Every few months, it happens again. A headline detonates across social media—another $100 million seed round, another paradigm shift, another reason for founders everywhere to question whether their fundraising plans are hopelessly small-time. The screenshots proliferate. VCs add fire emojis. And somewhere in Brooklyn or Austin or Palo Alto, an entrepreneur stares at their pitch deck and wonders if asking for $3 million makes them look naïve.
The numbers, though, tell a story that rarely trends.
According to Carta's snapshot from 2025, 68% of seed rounds were under $5 million. According to Carta's 2025 data, the median priced seed round was $3.7 million with approximately 20% dilution. These figures don't generate much engagement on LinkedIn. They don't make for viral moments on X. But they represent the actual experience of the vast majority of companies being built right now, in this market, with real founders navigating real cap tables.
The gap between what breaks through the noise and what happens behind closed doors has, perhaps, never felt wider.
What Actually Happens at Seed
In Q2 2024, Carta logged a median seed round size of $3.4 million in cash. Seventy-one percent of all seed rounds came in under $5 million. By early 2025, the median for priced seed rounds was $3.7 million with a $19.2 million post-money valuation according to Carta.
For seed rounds structured as SAFEs in the $2 million to $4.9 million range, the median raised was $2.8 million with an $18 million valuation cap. Implied dilution: roughly 18.3%.
These numbers have been remarkably stable, even as the headlines have grown more breathless. Cooley's law firm data from early 2025 paints a similar picture: Series Seed median pre-money valuations held roughly flat quarter-over-quarter, while Series A medians climbed to $52.5 million. What's shifted isn't the typical seed round. It's the composition of outliers, and the volume of attention they command.
The Outlier Problem
Mistral AI's €105 million seed round in June 2023—roughly $113 million at the time—became one of Europe's largest on record. It also became a distorting reference point for thousands of founders who lack, well, Mistral's very particular combination of world-class AI research talent and foundation model ambitions.
These mega-rounds are statistical outliers. They skew averages without budging the median, which is precisely the problem. When a handful of AI companies raise $50 million, $75 million, or $100 million at seed, the mean creeps upward even as the lived experience of the typical founder remains unchanged. This is what happens when power-law distributions meet a media ecosystem optimized for extremes.
The pattern continues. Flourish Care announced a $5.7 million seed round in March 2026 for maternal health technology—a solidly mid-range figure that generated little fanfare. UnityAI raised $8.5 million for healthcare workforce agents, landing at the upper end of normal rather than anywhere near the viral threshold. These are the rounds that actually close, quarter after quarter. They fund real companies solving real problems, and almost no one screenshots them.
Where the Money Is Going
The market has bifurcated, though maybe that's too polite a term. Call it what it is: concentration.
The final quarter of 2025 saw seed post-money valuations hit a record median of $24 million, per Carta data published in March. Around the same time, the first quarter of this year reportedly reached an all-time high for global startup funding—approximately $297 billion, according to Crunchbase figures cited by TechCrunch. The money is flowing. It's just flowing to fewer places.
PitchBook's 2026 U.S. venture outlook, released late last year, flags AI as attracting a disproportionate share of capital. Multistage firms are moving earlier, deploying seed checks amid the AI frenzy. Andreessen Horowitz reportedly raised roughly $15 billion across new funds in January, adding dry powder that inevitably influences check-writing behavior at seed and Series A. These firms can write $10 million, $20 million, even larger seed checks to companies they believe have escape velocity—often in AI, often in deep tech.
For everyone else? The game has gotten harder, just in different ways.
Bridge rounds and extensions now represent 46% of all seed deals in early 2025, the highest rate Carta has ever recorded. For the full year prior, 39% of seed deals were bridges, up from 31% in 2022. The implication: founders are extending runway with smaller top-ups rather than graduating to Series A at the clip seen in previous cycles.
Time between rounds has stretched accordingly. The median span between Series A and B hit an all-time high of 2.8 years as of last May. Seed-to-Series A graduation has slowed materially since the 2021 peak, with falling two-year graduation rates across seed cohorts becoming a recurring theme in Carta's commentary.
The SAFE Default
When you're raising less than $4 million, you're almost certainly doing it on a SAFE. Carta and related analyses from the past year or so show cap-only, post-money SAFEs as the dominant instrument for pre-seed and small seed rounds. Y Combinator's standard accelerator deal, as of this year, remains $500,000 total: $125,000 for 7% equity on a post-money SAFE, plus $375,000 on an uncapped most-favored-nation SAFE that converts at the next priced round.
It's become a template, and for good reason. It sidesteps the complexity and legal expense of a priced round for companies still hunting for product-market fit. It defers valuation questions to a stage when the company has more negotiating power. And it reflects a pragmatic acknowledgment that most seed-stage companies are raising amounts that fit comfortably within this structure.
Priced seed rounds, when they do happen, typically involve around 20% dilution—a benchmark that's held steady across multiple datasets from 2024 through 2025. Pre-seed dilution on conversion tends to run about 10%, with final dilution set at the first priced round. These are the mechanics that govern the bulk of early-stage fundraising, far removed from the valuation gymnastics of nine-figure rounds.
The View from Elsewhere
The U.S. market sets benchmarks, but other ecosystems paint their own portraits. Australia's 2025 State of Startup Funding report, published in February by Cut Through Venture and Folklore Ventures, shows a median seed of approximately AU$2.5 million (roughly US$2.5 million equivalent), with angel and pre-seed rounds around AU$1.0 million and Series A at AU$11.0 million.
Latin America's LAVCA Trends in Tech report for 2025, covering data through late last year, documents a contraction in activity as the region adjusts to the global pullback—though interest at seed and Series A persists. Europe's Atomico report for 2025 shows medians by stage reflecting a retreat from 2021 peaks but some stabilization. In each market, the median remains far below the headline-grabbers.
What This Means Going Forward
The disconnect between viral fundraising stories and actual median round sizes shows no signs of closing. It may widen.
Cambridge Associates' 2026 outlook, published late last year, advises investors to moderate commitments to seed-focused strategies unless truly exceptional, citing maturation of seed investing, elevated early-stage valuations, and a higher bar to exit. That caution applies to the median, not to the handful of companies that can command Series A-sized seed rounds.
For founders, the most useful framing comes from the cumulative evidence across Carta, Cooley, and NVCA/PitchBook datasets: median U.S. seed cash raised remains around $3 million to $4 million. The majority of seed rounds are under $5 million, even now. These are the numbers to build a pitch around. These are the expectations to set with co-founders and early employees over late-night whiteboard sessions.
The $100 million seed round exists. But calling it typical is a category error. It's a different species of financing, available to a different species of company—often in AI, often in deep tech, often led by repeat founders with established track records and Rolodexes that open doors most entrepreneurs can't reach.
For the other 68%, the path remains what it has been: raise a few million dollars, prove the concept actually works, manage dilution carefully, and either graduate to Series A or extend with a bridge when timing demands it.
The myth sells better than the median, obviously. But the median is where most companies are built—one modest seed round at a time, no fire emojis required.
