Is the IPO Window Really Open

Start with the basic math. Global unicorns are now worth about 5.2 trillion dollars on paper, spread across roughly 1,300 companies. In 2021, the hottest IPO year on record, global IPOs raised just over 600 billion dollars. In the more "normalizing" environment of 2023-2025, global IPO proceeds have bounced between roughly 120 and 170 billion dollars per year.

Even at the 2021 peak, public markets absorbed only ~10-15% of today's unicorn value in a single year. And at today's run-rate, the IPO market is structurally sized to clear only a small slice of that backlog annually.

The IPO market is not waiting to reopen. It is continuously rationing a fixed number of slots. There is a limited number of growth stories that public investors can support each year at scale. For growth investors, the job is to determine which businesses are eligible to claim a slot in that quota on structure and category signal.

Look at who actually makes it through the window. Historically, US IPOs came public at around 9 years old on average from 1980-2019. In the last few years, that average has jumped into the low teens, with one recent read putting 2024 tech IPOs at roughly 14 years old. On the fundamentals side, median revenue at IPO (in today's dollars) has climbed from about 64 million in 1980 to over 200 million by 2024.

Public markets are no longer the place where product-market fit is discovered. They are where margin visibility and cash-flow credibility are priced.

Layer on ownership. In 2024, nearly half of global IPO proceeds came from PE- and VC-backed portfolio companies. In the US, sponsor-backed IPOs rose to almost 30 percent of all listings in 2024, up from 17 percent in 2023. Sponsors are using the public markets to recycle risk and term out capital instead of discovering early growth.

For LPs, that has a simple implication: if you buy generic late-stage exposure, you're often funding someone else's exit timing. The marginal dollar you put into a broad "pre-IPO" pool is frequently the dollar that lets a mature sponsor hand the risk to public markets at their chosen moment.

Inside that 5 trillion-dollar backlog, there are really two very different stacks of companies.

The first stack is the legacy unicorn cohort-many born in the 2012-2021 zero-rate era. They raised heavily, hired aggressively, and often grew into their valuations with bloated cost structures. Across the unicorn universe, older names tend to show significantly lower revenue per employee than newer entrants. Their cap tables can be layered with multiple preference stacks, structured rounds, and insider recaps. They're long-duration, path-dependent bets.

The second stack is the AI-native cohort. Recent analyses of AI unicorns show they're reaching billion-dollar valuations in roughly two years versus about nine years for non-AI peers, and doing it with around half the headcount (roughly 200 versus 400 employees). In the top quartile, "lean AI" startups can achieve revenue per employee 4-5x higher than traditional SaaS benchmarks: on the order of 3-4 million dollars per employee at the top end, versus roughly 600 thousand for classic software.

These companies are short-duration, high-productivity, but they carry much higher narrative risk. They're all competing for a fixed pool of public-market attention, and AI accounts for a majority of net new unicorn valuation created since 2021. That sets up brutal dispersion: a few names take most of the oxygen; many others never get a real shot at the window.

In a rationed attention environment, second-order beneficiaries often offer superior risk-adjusted outcomes: they participate in AI-driven demand without requiring narrative dominance to access liquidity.

When we talk about "the IPO window" at SPLY, we're asking four practical questions about each company: