The Startup Graveyard: What 2,293 Dead Startups Reveal About How Companies Actually Die

Startups die quietly. A press release rarely goes out. The website 404s, the Slack goes dark, and a few months later a founder writes a thoughtful post-mortem that gets 400 upvotes and then disappears into the archive. Individually, each shutdown is a private tragedy. In aggregate, they are the most honest dataset in venture capital.

We built one. Our Startup Graveyard database now tracks 2,293 venture-backed companies that have closed their doors, with the real funding data behind each one: rounds raised, lead investors, founders, and last known financing. Together these companies raised and lost $147.2 billion. That is not a rounding error in the venture economy. That is a decade of conviction, marked to zero.

This report is what the data says when you stop looking at any single obituary and start looking at all of them at once. The patterns are sharper than the folklore. Startups do not mostly die as flameouts of overfunded unicorns. They die small, they die young, and lately they die in one very specific vintage.

The typical dead startup dies small. The money dies big.

Here is the single stat that reframes everything. Across the 2,293 closed companies, the average amount raised before shutting down was $64.2 million. The median was $4.95 million.

That gap is not noise. It is the power law running in reverse. Venture returns are famously concentrated: a handful of winners generate almost all the gains, and the average outcome is dragged upward by a few enormous exits. Failure obeys the same math in the opposite direction. Most companies that die never raised much, so the typical corpse is a sub-$5 million seed-stage company. But a small cluster of catastrophic, nine- and ten-figure failures drags the average up by a factor of thirteen.

The concentration is stark. Just 24 companies raised more than $1 billion before closing, our billion-dollar club. 227 raised over $100 million. 850 crossed $10 million. Everyone else, more than 1,400 companies, is in the long tail below that, the seed rounds and bridge notes that quietly ran out of runway. When people say "startups fail," they picture Quibi. The data says they should picture a two-person company that raised $3 million, hired a few engineers, and could not find product-market fit before the money ran out.

For founders, this is the honest baseline. The modal failure is not a governance scandal or a fraud. It is an ordinary company that ran the experiment, got a "no" from the market, and closed responsibly.

Almost half never made it past seed

Sort the graveyard by the last round each company raised before dying, and the shape of startup mortality becomes obvious.

• Seed: 729 companies
• Pre-Seed: 307
• Venture (round unknown): 278
• Series A: 263
• Series B: 119
• Debt Financing: 93
• Grant: 83
• Post-IPO Equity: 70
• Convertible Note: 53

Add seed and pre-seed together and you get 1,036 companies, roughly 45 percent of the entire graveyard, that never raised a priced round beyond seed. The Series A is where the funnel narrows hard: only 263 companies made it to a labeled Series A before dying, and just 119 to a Series B. This is the quantitative shape of the famous "Series A crunch." Getting seed funding is a permission slip to try. It is not evidence that the business works.

There is a subtle warning in the debt and post-IPO buckets, too. 93 companies took on debt financing as their last raise, and 70 were public via post-IPO equity. These are later-stage deaths, companies that had graduated past the "will anyone want this" question and still failed on unit economics, timing, or capital structure. Survival is not monotonic. Raising more does not remove the risk of dying. It just changes the cause of death.

The 2022 cohort is the ZIRP hangover

If you want to understand why the graveyard is filling up right now, look at when these companies last raised money.

• 2021: 559 companies, $29.8B raised
• 2022: 925 companies, $51.1B raised
• 2023: 504 companies, $24.3B raised
• 2024: 236 companies, $32.9B raised
• 2025: 65 companies, $9.0B raised

The 2022 vintage is not just the largest cohort in the graveyard. It is the largest by a wide margin, both in company count and in dollars incinerated. This is the zero-interest-rate-policy hangover, made visible.

The story is well known by now. Through 2021 and into early 2022, money was effectively free, valuations detached from fundamentals, and rounds closed in days on the strength of a deck. Companies raised 18 to 24 months of runway at prices that assumed the good times would continue. Then rates rose, the growth-at-all-costs playbook stopped working, and the next round never materialized at a price anyone would accept. Do the arithmetic on 24 months of runway from a 2022 raise and you land squarely in late 2023 and 2024, which is exactly when the deaths cluster.

The 2024 and 2025 numbers look smaller, but do not read them as a recovery. They are a reporting lag. A company that raised in 2024 may still be burning through runway today, and its shutdown will not enter a graveyard until it actually closes. The most recent vintages always look artificially healthy because the deaths have not happened yet. That is precisely why a live database beats a one-time snapshot: the 2024 cohort is going to get worse, and a static list published in 2024 will never tell you.

The AI paradox

Sort the graveyard by industry and you get a list that doubles as a map of the last decade's hype cycles.

• Software: 622 companies
• Artificial Intelligence: 355
• Information Technology: 351
• Health Care: 256
• Financial Services: 235
• FinTech: 220
• SaaS: 207
• Machine Learning: 185
• Biotechnology: 162
• Blockchain: 159
• E-Commerce: 154
• Analytics: 131

Note that companies are tagged with multiple industries, so these buckets overlap. But one pattern jumps out and refuses to leave. Artificial Intelligence is the number two failure category, with 355 dead companies, and Machine Learning adds another 185. AI is simultaneously the hottest funding sector on the planet and one of the largest graveyards we track.

That is the AI paradox, and it is not a contradiction. It is what a gold rush looks like from the inside. Capital floods a category, thousands of companies form to chase it, and the base rate of failure applies to all of them regardless of how exciting the category is. A hot sector does not lower your odds of dying. It raises the number of companies born into the same fixed failure rate, which means more absolute deaths, not fewer. The founders raising AI seed rounds today are drawing from the same distribution that produced 355 AI headstones. Blockchain, with 159 dead companies, already ran this exact movie a cycle earlier.

The lesson is not "avoid AI." The lesson is that being in a hot category is not a moat, a business model, or a substitute for demand. It just gets you funded faster, which, if the underlying company does not work, only means you reach the graveyard with more money attached.

The billion-dollar club: the biggest failures we track

The 24 companies that raised over a billion dollars before dying are where the $147 billion in losses concentrates. These are not scrappy seed-stage experiments. They are category bets from the most sophisticated investors in the world, and they still went to zero. Here are the ten largest by total capital raised.

Company	HQ	Raised	Sector	What happened
Cruise	San Francisco, CA	$16.0B	Autonomous vehicles	GM pulled funding and wound down the robotaxi unit
OneWeb	London, UK	$4.7B	Satellite internet	Filed for bankruptcy; rescued by the UK government and Bharti
Argo AI	Pittsburgh, PA	$3.6B	Self-driving	Ford and VW withdrew backing and shut it down
Britishvolt	Blyth, UK	$2.4B	EV batteries	Collapsed before its gigafactory was built
Hipgnosis Songs Fund	London, UK	$2.2B	Music royalties	Wound down after valuation and governance disputes
Marlette Funding	Wilmington, DE	$2.1B	Fintech lending	Consumer lending platform ceased operations
EQRx	Cambridge, MA	$1.9B	Biotech	Its "affordable drugs" thesis unwound before launch
Lacework	San Jose, CA	$1.9B	Cloud security	Once valued at $8.3B, sold in distress
FTX	Nassau, Bahamas	$1.8B	Crypto exchange	Collapsed in fraud; founder convicted
Fisker	Manhattan Beach, CA	$1.8B	Electric vehicles	Henrik Fisker's second EV company to go bankrupt

Look at what dominates this list: capital-intensive, hardware-heavy, physics-bound bets. Autonomous vehicles account for three of the top three. Batteries, satellites, and electric cars fill out most of the rest. The pattern is that the biggest failures tend to be businesses where software-style iteration does not apply, where you cannot ship a minimum viable robotaxi, and where the gap between "impressive demo" and "profitable at scale" costs billions to cross. Cruise alone raised more than the entire 2025 death vintage combined. The one clean outlier is FTX, which did not run out of money so much as run out of other people's.

Why we built a live graveyard, not another static list

Curated failure lists are having a moment, and rightly so. They are genuinely useful to idea-stage founders who want to see who tried their idea before and why it did not work. But most of them share the same limitation: they are hand-assembled snapshots. Someone curated a few hundred or a couple thousand companies once, wrote a sentence about each, and moved on. The list is frozen the day it ships.

Failure is not frozen. Companies die every week, the 2024 vintage is still being written, and yesterday's "well-funded startup" is tomorrow's headstone. So we built the Startup Graveyard on top of our live funding database instead of as a one-off content project. Every entry carries the real numbers: total raised, round history, named lead investors, and founders, not just a vibe and a cause of death. Because it is tied to a live dataset, it updates as the graveyard grows, and you can filter it the way you actually think, by sector, by stage, by amount raised, by investor.

That last point matters more than it sounds. If you are a founder, you can pull every dead company in your exact category and read the pattern before you pitch. If you are an investor, you can see which of your peers led the rounds that did not make it. If you are a recruiter or a job seeker, you can check whether a company's last raise was a healthy Series B or a quiet bridge note that suggests trouble ahead. The same dataset that powers the graveyard powers our full funding intelligence platform, tracking live rounds, investors, and founders across the market, not only the companies that failed.

If you want that full picture, the living companies as well as the dead ones, create a free VCbacked account at app.vcbacked.co/signup and search the entire funding database. The graveyard is the cautionary half. The other half is who is winning right now.

Frequently Asked Questions

What is a startup graveyard?

A startup graveyard is a database or curated list of companies that have shut down, usually venture-backed ones, collected so founders and investors can study why they failed. The best versions include real data on how much each company raised, who funded it, and what killed it. Our Startup Graveyard tracks 2,293 defunct venture-backed startups that collectively raised and lost $147.2 billion, with live funding data behind every entry.

What percentage of startups fail?

The widely cited figure is that roughly 75 percent of venture-backed startups fail to return investor capital, and around 90 percent of all startups fail over the long run. Our data reinforces why the venture number is so high: almost half of the companies in our graveyard never raised a round beyond seed, and the failure rate applies regardless of how hot the sector is. Even well-capitalized companies die. Reaching a Series B is not a guarantee of survival.

What is the biggest startup failure ever?

By total capital raised before shutting down, the largest failure in our database is Cruise, the autonomous vehicle company, which raised roughly $16 billion before GM ended funding for its robotaxi unit. It is followed by OneWeb ($4.7B, satellite internet) and Argo AI ($3.6B, self-driving). FTX is the most notorious, having collapsed in fraud rather than by running out of runway, but it raised less than the capital-intensive hardware bets above it.

Why do most startups fail?

Most startups fail for ordinary reasons rather than dramatic ones. The single most common cause is a lack of market demand, followed by running out of cash, being outcompeted, weak unit economics, and mistimed markets. Our vintage data shows a macro cause too: the 2022 ZIRP cohort raised at inflated valuations on cheap money, then could not raise the next round when rates rose. The typical dead startup in our data raised only $4.95 million, meaning it ran the experiment, did not find product-market fit, and closed before the money became large.

How do I research failed startups in my market?

Start by filtering a failure database to your specific sector and stage rather than reading a generic list. In the Startup Graveyard you can narrow to your industry, see every venture-backed company in it that has died, and read the amount raised, investors, and last round for each. That tells you which approaches have already been tried and funded and still failed, which is exactly the context you want before pitching or investing. For live competitive research on companies that are still operating, you can search the full funding database with a free VCbacked account.