Seed Capital Is Up 37%. Seed Rounds Are Down 20%.

Two numbers from Carta's Q2 2026 data tell most of the story. Capital raised at seed is up 37%. The number of seed rounds is down 20%.

More money. Fewer winners.

The pattern repeats up the stack. Series A capital up 16% with rounds down 12%. Series E and later, capital up 142%. Series B is the odd one out, where capital fell 30% even as deal count ticked up slightly.

So the pool got bigger and the door got narrower at the same time.

If you are raising your first round this fall, you are competing for fewer slots than the founder who raised the same round eighteen months ago. The check sitting behind each slot is bigger. That is a real improvement for the people who get one and a real problem for everyone else.

Here is the part that trips up first-time founders. The headlines say funding is back, and funding is back. Access is not. Those are different claims and only one of them shows up in your bank account.

Because you cannot raise an average.

Average round size is a summary statistic. It describes a distribution you are either inside or outside of. Round count is closer to a headcount: it tells you how many companies actually got funded. When that number drops 20% year over year, roughly one in five founders who would have closed a seed round in the old market does not close one now.

That founder did not get a smaller check. They got nothing.

PitchBook has been tracking the same squeeze from a different angle. Deals under $5 million used to be more than 70% of all US venture deals a decade ago. They are now less than half. Seed funds and multistage firms are competing for the same companies, which pushes deal sizes up and crowds out the small, scrappy, get-it-done round that most first-time founders were counting on.

The $750K friends-and-angels round that gets you to a real product has not disappeared. It has just stopped being the thing institutional money wants to write.

And the concentration goes further than seed. Megadeals of $100 million or more took 87.5% of the $412.7 billion deployed across US venture in the first half of 2026. Read that again. Records at the top, contraction almost everywhere underneath.

Three things, and they compound.

First, your raise takes longer. Fewer slots means more conversations per slot. Budget for a process that runs four to six months from first meeting to wired money, not the six weeks you keep reading about on X.

Second, the bar moves before you get there. Investors writing larger checks into fewer companies want more proof per check. Paying customers, not signups. Retention curves, not a waitlist. A burn number you can defend line by line. This is not gatekeeping for its own sake; it is what happens mathematically when a fund concentrates the same dollars into half as many names.

Third, and this is the one nobody warns you about, your fallback options thin out at the same time. Bridge rounds got easier to talk about and harder to get on decent terms. If your plan assumes you can patch a gap with a quick insider extension, check whether your existing investors have the reserves and the appetite. Many seed funds are now holding reserves for their two or three obvious winners and quietly triaging the rest.

So the failure mode is not raising a smaller round. It is running a process on the old timeline, missing, and discovering that the safety net was also repriced.

Stop asking how much you can raise. Ask how much time you need to buy.

A plan that works in this market has four components, and you should be able to write each one on a single line:

1. The one metric that makes your next round possible. Not five metrics. One, with a number attached.
2. How long it takes to hit that number, using your actual shipping pace rather than the pace you would have if everything went right.
3. Six months to run the raise itself.
4. Six more months of slack, because the first attempt at step two will be wrong.

Add those up and you usually land somewhere between 24 and 30 months of runway, which is well past the 18 months that every template still suggests. If the resulting number is a round you cannot raise, that is useful information. It means you need a cheaper path to the same milestone, not a more optimistic spreadsheet.

This is where a lot of first-time founders stall, because building the model means making explicit guesses about pricing, conversion, and hiring that feel uncomfortable to write down. You can do it in a spreadsheet, in Notion, or in a planning tool like Foundra that walks you through the sections one at a time. The tool matters less than the discipline of putting a number next to every assumption so you can watch which ones break first.

One more thing. Model the version where you raise nothing. If that version still gets you somewhere real in twelve months, you will negotiate better in every conversation you have.

Roughly the same things they always wanted, weighted differently.

Proof of demand now outranks proof of vision. A few paying customers with a retention curve beats a large market sizing slide. Show the cohort, not the TAM.

Clean ownership matters more than it used to. Who owns the IP, who has equity, which contractors signed what. Diligence has gotten longer because the checks are bigger, and messy cap tables now kill deals that would have survived in 2021.

Burn discipline gets read as judgment. A founder who can explain their burn multiple and defend each hire reads as someone who will not need an emergency bridge in month nine.

And category clarity has become almost a filter. If a partner cannot repeat back what you do in one sentence after your first meeting, you are unlikely to survive the internal conversation you never see.

None of that is a trick. It is the natural consequence of an investor writing half as many checks and having to defend each one to a partnership.

Then you build a company that does not require one, which is a better outcome than it sounds.

The cost of starting has kept falling while the cost of capital has gone up. That is an unusual combination and it favors small teams with revenue. A company doing $30K a month with two people has options: stay small and profitable, raise later from a position of strength, or take an acquisition conversation seriously. A company doing $0 with eight people has exactly one option, and it is the one that just got 20% harder.

Practical moves that work when institutional money is scarce:

• Charge earlier than feels comfortable. Revenue is the only unconditional funding source.
• Look at non-dilutive money. Grants, accelerator programs, revenue-based financing, and customer prepayments all exist and none of them require a partner meeting.
• Shrink the milestone. If the metric you need is out of reach on current cash, find the smaller version of it that still proves the thesis.
• Extend the team runway before the company runway. Fewer, better-paid people beat more, cheaper ones almost every time.

The founders who come out of a concentrated market well are rarely the ones who out-pitched everybody. They are the ones who needed less.

• Carta's Q2 2026 data shows seed capital up 37% while seed round counts fell 20%. Bigger checks, fewer recipients.
• Round count matters more than average round size, because you cannot raise an average.
• Sub-$5M deals have dropped from over 70% of US venture a decade ago to under half, squeezing the small first round.
• Megadeals of $100M or more took 87.5% of the $412.7B deployed in H1 2026.
• Size your raise in milestones plus slack, not in months. Most honest plans land at 24 to 30 months.
• Model the no-raise scenario. It improves your terms in every conversation, including the ones that go well.

Is it a bad time to raise a seed round?

It is a harder time to get one and a better time to have one. Fewer rounds are closing, but the ones that close are larger. Plan for a longer process and a higher evidence bar rather than assuming the door is shut.

How much runway should a seed round buy in 2026?

More than the 18 months most templates assume. Work backward from the milestone that unlocks your next round, add six months to run the raise, then add six months of slack. Most realistic plans land between 24 and 30 months.

Why is Series B down when everything else is up?

Carta's data shows Series B capital down 30% even with slightly more deals, which usually points to valuation resets at that stage rather than a lack of interest. Companies are getting funded at prices closer to where their metrics actually sit.

Do I need revenue to raise a seed round now?

Not always, but the alternatives have to be strong. Deep technical advantage, a team with relevant history, or unusually fast usage growth can substitute. What rarely works anymore is a deck with a market size and no evidence of demand.

Should I take a bridge round if it is offered?

Depends entirely on whether it buys a milestone or just buys time. A bridge that funds a specific, dated proof point is fine. A bridge that funds the same plan for six more months usually just moves the problem, and it prices the next round lower.

1. TipRanks: Carta Data Points to Larger Venture Rounds and Tighter Selection in Q2 2026
2. Carta: State of Private Markets, Q1 2026
3. PitchBook: The incredible disappearing sub-$5M round
4. PitchBook-NVCA Venture Monitor, Q2 2026
5. National Venture Capital Association: PitchBook-NVCA Venture Monitor
6. Carta: Data Desk, private market insights