Last Updated on July 7, 2026 by Taya Ziv
Almost every founder I meet writes the same number on slide one. We’re raising $1.5 million. It is such a natural number that nobody questions it. It feels grown up without being greedy. It is enough to hire two or three people, run a real pilot, and buy yourself maybe eighteen months of not panicking. It is the round that lets you stop selling for a while and actually build the thing. For about a decade, that round was the default starting move of a startup. The training wheels everyone got to use.
Here is the part that should make you sit up. According to the cleanest pre-seed data we have, that exact round just became the single hardest one to close. Not the tiny scrappy round. Not the giant headline round. The sensible one in the middle, the one you probably wrote down without thinking, is the one quietly slipping away.
What the data actually says
Carta runs the numbers on this every quarter, and their reports are about as honest as it gets, because they’re reading off real cap tables, not vibes or LinkedIn brags. Cap tables don’t exaggerate. Their State of Pre-Seed report for the first quarter of 2026 covers roughly 3,000 US startups that raised at the earliest stage, adding up to about 2.3 billion dollars, on track to land near 2.9 billion once the late data files in. So the first thing to know is that the money itself didn’t vanish. Total pre-seed cash is basically flat, sitting where it’s sat for a while. If you only read the top line, you’d think nothing changed.
But look at the shape instead of the size and the floor moves under you. Rounds between 1 million and 2.5 million dollars made up 24% of all pre-seed rounds back in the first quarter of 2023. By the first quarter of this year, they’re down to 18%. Meanwhile the rounds under a million got more common, and the big rounds above 2.5 million held steady. The same pile of money, spreading toward the two ends and draining out of the middle.
That’s the whole story in one sentence. The pre-seed market is turning into a barbell. Heavy on the small end, heavy on the big end, and getting thin exactly where most founders are standing.
The middle was always the trap
Now here’s where I’m supposed to tell you this is a disaster. I’m not going to, because I don’t fully believe it. I’ve watched more than a hundred startups go through this stretch, and if I’m being honest with myself, the comfortable middle round was never as friendly as it looked.
Think about what 1.5 million dollars actually does to a young company. It’s exactly enough money to let you stop asking the only question that matters. Does anyone actually want this. With a tiny round, you can’t afford to be wrong for long, so you stay close to customers because you have no choice. With a giant round, you’ve usually already proven something real, or you’re in a category where being early and loud genuinely wins. The middle round is the one that funds a year of looking busy. You hire a couple of people, you build the roadmap, you get a nice office plant, and you spend eighteen months politely avoiding the market because you can. Then the money runs low, you finally go sell the thing, and you discover the answer was no the whole time. You just paid a million and a half dollars to delay hearing it.
I’ve said for years that technology is a distraction until you have proven demand, and that selling before you build beats building before you sell. The middle round is the financial version of breaking that rule. It buys you permission to skip the scary part. So when I see the data showing that round getting harder to raise, a part of me doesn’t read it as the market getting meaner. I read it as the market quietly removing the most seductive way for a good founder to waste a year.
Which end of the barbell are you on
That’s the real question this data forces, and it’s worth answering on purpose instead of by accident.
The small end, under a million, is not a consolation prize. It is a completely valid way to start, and in some ways it’s the most honest one. A smaller round keeps you lean by force. It keeps you in the conversation with customers because you can’t afford a moat of employees between you and the people paying you. It treats validation as the job, not as a phase you’ll get to after the hires. If you can prove that real people will pay, on a small check, you walk into your next raise holding real cards instead of a story. The catch is that this end is unforgiving for anything capital heavy. If you’re building hard deep tech, or anything that needs real money before it can show a single result, the sub-million world is genuinely brutal, and I won’t pretend otherwise.
