: Q1 2026 (Carta)
3. Pre-Seed Funding Report: H1 2026 Data, Caps, Investors (Causo Hub)
4. No product? No problem. Getting pre-seed funding with conviction and storytelling (TechCrunch)
5. Almost 90 new unicorns have been minted so far this year (TechCrunch)