Cheap Money Didn't Inflate Startups. It Rewired How They Were Built.

In 2021, a startup could raise money on a slide that said it would lose more next year than this year — and that was the point. Burn was a feature. Spend faster, grab the market, raise the next round at three times the price, and let the next investor worry about a path to profit. The math worked because money cost nothing. US venture-backed companies raised nearly $330 billion that year, roughly double the previous record of $166.6 billion set in 2020.³ Then, in seventeen months, the Fed took its target rate from near zero to 5.25–5.50%.² The slide that said 'we'll lose more next year' stopped being a feature and became an obituary.

The official story is that ZIRP inflated startup valuations, and higher rates popped the bubble — a price problem with a price solution. That framing is comforting because it implies the cure is just waiting it out. The real story is worse: cheap money didn't only set the prices. It set the company designs. A decade of free capital didn't inflate startups so much as build them — wiring 'growth at any cost' into hiring plans, cap tables, and boardrooms — and you cannot un-inflate a business model the way you can un-inflate a multiple.

First, kill the myth of one long decade

The 'decade of cheap money' is a useful story and a slightly false one. The Fed actually ran two separate near-zero stretches: December 2008 through December 2015, then again March 2020 through March 2022, with a tightening cycle in between.¹ This matters more than pedantry. The second episode landed on an ecosystem that had spent the first one learning that aggression got rewarded — and then poured pandemic-era liquidity into it. The result wasn't a slow inflation; it was a vertical one. As of 2015 there were roughly 140 unicorns. By 2018, around 340. By the end of 2021, more than 1,100 — a stretch one analysis flatly calls a frenzy.⁷ The herd didn't grow. It exploded.

Notice the engine, though, not just the output. In 2021, nontraditional investors — hedge funds, crossover funds, sovereign money chasing yield that bonds no longer offered — participated in nearly 77% of total US deal value.³ These were not patient venture investors building over a decade. They were tourists from the public markets, and they were there for one reason: when safe assets pay nothing, money goes looking for risk it would never normally touch. ZIRP didn't make startups better. It made everything else worse, and venture caught the overflow.

The mechanism: free money is a different kind of money

Here is the causal core. An interest rate is the price of waiting. When that price is near zero, a dollar of profit ten years from now is worth almost as much as a dollar today — so a business that promises enormous future cash flows in exchange for enormous current losses looks rational, even brilliant. Discount rates do the math, and at zero the math always favors the company that grows fastest, regardless of whether it earns anything. So the optimal strategy under ZIRP was genuinely to burn: hire ahead of revenue, subsidize customers below cost, buy growth, and trust the next round to refill the tank. That is not recklessness. Given the inputs, it was correct.

Which is exactly why the hangover is structural rather than cosmetic. The companies built this way didn't merely carry high valuations; they carried high metabolisms — cost structures, headcounts, and unit economics tuned to a world where the next round always came and always came up. When the Fed lifted the discount rate, it didn't just lower the price of these companies. It invalidated the assumption their entire operating model was built on. A valuation can reset overnight. Re-engineering a company that was designed to lose money toward one that makes money takes years — and a lot of the time, layoffs.

The trough came later than everyone remembers

The popular memory is that 2022 was the crash. It wasn't. US VC deal value fell 30.8% that year to $238.3 billion — a real decline, but still 39% higher than 2020.⁴ 2022 was the car coasting after the foot left the gas. The actual floor came in 2023, when the contraction reached every stage at once: global funding fell 53% year-over-year in Q1 2023 to $76 billion, and even counting OpenAI's $10 billion and Stripe's $6.5 billion raises, every stage was down 44–54%.⁵ The tourists left first. Crossover investors backed 1,046 US rounds in 2023, a 50% drop from the prior year.⁸ New unicorn formation collapsed to 225, the lowest since 2019, and growth-stage median valuations fell 64% from their 2021 peak.⁸ The delay is the tell. A pure bubble pops on impact. A structural reset bleeds through the system one fund, one bridge round, one runway extension at a time.

Isn't this just a normal cycle that'll bounce back?

The fair objection is that venture is cyclical by nature — funding rises, funding falls, the dot-com bust came and went, and survivors always emerge stronger. By that read, the ZIRP hangover is just the down phase, and capital will flood back the moment rates ease. There's truth in it. The numbers themselves push back on the apocalypse framing: 2022 funding still topped 2020, and the best companies of any era are forged when money is hard to get. A reset that kills businesses which only ever worked at zero is doing its job, not breaking the system.

But the cyclical read mistakes a price for a structure. What changed wasn't sentiment; it was the design assumptions of more than a thousand companies that had raised something on the order of $700 billion against a combined valuation near $4 trillion by the end of 2021.⁷ Those cap tables carry preference stacks and last-round prices that a returning bull market cannot simply forgive. A down round isn't a mood — it triggers anti-dilution clauses, resets employee equity, and reorders who gets paid in an exit. That is governance debt, not market timing. Even if rates fell tomorrow, the companies engineered for free money would still have to be rebuilt for expensive money, and that rebuild runs on the slow clock of a full fund cycle, not a single rate decision. The hangover clears when the 2025–2030 vintages have written off the last of what the prior decade financed — not when the Fed blinks.

ZIRP's real legacy isn't a list of dead unicorns. It's a generation of operators and investors who learned, in their formative years, that the right answer was always 'grow faster, raise more.' That instinct was rational at zero and ruinous at five and a quarter. The hangover is the long, unglamorous work of unlearning it — re-running the math at a discount rate the industry forgot it would ever see again. Cheap money didn't just buy startups too much runway. It taught them to fly a plane built for an atmosphere that no longer exists, and the crash everyone called a bubble is really the slow business of learning to land.