The big end, above 2.5 million, is real too, but it’s earned differently now than it was two years ago. And this is where the second half of the Carta data matters. AI startups now take about half of every pre-seed dollar, up from roughly 30% a few years back. Caps on the standard SAFE are drifting upward even as round sizes stay flat, which means investors are paying more for the same slice, mostly on the bet that AI makes small teams absurdly productive. So the heavy end of the barbell is increasingly an AI end, and the concentration gets more extreme the higher you climb, the same force behind a single quarter where four companies soaked up most of the venture capital raised on the entire planet. If you’ve got a credible AI core, that lane is open and the caps are friendlier than they’ve been. If you don’t, you’re not locked out, but you are fishing in the half of the pond that’s getting smaller, and you should plan like it.
What you don’t want to do is aim straight at the vanishing middle out of habit. Targeting 1.5 million because it’s the number everyone writes is now the most crowded, most competitive corner of the entire market, and you’d be walking into it because of muscle memory, not because of a decision.
The boring detail that’s actually costing founders money
One more thing buried in the same report, and it’s the kind of unglamorous detail that separates founders who keep their company from founders who give it away by accident. SAFEs are now 93% of pre-seed rounds. Convertible notes fell to a record low, 7% of rounds and 8% of dollars. The note is basically extinct at this stage, and good riddance, because notes carry maturity dates and interest that turn into pressure exactly when you can least handle it.
But “everyone uses a SAFE now” hides a trap. People stack them. A 500k SAFE at one cap, then a 750k at a higher cap, then another at a higher one still, and each tranche converts at its own price when the priced round finally lands. The math isn’t hard. The problem is that almost nobody runs it. I’ve watched founders walk into a Series A conversation genuinely unable to say what percentage of their own company the SAFE holders will own after conversion. They were running on a feeling about “the cap” instead of the actual waterfall. That’s the moment an investor smells blood. If you’re raising on stacked SAFEs, and in this barbell market a lot of you will string together several small ones, model the conversion before you sign the next one, not after. Know your real dilution cold. It’s free to do and expensive to skip.
My take, and where I might be wrong
What I actually believe is that the disappearing middle is good news wearing a scary mask. It pushes founders toward the two honest positions, prove it cheap or have a real reason to go big, and away from the cozy middle where companies quietly coast into the graveyard with a full team and an empty pipeline. The barbell is the market telling you to make a decision instead of defaulting to a number.
Now let me argue against myself, because I think there’s a fair chance I’m romanticizing scarcity from a comfortable seat. It’s easy to call the small round “honest” when you’re not the one trying to build something that physically can’t ship on 800k. For a chunk of genuinely good, capital heavy founders, the hollow middle isn’t a clarifying nudge, it’s a closed door, and the people who sail through the big end aren’t always the ones with the best company, just the ones with the right two letters in their pitch. AI in the deck shouldn’t be a cheat code, and the fact that it sort of is right now is its own quiet problem. So I hold my take a little loosely. The middle dying is probably healthy for most pre-seed founders, and unfair to a real minority of them, and both of those things are true at once.
What to actually do about it
Three things, and none of them is “wait for the middle round to come back,” because the trend has been pointed this way for several quarters and it isn’t a blip.
First, pick your end on purpose. Before you write a number on slide one, decide whether you’re running the lean play or the big-bet play, and size the raise to that decision instead of to habit. If you can’t articulate why you need more than a million, you probably should be raising less than a million and staying close to your customers.
Second, if you go lean, treat validation as the entire point of the money, not as something you’ll get to after the hires. Spend the small round proving people will pay, which is really just the discipline behind the 90-day revenue rule that’s quietly replacing MVP culture, because that proof is the thing that buys you a real next round on real terms.
Third, know your cap table math before you need it. Model your SAFE stack, know your dilution at a few different priced-round scenarios, and benchmark your cap against this year’s data, not the comps you remember from 2023. Caps moved. Don’t leave money on the table by anchoring on an old number, the same way you shouldn’t chase a round size just because it used to be standard.
The headline says pre-seed funding is stable, and technically that’s true. It’s the same trick as when seed funding hit record highs while five out of six of those startups still won’t reach a Series A, a cheerful number sitting on top of a much harder story. Stable is hiding a real shift, and the founders who keep writing 1.5 million on slide one out of habit are aiming at the one spot on the board that’s emptying out. Look at the barbell, pick your end, and raise like you actually decided.
