VC新门槛：Series A需展示250亿美元退出路径

So my co-investor in $2.3B Owner.com (Vertical AI for restaurants) just moved to growth partner at Menlo Ventures and put out a post and chart this week that explains how venture capital works in 2026:

10 years ago, VCs were hunting unicorns. Then decacorns. Now? $25B+ exits.

There are 60+ of them now, so while they are still outliers, they’ve become … predictable. And that’s a 13.5x increase in the supply of the outcome large funds are chasing. At that supply, $25B becomes the target.

The word “unicorn” was coined in 2013 to describe something rare. There are now more than 1,300 of them. $25B is the number partners are running in their heads when they take your Series A meeting.

6, Then 23, Then 81

Break the bars apart and the public and private lines move at different rates.

• Public companies at $25B+ market cap: 6, then 18, then 60. A 10x over twenty years.
• Private companies at $25B+ valuation: 0, then 5, then 21.

The private column went from not existing to a quarter of the total. Twenty years ago, if you wanted to be worth $25B you had to be public. There was no other mechanism. Today 21 companies carry that value while still private.

PitchBook’s July 2026 data says the same thing from another angle: 63 active US decacorns, up from 53 last year and 26 in 2021. Nineteen new ones crossed $10B in the first half of 2026, already past 2025’s full-year total of 18. Mega-deals of $100M or more accounted for 87.5% of the $412.7B invested in the first half.

Capital that used to arrive at the IPO now arrives at the Series D, E and F. The returns that used to come from the public market have to be manufactured while the company is still private.

$60B in August. $48B Three Weeks Later

The 81 already needs updating.

On August 14, SpaceX closed its acquisition of Anysphere, the company behind Cursor, in an all-stock deal valued at $60B. About 391 million Class A shares issued, confirmed in an 8-K, and widely described as the largest acquisition of a venture-backed startup on record. Cursor was worth $2.5B at the start of 2025 and $29.3B at its Series D last November. It exited at roughly 15x its ~$4B in annualized revenue, four years after being founded.

Then on September 8, Cognition announced a Series E of more than $2B at a $48B valuation, led by Andreessen Horowitz and Accel with Founders Fund, General Catalyst and Avenir. In May the same company raised at $26B on a $492M run rate. Run rate is now approaching $900M. The valuation nearly doubled in a little over three months.

The world needs far more software than it can build. Cognition exists to change that.

We’ve just raised over $2B at a $48B valuation, led by a16z, Accel, Founders Fund, General Catalyst, and Avenir.

Since our round in May, run-rate revenue has grown from $492 M to almost $900… pic.twitter.com/Z7Ph29GG6B

— Cognition (@cognition) September 8, 2026

Two AI coding companies, both founded within the last four years. One cleared $25B and exited at 2.4 times it. The other cleared $25B in May and passed $48B by September.

$25B is now the entry point to the top tier, and companies are moving through that tier in quarters. When a partner is underwriting against these two comps, a company with a credible path to $2B doesn’t come up.

The Cursor deal deserves one caveat given the “exit” framing. SpaceX paid in stock of a company that had been public for a matter of days. The liquidity is real. So is the fact that the largest startup acquisition on record required a buyer with a newly minted public currency, and there are very few of those.

A $3B Fund Needs Four $25B Companies. Here’s the Arithmetic

Menlo announced $3B in new capital in June, the largest raise in the firm’s 50-year history. They are not unusual.

Most versions of this math assume 15-20% ownership at exit, and nobody holds that anymore. Menlo’s own stated Series A target is around 20% at entry, and entry is the high-water mark. Carta’s 2026 data has the median founding team going from 56% at seed to 36% after the A to 16.1% by Series C. Investors ride the same slide unless they pro-rata into every round, and in the AI cohort the rounds are large enough that most can’t. Call it 8-12% at exit for an early lead that defends its position well.

A $3B fund needs roughly $9B-$10B gross to return 3x. At 10% ownership, a $25B outcome returns $2.5B. That’s four of them. Four, out of the 81 that exist in the entire world, in one fund, in ten years.

Growth funds land in the same place by a different route. Their constraint is entry price, not ownership. A $100M check into a $5B round buys 2%, and the return on that position is set by the multiple on entry: $5B to $25B is 5x whether you own 2% or 6%. To get a 3x fund out of positions bought at $3B-$5B, the companies have to reach $25B and keep going. So the threshold binds hardest at the stage where the most capital sits.

Run the same math on a $1B outcome. At 10% ownership that’s $100M, about 1% of what a $3B fund needs to return itself once. A billion-dollar exit, the thing the industry named a mythical creature after, moves a large fund’s DPI by almost nothing.

Fund sizes went up roughly 10x over this period. Exit ownership went down. That’s the pass most founders are getting, dressed up as feedback about their market.

Three Things the Chart Doesn’t Say

The direction is right and it’s large. The magnitude deserves scrutiny before anyone builds a strategy on 6 to 81.

• The threshold is nominal. $25B in 2006 dollars is about $41B in 2026 dollars, on 65.6% cumulative CPI inflation. Hold the bar constant in real terms and the count grows well under 13.5x. Some of the third bar is the dollar.
• Private valuations aren’t exits. The framing here is “$25B+ exits,” and 21 of the 81 haven’t exited anything. A private mark is the price one buyer paid for one slice, usually with liquidation preferences and structure attached. The last two cycles produced a long list of companies that carried a mark for years and never got near it in a liquidity event.
• Counting companies isn’t counting new companies. Some of the 60 public names on today’s bar were on the 2006 bar and simply got much bigger. A rising market produces $25B+ market caps mechanically. The chart blends new giants created, existing giants compounding, and multiple expansion across the whole index.

The conclusion holds either way. The honest version is that the bar moved up a lot, not that it moved up 13.5x.

What This Changes About Your Series A Pitch

The question the partner is answering internally isn’t “will this work.” It’s “if this works, how big does it get.”

The ceiling matters more than the floor in that room, and a credible path to $25B beats a highly probable path to $500M. What you have to show:

• A market where the winner does $2B+ in revenue, not $200M
• A pricing model that keeps expanding after seat growth stops
• Distribution that compounds instead of being repurchased every year
• Growth on the order of $1M to $100M ARR in about five years, which is the venture bar in practice now

The AI cohort made this bar enforceable. Cursor went from $1M to $100M ARR in about 12 months, then from $2.5B to a $60B exit in roughly 18 months after that. Lovable hit $100M in eight months with 45 people. Once a handful of companies do that, every underwriting model gets rewritten around them, including for the companies that will never be in that category.

The Best Outcome for Most B2B Companies Is Not a $25B Outcome

A $300M exit is life-changing for a founder, an excellent outcome for a seed fund, and irrelevant to a $3B fund.

Raising venture is choosing a distribution of outcomes and agreeing to be optimized toward the right tail of it. If your realistic ceiling is $500M, venture is an expensive way to get there, and the pressure to swing for a ceiling you can’t reach will cost you the outcome you could have had.