The Seed Round You Raise Now Has To Last 25 Months

It got cut in half, slowly enough that most founders never noticed the ground move.

Carta tracks how many companies that raise a seed round go on to raise a Series A within two years. For companies that closed seed in the first quarter of 2018, that figure was 30.6%. For the 2022 cohort, it fell to roughly 15.4%.

One in three, then one in six.

The ones that do graduate also take much longer. Median time from seed to Series A stretched to about 774 days by late 2024. At the 2021 peak it was 420 days.

The whole story in two data points. Fewer companies clear the bar, and the ones that clear it wait almost twice as long.

Here is why it matters. Most seed rounds are still sized and modeled as if the old clock applies. Founders raise for "18 months of runway" because that is the number in every template. Eighteen months is about 540 days. The median graduating company needs 774.

So the default plan runs out of money roughly eight months before the median version of the event it exists to fund.

Because runway does not fail gradually. It fails all at once, and it takes your negotiating position with it.

Run the sequence. You close seed with 18 months of cash. Month 12 you start Series A conversations, because everyone says start six months out. A real process takes four to six months when it works. Month 16 or 17 you either have a term sheet or you do not. If you do not, you have weeks of payroll left and every conversation becomes about survival. Bridge notes priced from that position are expensive, and everyone in the room knows why you are there.

Now run it with 30 months. Month 18 you start. It stalls on retention, or a metric a partner could not get comfortable with. You have twelve months to fix the actual thing and come back with better numbers instead of a smaller ask.

Those two companies can be identical on every dimension that matters. The only difference is calendar.

So stop sizing rounds in months. Size them in milestones plus a failure buffer:

1. Pick the specific metric that makes an A possible in your category.
2. Estimate how long it takes to hit, then strip the optimism out of that estimate.
3. Add six months for running the raise.
4. Add six more for the version where the first pass does not close.

That sum is your number, and it usually lands 40% to 60% above your original model. Founders flinch, because a bigger round means more dilution today. But dilution from a bigger seed is a cost you choose. Dilution from a distressed bridge is a cost someone else chooses for you.

The bar moved on every axis at once, which is why partial progress feels so unrewarded.

Where a strong team and under $1M of revenue could raise an A in 2021, the 2026 version wants roughly $1M to $2M ARR growing about 3x year over year, with retention that holds. Some benchmark sets put the median entry point closer to $3.5M ARR.

But ARR alone stopped being the test. The composite is what gets diligenced:

• Net revenue retention around 120% at the median, meaning existing customers grow on their own
• Burn multiple under 2x, meaning less than two dollars of net burn per dollar of new ARR
• Growth that has held for three or four consecutive quarters, not one good one

The round got bigger too. Median Series A post-money now sits around $75M to $85M on a median raise of $13M to $15M, up from $8M to $10M.

That combination is the trap. A $75M post-money A carries a Series B expectation attached to it, and the number of companies that can support that arc is exactly why the graduation rate fell. Funds are not writing more A checks. They are writing larger checks to fewer companies.

If you are at $600K ARR growing 2.2x with 105% retention, you are doing fine work. You are also not raising an A this quarter, and no narrative fixes that. Better to know it in month 10 than month 22.

Almost certainly not, and the arithmetic is worth doing yourself.

Global startup investment hit about $297 billion in the first quarter of 2026, roughly 2.5x the $118 billion of the previous quarter. Every headline framed it as a boom.

Then look at composition. OpenAI collected about $122 billion in a single round at an $852 billion valuation. That one company took roughly 41% of the quarter.

The concentration continues down the stack. AI now captures roughly 86% of US venture dollars. OpenAI and Anthropic together accounted for about 43% of all first-half funding. The top ten VC funds took close to 43% of all venture capital deployed in Q3 2025, the highest share in a decade.

Strip the megarounds out and the market available to a normal seed-stage company looks tighter than 2023.

So if you are building something that is not AI-native, plan longer runway, earlier revenue, and less reliance on a follow-on arriving on schedule. If you are AI-native, the money exists and so does everyone else's pitch.

Either way, stop reading market totals and start reading your slice. Ask three investors in your category how many A checks their fund wrote in the last twelve months. The answer is usually smaller than you expect.

Sometimes the bigger round is not available. Then you have three levers, and only three.

Lower the cost base. Not a hiring freeze after the fact, but a plan built around smaller headcount holding longer. The biggest driver of seed-stage burn is the fourth through eighth hires, usually made in the two months after close, when confidence is highest and evidence thinnest. Delay each by a quarter and you buy months.

Pull revenue forward. Annual prepay with a discount, pilot fees instead of free pilots, a services wrapper that funds the product. Services revenue is unloved by investors. It is also cash you do not need permission for.

Change the shape of the raise. A seed extension planned in advance, from investors who already believe, is a different instrument from a bridge negotiated in a panic.

What makes this work is modeling all three before you need them, with written trigger conditions. If ARR is below X by month 14, we cut Y and start Z. Deciding that in month 14 with a board watching is much harder than deciding it in month 2.

You can lay this out in a spreadsheet, in Notion, or in a planning tool like Foundra that walks first-time founders through the projections section step by step. The tool is not the point. Having the scenarios written down before month 14 is.

Then you join the roughly five in six companies that do, and you should know in advance what that means.

It does not mean the company dies. A meaningful share of that cohort is still operating, just not on venture's schedule. The failure is rarely the missed round. It is having built a cost structure that only works if the round arrives.

The options, in rough order of how much optionality they preserve:

Default alive. Cut to a burn your current revenue covers within twelve months. The only option that does not require anyone else to say yes, and it improves every other option, because a company that does not need money negotiates differently.

Seed extension. Best done early, from insiders, before the metrics soften.

Revenue-based financing or venture debt. Works with predictable recurring revenue and real gross margin. Not a substitute for product-market fit, and lenders can tell.

Acquisition. Most seed-stage acquisitions happen because the buyer wants the team.

Avoid the option nobody names: a small structured round on bad terms that postpones the decision by six months. That is not a bridge. It is a slower version of the same problem with a worse cap table.

• Seed to Series A graduation fell from 30.6% for the Q1 2018 cohort to about 15.4% for the 2022 cohort.
• Median time from seed to A stretched to about 774 days, up from 420 at the 2021 peak. An 18 month plan is short by roughly eight months.
• Size by milestone plus buffer: time to metric, six months to raise, six months for a failed first attempt.
• The 2026 A bar is a composite: $1M to $2M ARR minimum, about 3x growth, roughly 120% net revenue retention, burn multiple under 2x.
• Q1 2026's record $297B included a single $122B OpenAI round. The headline market is not your market.
• Model the three survival levers in month 2 with written triggers, not in month 14.

How much seed money should a first-time founder raise in 2026?

Enough to reach a Series A metric plus twelve months of buffer, which usually means 30 to 36 months of runway rather than 18. For most software companies that lands between $2.5M and $5M depending on team size and salary levels.

Should I raise a bigger seed even though it means more dilution?

Compare it against the alternative. A larger seed at a fair price costs known equity today. A distressed bridge at month 17 costs unknown equity plus structure plus control terms, at a valuation set by your weakness. The larger round is usually cheaper on a risk-adjusted basis.

What is a burn multiple and why do investors ask about it now?

Net burn divided by net new ARR over the same period. Spend $1.6M to add $1M of ARR and your burn multiple is 1.6x. Under 2x reads as efficient at Series A. It became standard because it exposes growth that was purchased rather than earned.

1. Carta: Graduation rate from seed to Series A
2. Chronograph: The Series A Crunch, Exploring Seed Graduation Rates
3. CRV: Series A Metrics VCs Expect in 2026
4. Value Add VC: Series A Traction Requirements 2026 vs 2021
5. Value Add VC: The State of VC Funding in 2026
6. TechCrunch: Startup funding shatters all records in Q1