The Startup Valley of Death: Why Companies Stall After Launch
Turn this article into takeaways for your work.
Each assistant summarizes the article only for you and suggests best practices for your work.
The valley of death is the stretch between a company's first funding or launch and the point where revenue covers costs. Spending runs ahead of income for months or years, cumulative cash falls, and the company has to survive on whatever it raised until the curve turns. Many young companies don't make it across.
This article explains where the term came from, what the cash-flow curve looks like, why the gap opens, the named models that describe it, and what survival data can and can't tell you. It's a concept piece. It doesn't tell any one founder what to do. For the broader picture of why young companies close, see why startups fail.
Two meanings of one phrase
"Valley of death" gets used for two related ideas, and mixing them up causes confusion.
The innovation-policy meaning. Here the valley sits between research and product. A university or government lab produces a promising result, but nobody funds the expensive, risky work of turning it into something a customer can buy. One of the clearest early uses in a policy setting appears in a 1999 National Research Council volume, where a speaker for US Representative Vernon Ehlers, describing his congressional science policy report, says that the gap between basic research and industry-driven applied research or product development is referred to as the "valley of death". The report in question, produced in 1997 and 1998 and approved by the House in October 1998, was a national science policy statement. That's a use of the phrase, not proof of who first said it. The term was already circulating in technology-transfer circles by then, and its exact origin isn't well documented.
The best-known academic treatment came a few years later. Lewis Branscomb and Philip Auerswald wrote a 2002 report for the US National Institute of Standards and Technology on funding for early-stage technology development, and followed it with a 2003 paper, Valleys of Death and Darwinian Seas, in the Journal of Technology Transfer. Their argument is that asymmetries of information and motivation, along with institutional gaps, can systematically deter private investment in early-stage technology development.
The startup-finance meaning. In startup talk, the valley is a cash-flow shape, not a funding gap between institutions. A company has launched, or raised its first money, and is spending to find customers, but revenue is still small. The two meanings overlap. A deep-tech company coming out of a lab can face both at once. But a software startup with a finished product can sit in the finance valley without ever touching the research gap.
The rest of this article is mostly about the second meaning.
The J-curve of cumulative cash
Plot a young company's cumulative cash position over time. It starts at zero or at the amount raised. Costs begin immediately: salaries, tools, infrastructure, marketing, legal. Revenue arrives later and smaller. So the line dips.
Eventually, if the business works, monthly revenue overtakes monthly spending, the line bottoms out and climbs. The shape is a J: a long slide down, a low point, then a rise that eventually passes the starting level. The low point is the deepest cumulative cash deficit the company reaches.
Three things define the valley on that curve.
- Depth. How much cash the company consumes before it turns. This is set by burn rate and by how long the dip lasts.
- Length. How many months pass between the first dollar spent and the first month in which the business pays for itself.
- Funding on hand. Whether the company's cash covers the full depth of the dip. This is the line between a valley that is crossed and one that ends the company.
A useful way to say it: the valley isn't dangerous because cash goes down. It's dangerous when runway ends before the curve turns. A company that raised enough to reach the bottom and climb out was never in mortal danger. A company that raised the same amount but found the bottom was twice as deep as planned was.
The "J-curve" label is also used in private equity and venture capital for the returns of a fund, which show negative early returns from fees and write-offs before gains arrive. That's a related shape for a different thing (an investor's fund, not a single company's cash), so check which one a source means.
Why the valley opens
The gap between spending and income isn't a failure of management by itself. Most of the reasons are structural.
Costs come first, revenue comes later. A company has to hire, build and launch before it can sell. Even after launch, early customers take time to find, trial and convert. With business customers, sales cycles of several months mean today's spend produces revenue next quarter or later.
Early revenue is small and uneven. The first customers are often discounted, on pilots, or on short terms. Revenue that exists may not yet be repeatable, so it can't be forecast with confidence.
The product isn't finished learning. Many startups launch before they know which customers value the product most. Finding that out takes iteration, and iteration costs months. This is why the valley overlaps with the search for product-market fit. Until the company has fit, adding spend tends to stretch the dip instead of shortening it.
Funding arrives in rounds, not continuously. Cash comes in lumps, and between lumps the company has to prove enough to raise the next one. The gap between rounds is where many valleys become fatal, because investors price a company on its trajectory, and a flat or unclear trajectory doesn't clear the bar.
Expectations shift between stages. A seed investor may back a team and a thesis. A Series A investor typically wants evidence: customers, retention, a repeatable way to acquire more. The change in what counts as proof is itself a hurdle. The seed stage is where that first proof gets built, and Series A, B and C funding covers how the rounds differ.
