How valuations are set
No formula gives the value of a young startup. Learn the four forces that set the price, and how to turn your target raise into a valuation cap.
In this lesson you will learn
- Why the methods that value mature businesses do not work for early startups
- How to set a valuation cap from the money you must raise and the part you agree to sell
- How to read market medians and percentiles, and how the AI premium works
- Why the highest valuation is not always the best offer
A seed investor tells Ada and Ben: “We think that Northwind is worth $16M.” Where does that number come from? Northwind has a product, a small number of customers, and no profit. No formula turns those facts into $16M. The number is a price, and four forces set it: the money that the company must raise, the market, similar companies, and the competition between investors.
Why no formula gives the answer
A valuation is the price at which one investor agrees to buy one part of the company. The lesson pre-money and post-money calls it the result of a negotiation, not a measurement. For a young startup, the usual tools of finance have little to measure.
The classic method for a mature business is a DCF, a discounted cash flow. It adds the cash that the business will make in future years, at the value of today. A startup at the seed stage has no history of cash flows. A small change in its growth forecast can change the result by a large amount. Thus investors rarely use a DCF for early startups.
Investors and founders use simpler tools. Each tool gives a range, and the negotiation selects a point in that range.
Start from the money and the percentage
Y Combinator, which wrote the SAFE, tells founders: “Pick the fundraise amount. The valuation cap follows.” The founders answer two questions first, and the valuation comes from the answers. For a SAFE, that valuation is the valuation cap, the highest price at which the SAFE converts into shares.
The first question is how much money the company must raise. The money must pay for the work that gets the company to the milestones for its next round, plus a safety margin. At the seed stage in 2026, the margin must be large, because the time between rounds is long. In Q1 2026, the median time from a seed round to a Series A on Carta was approximately 2 years. The lesson when and how much shows how to calculate this amount.
The second question is what part of the company the founders agree to sell. Market data gives a guide. In Q2 2026, the median priced seed round on Carta sold 19.4% of the company. In 2025, the median pre-seed deal on PitchBook-NVCA was $0.5M at an $8.3M pre-money valuation, approximately 5.7% of the company.
Predict first
You want to raise $1M on a post-money SAFE, and you agree to sell 15% of the company. What post-money valuation cap does that give?
Type a guess before you continue. A guess first, even a wrong one, makes the answer easier to remember.
The cap is $6.67M. The investors own their money ÷ the post-money cap, so the cap = $1M ÷ 15% = $6,666,667. A common error multiplies $1M by 15 and gets $15M. The YC Post-Money Safe User Guide uses this example and rounds the cap to $6.7M. The equivalent pre-money valuation is approximately $5.67M.
On a post-money SAFE, the investors own their money ÷ the cap. The formula is the equation of the first lesson, in the other direction:
Post-money = Money ÷ Part sold
$6.67M = $1M ÷ 15%
Pre-money = $6.67M − $1M
= $5.67M
“Money” is the money that you raise, and “part sold” is the part of the company that you sell.
The cap fixes a price, not an amount of money. If you raise only $500K on that cap, you sell $500K ÷ $6.67M = 7.5%. If investors offer $2M on the same cap, you sell 30%. The SAFEs module shows how several SAFEs on different caps add together, in the lesson post-money cap math.
Look at the market
Investors compare each deal with recent deals at the same stage. The median is the middle deal: half of the deals are higher and half are lower. Two data sources give different medians, because they count different deals:
A median is not a target. Many deals are much higher. A percentile shows the spread. The 90th percentile is the deal with 90% of deals below it. In Q1 2026, the 90th percentile valuation on Carta was almost 4 times the median at the seed stage. At the Series A, it was almost 5 times the median. News stories usually report the deals at the top, so the “normal” rounds in the news are not normal.
For SAFE rounds, Carta gives the median cap by the size of the round:
The table shows that larger rounds get higher caps, and that caps increased from 2022 to 2026. Remember the two facts when you read an example in a guide. YC’s example of $1M for 15% gives a $6.67M cap. In Q2 2026, the median cap for rounds of $1M to $2.4M was $18M. At $18M, the same $1M sells 5.56%.
Calculator
Set the money and the percentage. The cap follows.
Type the amount you want to raise on post-money SAFEs, and the part of the company you agree to sell.
Type an amount from $10K to $25M.
- Post-money cap
- $6.67M
- $1M ÷ 15%
- Pre-money
- $5.67M
- $6.67M − $1M
- Median cap
- $18M
- Q2 2026 only, rounds of $1M to $2.4M
- At the median cap
- 5.56%
- $1M ÷ $18M
Your cap against caps for rounds of $1M to $2.4M, Q1 2025 to Q2 2026
- Your cap
- $6.67M
- 10th percentile
- $8M
- 25th percentile
- $10M
- Median
- $15M
- 75th percentile
- $25M
- 90th percentile
- $40M
What you sell if the round changes size
- $500K raised
- 7.5%
- $1M raised
- 15%
- $1.5M raised
- 22.5%
- $2M raised
- 30%
To sell 15% for $1M, set a post-money cap of $6.67M. Your cap is below the 10th percentile of caps for rounds of $1M to $2.4M.
Three of four rounds of this size had a higher cap, so their investors paid more for each part of the company. At the Q2 2026 median cap, the same $1M sells 5.56%.
Move the percentage from 15% to 5%, and look at where your cap goes among the market caps. Then change the amount. At $250K, $500K, $1M, and $2.5M, the widget changes to a different band of round sizes.
