The exit that didn’t come

For most of the last decade, the exit was less a plan than an assumption — a date nobody had set but everybody expected, an IPO or an acquisition large enough to make the cap table whole. Then the door closed, and it closed fast. In 2022, exactly fourteen venture-backed companies went public in the United States, raising $1.7 billion between them, according to Renaissance Capital. Four years earlier the count had been eighty-eight. The market founders had spent a whole cycle building toward had, quietly, stopped being there.

The window, reopened a crack

It is fashionable now to say the window is back, and by one measure it is. The US saw 202 IPOs raise $44 billion in 2025, the busiest year in four. But the venture-backed slice of that recovery stayed thin: 34 IPOs raising $14.7 billion — up 85% in proceeds on the year before, and still well below the ten-year average of roughly seventy IPOs and $18.5 billion a year, by Renaissance Capital’s own reckoning. A recovery that arrives at half the normal volume is a strange kind of recovery.

It was also a narrow one. Just nine venture-backed technology companies went public all year, against a ten-year average of twenty, and the ones that worked shared a profile: artificial intelligence, digital assets, or a visible path to profit. Figma’s debut popped 250% on its first day — the biggest first-day jump for a billion-dollar IPO on record. CoreWeave and Circle rode the AI and stablecoin trades. For most of the rest, the message from public investors was to wait, and most did, taking private money rather than test a valuation they had set in a louder year.

The companies that stayed

Behind the trickle of listings sits the reason the window feels so crowded. More than 1,500 companies now sit on Crunchbase’s unicorn board, worth a combined $6 trillion, and over 60% of them have not raised at a disclosed valuation in more than three years. These are not failures. They are companies too valuable to be cheap to run, too richly priced to go public at the number on paper, and too far along to pretend the clock isn’t running. SpaceX, the oldest name on the board, has stayed private for more than two decades and never listed at all.

The mechanism is dull and unforgiving. A valuation set in 2021 becomes a ceiling the company has to grow into before it can raise or sell again — and for most, growth slowed before it got there. Staying private, even indefinitely, turns out to be cheaper than admitting the number was too high. So the backlog builds, one deferred exit at a time, and the pile of aged, unexited value gets deeper than any plausible year of IPOs and acquisitions could drain.

What building now actually means

None of this is an argument against starting something. It is an argument for starting it with clearer eyes about the ending. If you are building now, a few things are worth holding as base cases rather than as fears:

• The IPO is a possibility, not a plan. The market reopens for a profile — AI, real margins, genuine scale — well before it reopens for the median company.
• Staying private longer is the normal path now, not the failed one. In that backlog you are in company, not in disgrace.
• The valuation you celebrate on the way up is the number you have to beat on the way out. Pricing high is a debt you repay at exit, if you exit.

Renaissance expects the thaw to continue — 200 to 230 IPOs in 2026, by its estimate — and it may well be right. But a thaw is not a return to 2021, and the founders who come through it best will likely be the ones who built as if the exit might never arrive: durable, close to profitable, worth owning whether or not a door ever opens. The exit was always the reward for the company, not the point of it. The last few years just made that harder to forget.