The eight-week-old startup had six-figure customer contracts and a valuation ask of $40 million. Nobody blinked.
That startup, presenting at Y Combinator's Winter 2026 Demo Day in March, was not an anomaly. It was the market. Ashley Smith, a general partner at the early-stage fund Vermilion, attended the Demo Day and told TechCrunch that companies were pricing rounds years ahead of their actual traction. The YC tax, the premium investors have long paid just for the accelerator's brand, no longer explains the numbers.
Seed valuations for AI startups have decoupled from the companies being valued. A $10 million seed round at a $40 million to $45 million post-money valuation is now typical for AI companies, according to Pete Martin, the founder of the cybersecurity startup Realm, who raised at those terms in 2024 and now watches newer companies clear the same bar as a baseline expectation. For non-AI startups at the same stage, the market looks nothing like that.
The data backs the anecdote. Carta, which tracks private market valuations across thousands of funding rounds, found that at Series A, the median AI startup carried a valuation 38 percent higher than the median non-AI startup. By Series E and beyond, that premium widened to 193 percent. Those numbers reflect companies that already have revenue. Seed-stage companies have less.
YC's Winter 2026 cohort is the sharpest available snapshot. Garry Tan, YC's CEO, confirmed that 14 companies in the roughly 200-company batch had crossed $1 million in annualized revenue before Demo Day concluded, three times the number from the prior cohort and the highest total in YC's history. Average weekly revenue growth across the cohort was 14 percent, the fastest ever recorded. The default post-money valuation was $40 million, up from $20 million for the equivalent cohort just three years prior. At the top end, one company's valuation reached $200 million post-money, the highest single-company Demo Day figure in six years of tracking.
These are not companies with five-year product roadmaps. They are companies with months of operating history and contracts that suggest the revenue is real. Investors are treating them as if the contracts are the floor, not the ceiling.
"The best seed-stage companies do not look like traditional seed-stage companies anymore," said Marlon Nichols, a managing general partner at MaC Ventures. His average entry check has risen from $2.5 million in 2019 to $5 million today, and the companies he is writing those checks into are already generating millions in revenue before he closes. His last two seed investments were each producing more than $2 million in revenue, with paid pilots from large enterprises and a clear line of sight to commercial agreements. He wrote $3 million to $4 million checks at post-money valuations of $25 million and $30 million respectively.
What has changed is not the quality of the companies. It is the reference point investors are using to price them. Cursor, the AI-powered code editor, generated $100 million in annual recurring revenue in just 12 months in early 2025. That did not merely set a record for the fastest SaaS company to that milestone. It became the template for what a successful AI company looks like in the minds of investors deciding what to pay today. "The pressure is at an all-time high, not to be a billion-dollar company, but a $50 billion," said Shanea Leven, the founder of the enterprise AI application platform Empromptu.
That reframe is where the valuation gap between AI and non-AI startups originates. VCs are not simply paying high prices for current AI capabilities. They are betting that the companies they are funding will deliver something closer to artificial general intelligence, or at least a step-change beyond today's models, within the lifetime of the fund. Mira Murati's $2 billion seed round for Thinking Machine Labs at a $12 billion valuation is the extreme expression of this logic, but it is not an outlier. It is the market's assumptions made explicit. When a founder with the right pedigree can raise at those terms with a pitch and a team, every seed negotiation starts from a different baseline than it did three years ago.
Seed deal count is down, but valuations are up, partly because large venture firms are moving earlier into rounds and squeezing smaller check-writers out of the deals they once dominated. "You can end up stuck in between," Martin warned. "Too expensive for new investors, but without the traction to justify the next round." Jonathan Lehr, a general partner at Work-Bench, put it another way: higher seed valuations mean less room for experimentation, fewer pivots, and more scrutiny if progress does not match the capital raised. The tolerance for the uncertainty that seed investing has historically required has compressed.
The counterargument is that the traction is real. Companies are reaching revenue milestones in months that used to take years, and AI tools are accelerating the pace at which founders can build and sell. That is a genuine shift in what is possible. Whether the valuation premiums being assigned at seed are consistent with that traction, or whether they reflect expectations that have migrated ahead of current capabilities, is the open question.
