How big is a normal cybersecurity seed round right now, and what must it prove for the A?

You are sizing a round against numbers that are not yours. Behind every median you will be quoted sits a lead investor doing ownership arithmetic, a buyer whose procurement calendar ignores your runway model, and a Series A investor who will judge the seed by the bar its price implies. Those three set your number. The median only tells you what other people announced.

The median is not your number.

You will be quoted two medians, and they count different things. One is a round size built from press releases, and it includes the foundry checks, the premium paid for AI security companies, and rounds that a cap table would call a Series A. The other is a valuation built from cap tables, with bridges and notes left out. Quote either one with its date, and never blend them. Geography moves the number more than sector does, and outside the two coastal metros you read a different line.

What sets your number is the lead's arithmetic, and it has three inputs. The ownership the lead wants. The share of the round it expects to take itself. And a post-money valuation it can defend to its own partners. The round size falls out of those three: target ownership times post-money, divided by the lead's share of the round. Ask for the first two numbers on the first call. A lead that will not say has not decided to lead.

The price is a bar you agree to clear.

You will be told to take the highest price you can get. The price is not a prize. It is the bar for your Series A, because the next investor measures the step up from your seed valuation, and a seed priced on ambition rather than traction quietly becomes an evidence requirement that this runway may never reach.

The behavioral cost is worse than the arithmetic. A founder defending a valuation makes decisions that protect the number, at exactly the stage that rewards searching. The seed is the round where being wrong is supposed to be cheap. Price it so that it is. And know which term you are protecting, because you repay whichever one it is. Protect the price and you repay it at the Series A. Protect your ownership and you repay it in the bridge.

Raise enough to be wrong for two quarters.

You size the round to survive being wrong for two quarters, not to hit the plan with the least dilution. We have watched founders raise the smallest amount at the highest price that covered the plan, miss the plan by two quarters, and pay far more in the financing that followed than the money they declined at the start. Minimizing dilution at the seed is treated as basic hygiene. It is a bet that the plan works to the quarter, and the instrument that rescues a founder who lost that bet is priced against their lack of alternatives.

In the companies we have watched, the bridge has been the ordinary path between a seed and a priced round, not the exception. Fund for it.

Count sales cycles, not months.

Runway modeled in months assumes revenue arrives on a software calendar. Your buyer imposes a different one. The security review of a security vendor, the questionnaire, a proof of concept on production data, legal and the data agreement, procurement, and signature each take their own time. Nobody measures that cycle for you. Follow one deal from first meeting to signature, and use the elapsed time as your cycle. Until one has closed you have not measured it, and saying so is better than modeling it.

Then count how many cycles must close before the Series A, and fund that many plus two quarters of miss. The buyer's fiscal year is the multiplier. A deal that slips past it costs a year rather than a quarter, and that is the difference between an eighteen-month raise and a thirty-month one.

Take the lower clean price.

You take the clean price, and you negotiate what happens on a miss before you sign. Founders optimize the headline valuation and treat the structure as detail their lawyers will handle. Structure determines the outcome in every scenario except the best one. A valuation that is conditional on hitting a number by the close is a repricing clause, and when the number is missed the repricing arrives as structure rather than as a price cut: warrant coverage, a participating or multiple liquidation preference, an inside round priced with warrants.

Investors concede structure faster than price, because a headline number is what everyone sees and structure is what nobody reads until it pays out. That is exactly why a competitive process is worth more for the terms it removes than for the valuation it adds. Structure in a seed term sheet is not the market's default. Ask why it is there.

A second term sheet sets the price.

You get a market price from a second term sheet, not from a narrative. We have watched founders work a wide list of funds for two quarters on the theory that enough interest eventually produces a lead. Interest without a lead is a stable state, and it can persist until your cash runs out. A corporate investor's interest is not a term sheet either.

Buyer money is priced differently.

Seed money in cybersecurity often comes from people who are also your buyer, a cap-table shape that general software does not have: groups of sitting CISOs investing their own money, design partners who ask for equity, and corporations. Three rules apply.

A buyer on the cap table is diligence and a reference, not pipeline. An investor's introductions buy warm meetings rather than budget, so size the round as if the introductions produce no new customers, and be glad when they do. Equity to a design partner is a discount that never expires and a conflict their own procurement team will ask about, so grant it only against a scope of work with an end date, and never instead of a price. A corporate's check is a wire, not a channel, so negotiate the commercial relationship as a separate, dated agreement with obligations in it, at the same time.

Know what the Series A investor reads.

You are judged on a handful of measures moving together, revenue, growth, gross margin, an efficiency ratio, a churn ceiling, and a payback period, and the company is judged on its weakest one. Nobody publishes a cybersecurity-specific Series A bar with a sample and a method behind it. The bars that surface in a search are fundraising advisers' marketing, with nobody who has to defend them, so check for a method before you let a bar size your round.

What a security Series A reads instead of the blended number is the security delta. Net revenue retention can stay excellent inside a market too small to matter, for years, and the thing that would have told you is new logos. Report the new-logo line separately. Show a pipeline with losses in it, because a pipeline with no losses is a pipeline nobody has qualified, and security deals slip rather than decline. Show paid production deployments, not design-partner logos sitting in procurement.