You've heard that 90% of startups fail. It's in every accelerator pitch, every LinkedIn hot take, every "brutal truths" thread. Here's the problem: that number is only true for one narrow definition of startup, and it's probably not yours. The actual startup failure rate in 2026 ranges from about 20% to about 90% depending on what you count as a startup and what you count as failure. That gap isn't a rounding error. It changes what you should do about it.
I've spent a lot of time with this data, and the pattern that keeps showing up isn't "startups are doomed." It's that most failures trace back to one preventable mistake, and the headline stats bury it. Let's walk through the real numbers.
What percentage of startups actually fail?
For all new US businesses, about 20% fail in year one, half by year five, and two thirds by year ten. For venture-backed startups chasing 10x returns, the failure rate climbs to 75-90%. Both numbers are real. They measure different things.
The Bureau of Labor Statistics tracks every new private-sector establishment in the country. Its most recent data puts the one-year failure rate at 20.4%, the five-year rate at 49.4%, and the ten-year rate at 65.3%. That covers everything: coffee shops, dental practices, freelance LLCs, and SaaS companies alike.
The scarier numbers come from a different population. Harvard Business School researcher Shikhar Ghosh studied roughly 2,000 venture-backed companies and found that 75% never return cash to investors. The Startup Genome Project pushed further: around 90% of scalable, innovative startups fail to achieve venture-scale returns.
So when someone says "90% of startups fail," ask which population they mean. If you're bootstrapping a small SaaS product to $10k a month, the BLS numbers describe your odds better than the VC numbers do. A coin flip over five years. Not great, not hopeless.
Why do most startups fail?
The number one root cause is building something the market doesn't want. In CB Insights' analysis of 431 failed VC-backed companies, 43% cited poor product-market fit as a root cause, while 70% cited running out of capital. And here's the part most people miss: CB Insights explicitly classifies running out of money as the final symptom, not the underlying disease.
Think about what that means. The companies in that study raised a combined $17.5 billion before shutting down. The median failed company raised $11 million. These weren't teams that couldn't find money. They were teams that spent money building things nobody wanted badly enough to pay for, and the bank balance just recorded the outcome.
The other root causes in the study: bad timing or macro conditions (29%) and unsustainable unit economics (19%). Percentages exceed 100% because dying companies rarely have just one problem.
There's a grim detail in the timeline data too. The median time from a company's last fundraise to its shutdown was 22 months, and nearly a quarter of failed startups operated as "walking dead" for three or more years before closing. Most founders knew something was wrong long before they admitted it.
Failory's independent analysis of 80+ failed startups used a different taxonomy and still landed in the same place: 56% pointed to marketing problems, which in practice usually means "we built it and nobody came." The demand side kills startups. Not the code.
When do startups actually fail?
The highest-risk window is years two through five, and the highest-risk stage is before Series A. First-year deaths are actually the minority: only about 1 in 5 businesses fails in year one.
The stage data is stark. Around 60-70% of seed-stage startups never reach Series A. Roughly 35% of Series A companies never reach Series B. By Series B and beyond, outright failure drops to around 1%. Carta's shutdown data confirms the shape of the curve: 74% of recorded startup shutdowns happened at pre-seed or seed stage, with 41% at seed alone.
Notice what that timing implies. Startups don't usually die during the build. They die after launch, when the product meets the market and the market shrugs. The dangerous period is exactly when most founders skip validation because they'd rather be building.
How many startups shut down recently?
Shutdowns spiked hard coming out of the 2021 funding boom. Carta recorded 966 startup shutdowns in 2024, up 25.6% from 769 the year before. AngelList tracked 364 winddowns in the same period, a 56% jump. The wave kept rolling through 2025 as the 2021 vintage of funded companies burned through its runway.
The sector breakdown is worth a look if you're building in software: enterprise SaaS led with 32% of shutdowns, followed by consumer (11%), health tech (9%), fintech (8%), and biotech (7%).
The cause wasn't mysterious. VCs didn't get better at picking winners in 2021. They just funded far more companies at the same hit rate, so the absolute number of failures had to rise a few years later. A lot of those companies raised on momentum and skipped the boring question of whether anyone would pay.
Which industries have the highest failure rates?
Tech has one of the worst survival curves of any industry. The BLS Information sector, which includes software, shows a 25.1% first-year failure rate and a 70.9% ten-year failure rate. Only mining and oil extraction fares worse over a decade.
Some context from the same dataset: agriculture businesses fail at just 49.5% over ten years, real estate at 57.8%, retail at 58.3%. The all-industry average is 65.3%. Software sits above it.
That surprises people. Software has near-zero marginal costs and infinite distribution, so shouldn't it be safer? No, and the reason is the same reason it's attractive: low barriers to entry mean crowded markets, fast-moving competition, and a flood of founders who can ship a product in weeks without ever checking if a market exists. Easy to build means easy to build the wrong thing.
Do first-time founders fail more often?
Yes, but by less than you'd expect. First-time founders succeed about 18% of the time, versus 30% for founders with a prior success. Founders whose previous startup failed succeed about 20% of the time, barely better than rookies.
Sit with that last number for a second. Failing once teaches you surprisingly little on its own. What separates the 30% group isn't scar tissue, it's process: successful repeat founders stop assuming demand and start testing it before they commit. That's learnable without the expensive first attempt.
This matches what the Startup Genome Project found years ago and what still holds up: 74% of failed startups scaled prematurely, spending on team and marketing before confirming product-market fit. And startups that pivoted once or twice showed 3.6x better user growth than those that never pivoted at all. Rigid founders who never update on evidence, and frantic founders who pivot constantly, both underperform the ones who test, learn, and adjust deliberately.
What can you actually do about these odds?
Attack the 43% problem before you spend money, because product-market fit is the one major failure cause you can address for free. You can't control macro timing. You can't will your unit economics into working after you've built the wrong product. But you can find out whether the problem you're solving is painful enough that people will pay, before you write code.
Practically, that looks like:
- Talk to 15-20 potential customers before building anything. Not friends. People who have the problem. Ask what they currently do about it and what it costs them. If they're not already spending time or money on the problem, it's probably not painful enough.
- Get a commitment, not a compliment. "Cool idea" predicts nothing. A pre-order, a signed letter of intent, a waitlist signup with a card on file, even a calendar hold for onboarding: those predict something. Compliments aren't demand.
- Size the market with a bottom-up model. Count reachable buyers and realistic pricing, not "1% of a $50B market." The 29% who died from bad timing mostly had top-down spreadsheets that said everything was fine.
- Write down your riskiest assumption and test it first. Most founders test the assumption that's easiest to confirm instead. That's how you end up 22 months from your last raise wondering where it went.
- Check your unit economics on paper before you scale anything. If the napkin math on acquisition cost and lifetime value doesn't work with generous assumptions, it won't work with real ones.
None of this requires money. It requires structure, which is the part most first-time founders lack. You can run this process in a spreadsheet or a Notion doc, or use a planning tool like Foundra that walks first-time founders through validation, competitive analysis, and financial modeling step by step. There are also free calculators for market sizing and runway math at foundra.ai/tools if you want to pressure-test your numbers without committing to anything.
The tool matters less than the discipline. The 18% of first-time founders who succeed mostly aren't smarter. They just checked before they built.
Key takeaways
- The "90% of startups fail" stat only applies to venture-scale startups. For all new US businesses, it's 20.4% in year one and 49.4% by year five (BLS).
- Running out of cash is the symptom in 70% of failures. The root cause, in 43% of cases, is poor product-market fit (CB Insights, 431 companies).
- Failed startups aren't underfunded: the median failed VC-backed company raised $11 million.
- The danger zone is years two through five, and 74% of shutdowns happen at pre-seed or seed stage (Carta).
- Software has one of the worst ten-year survival rates of any industry at 70.9% failure (BLS Information sector).
- First-time founders succeed 18% of the time vs 30% for previously successful founders. The difference is validation discipline, and it's learnable.
- 74% of failed startups scaled prematurely, before confirming product-market fit (Startup Genome).
FAQ
What percentage of startups fail in 2026?
It depends on the population. Per BLS data, 20.4% of all new US businesses fail within one year, 49.4% within five years, and 65.3% within ten. For venture-backed startups, Harvard research found 75% never return investor capital, and roughly 90% fail to hit venture-scale returns.
What is the number one reason startups fail?
Poor product-market fit. CB Insights found 43% of 431 failed VC-backed companies cited it as a root cause. Running out of capital appeared in 70% of failures but is classified as the final symptom, not the underlying cause.
Do most startups fail in the first year?
No. Only about 1 in 5 new businesses fails in year one. The highest-risk period is years two through five, after launch, when founders discover whether real demand exists for what they built.
Is the 90% startup failure rate true?
Only for a specific population: innovative, scalable startups chasing venture-scale returns. For ordinary new businesses, the ten-year failure rate is 65.3%. Quoting 90% for all startups mixes two different datasets.
How many startups shut down in 2024?
Carta recorded 966 shutdowns in 2024, up 25.6% from 769 in 2023. AngelList tracked 364 winddowns, up 56%. Enterprise SaaS accounted for 32% of Carta's recorded shutdowns.
Can validation really change your odds of failure?
The data points that way. The top root cause of failure (poor product-market fit, 43%) is addressable before you spend money, 74% of failed startups scaled before confirming fit, and repeat founders' higher success rate tracks with disciplined demand testing rather than raw experience.
Top comments (0)