Compare with similar companies
Comparables are recent deals of similar companies: the same stage, the same type of customer, and similar traction. An investor says, for example: “Companies like yours raised their Series A at a value of 15 to 25 times their annual recurring revenue.” Annual recurring revenue (ARR) is the yearly value of the recurring contracts that are active today. Those numbers are a revenue multiple:
Revenue multiple = Valuation ÷ ARR
Valuation = Multiple × ARR
$45M to $75M = 15x to 25x × $3M
These multiples are an invented example, not market data. Real multiples change with the sector, the growth rate, and the market. A company with no revenue has no revenue multiple, so comparables at the pre-seed stage use the caps and valuations of similar rounds.
Comparables can mislead in two ways. First, a headline valuation can include terms that give the investors more, such as a large option pool or a large liquidation preference. Second, the deals that you read about are usually the largest deals, not the typical ones. Use a comparable deal as a rough guide, unless you know its full terms.
The competition between investors
A price is not possible without a buyer. If only one investor offers a term sheet, that investor has much control over the price. If three investors want the same round, the founders can ask for a higher price or for better terms. Then each investor must make its offer better. For this reason, many founders meet many investors in a short period. The lesson meetings to term sheet shows how founders get that competition.
The AI premium
In 2025 and 2026, investors paid much higher prices for some AI startups. The AI premium is the difference between those prices and the prices for other startups. It is large at some stages, and it goes mostly to one type of company.
The premium at the seed stage was small. At the Series A, it was large, but Carta’s data shows where it goes. Companies that use AI in an application got approximately 18% more than other companies. A foundation model is a large AI model that other products use, and few teams can build one. Companies that build foundation models got 5.45 times the median of other companies. A startup does not get the premium because its pitch says “AI”.
Why the highest valuation is not always the best
A higher valuation sells less of the company for the same money. That is a real benefit. But a very high price also has costs:
- The next round must pay a higher price per share. If the value of the company does not increase to that price, it gets a flat round or a down round. In Q1 2026, 11.4% of rounds on Carta were down rounds.
- A high price can come with terms that pay the investors first in a sale. The lesson two offers compares a high valuation with dirty terms against a lower valuation with clean terms.
- A high cap on a SAFE also sets a bar. Investors compare the next priced round with the cap. If the round values the company at less than the cap, the round looks flat or down to them.
Is a lower valuation a sign of a weak company?
No. A valuation depends on the stage, the sector, the market at that time, and the number of interested investors. Think of a founder who raises the necessary money at a fair price, with clean terms. That founder can do better than a founder who waits months for a higher price. Time without money is also a cost.
A worked example
Worked example · step 1 of 5
Northwind sets the cap for its pre-seed SAFEs
- 1
Calculate the money
Northwind plans to spend $50,000 a month. Ada and Ben want 24 months of runway to get to their seed milestones.
Money = $50,000 × 24 months = $1,200,000
Now do the steps yourself.
Work it out
You want to raise $750K and sell 10% of the company. What post-money cap does that give?
Post-money cap = $750,000 ÷ 10% = $7.5M. The pre-money equivalent is $7.5M − $0.75M = $6.75M.
Work it out
In the YC standard deal, YC invests $125,000 on a post-money SAFE for 7% of the company. What post-money valuation does that part of the deal imply?
$125,000 ÷ 7% = $1,785,714, approximately $1.79M. This number is an implied valuation: YC sets the money and the percentage, and the valuation follows. The other $375,000 of the YC deal uses a different SAFE, which the lesson the YC deal explains.
Work it out
Maple Data planned to raise $1M on a $10M post-money cap. Investors put in $2M on the same cap. What part of the company did the SAFEs sell?
$2,000,000 ÷ $10,000,000 = 20%, twice the 10% of the plan. The cap stayed the same, so each extra dollar sold more of the company.
Quiz · question 1 of 6
Test what you learned
You want to raise $1M and sell 15% of the company. What post-money cap does that give?
Key takeaways
- 1.An early-stage valuation is a negotiated price. The money that the company must raise, the market, comparable companies, and the competition between investors set it.
- 2.Post-money valuation = the money you raise ÷ the part of the company that you sell. $1M for 15% gives a cap of approximately $6.67M.
- 3.A median is the middle deal, not a target. In Q1 2026, the 90th percentile seed valuation on Carta was almost 4 times the median.
- 4.The AI premium is real but narrow. Most of it goes to companies that build foundation models.
- 5.A very high valuation sets a high bar. The next round must pay a higher price, or the company gets a flat round or a down round.
Sources
- 1.The SAFE. Y Combinator, Accessed 2026-11-04.
- 2.Post-Money Safe User Guide (v1.2). Y Combinator, 2023-02.
- 3.The YC Deal. Y Combinator, Accessed 2026-11-04.
- 4.In the pre-seed market, valuation caps on new SAFEs keep getting bigger. Carta, 2026-08-25.
- 5.Seed Funding: A Startup's Guide to Raising a Seed Round (Q2 2026 benchmarks). Carta, Accessed 2026-11-04.
- 6.Series A Funding: How to Raise Your First Priced Round (Q2 2026 benchmarks). Carta, 2026-10-07.
- 7.State of Private Markets: Q1 2026 (full report). Carta, 2026-05-29.
- 8.Q1 2026 PitchBook-NVCA Venture Monitor. PitchBook and NVCA, 2026-04.
- 9.Q3 2026 PitchBook-NVCA Venture Monitor. PitchBook and NVCA, 2026-10.
- 10.Q4 2025 PitchBook-NVCA Venture Monitor. PitchBook and NVCA, 2026-01.
- 11.State of Pre-Seed: Q2 2026. Carta, 2026-08-13.
- 12.Carta Data LinkedIn archive (Peter Walker, 2025 benchmarks post of December 22, 2025). Carta, 2025-12-22.