The Series A crunch
The best-known version of the funding gap is the so-called Series A crunch: companies that raised a seed round find it hard to raise the next one.
Crunchbase, a venture data provider, tracks this for US startups that raised seed rounds of $1 million or more. In a May 2026 analysis by Gené Teare, Crunchbase reported that of the companies that raised a $1 million-plus seed round in 2023, only 24% had progressed further. The same piece says US startups have been taking longer to raise a Series A after a seed round of that size, with the gap now stretching to more than two years.
Three cautions apply. This is a population of venture-backed US startups with seed rounds of $1 million or more, not startups in general. "Progressed further" is Crunchbase's measure of later-stage funding, which isn't the same as the company being healthy or unhealthy. And recent cohorts haven't had much time yet, so their numbers will change as more companies reach their next round.
What the data supports is the mechanism, not a precise rate: the longer the wait between rounds, the more of the valley a company has to fund from its existing cash, and the more a burn rate set for a faster timeline becomes a risk. A bridge round is one stopgap, and a down round is one cost of raising in a weaker position.
Named models that describe the same stretch
Several well-known frameworks describe parts of the valley from different angles. They aren't competing; they look at different parts of the same problem.
| Model | Source | What it describes | Population |
|---|---|---|---|
| Valley of death (policy) | Technology-transfer literature, discussed in a 1999 National Research Council volume | The gap between basic research and industry-driven product development | Research institutions and technology commercialization |
| Valleys of Death and Darwinian Seas | Auerswald and Branscomb, 2003 | Early-stage technology funding gaps, and the later market selection that follows | US technology-based innovation |
| Trough of sorrow | Drawn by Paul Graham and other Y Combinator partners, as described by Andrew Chen | The middle phase after the initial excitement fades and before product-market fit | Venture-backed startups |
| Chasm | Geoffrey Moore | The gap between early adopters and the mainstream market | Technology product adoption |
| Greiner's crises | Larry Greiner | The crises that force a growing company to change how it is managed | Growing organizations |
Notice what differs. The policy model is about who pays for risky early work. The trough of sorrow is about morale and metrics in the middle. The chasm is about customers. Only the cash-flow view, the J-curve, measures the thing that actually ends a company: money.
What the survival data says
The cleanest public data isn't about startups specifically. The US Bureau of Labor Statistics publishes survival of private sector establishments by opening year, covering every new private-sector establishment, from restaurants to software offices. An establishment is a single physical location, not a company, so a closure can mean a business failed, moved, merged or was sold.
For the cohort of establishments that opened in the year ended March 2013, the table gives these figures.
| Years after opening | Share of the original cohort still operating | Share of the previous year's survivors still operating |
|---|---|---|
| 1 | 79.6% | 79.6% |
| 2 | 68.9% | 86.6% |
| 3 | 61.4% | 89.0% |
| 4 | 55.3% | 90.1% |
| 5 | 50.6% | 91.5% |
| 10 | 34.7% | 92.2% (year 10 only) |
The most recent cohort with a full first year, the year ended March 2024, shows 77.9% still operating after one year. The latest data point in the table is March 2025, when the 2013 cohort's survival rate since birth was 30.4%.
Two readings matter here.
First, the steepest drop is in year one. About one in five establishments in the 2013 cohort was gone within twelve months. That's consistent with the idea that the earliest stretch is the riskiest, though the data can't tell you why any individual establishment closed.
Second, the annual loss rate falls over time. Survivors from earlier years are more likely to make it through the next one: the share rises from 79.6% in year one to 91.5% in year five. In plain terms, the odds improve the longer a business has been operating. That fits the shape of the valley, where the hard part is early and the risk eases once revenue is established. But it doesn't prove the cause. Older businesses differ from newer ones in many ways, and survivors are by definition the ones that weren't filtered out.
The population caveat is the important one. These are establishments, not firms, and not venture-backed startups. A new venture-backed software company, a new restaurant and a new branch office of an existing company are all in the data. Use it as a baseline for how hard it is to keep any new business running, not as the odds for a technology startup. The Crunchbase graduation data above covers a narrower population (US startups with $1 million-plus seed rounds), and the two series answer different questions. One is about staying open. The other is about raising the next round.
How runway and burn rate frame the valley
If the valley is a curve, runway and burn rate are how a founder reads where on the curve the company stands.
Burn rate is the net cash the company loses per month. Runway is cash on hand divided by that burn, expressed in months. Together they give a deadline: the date cash reaches zero if nothing changes.
The valley adds a second question. How many months until the business pays for itself? If runway is 18 months and the company expects to break even in 24, the plan has a six-month hole. There are only a few ways to close a hole like that, and each one trades something off:
- Raise more money, which dilutes ownership and depends on investors saying yes.
- Cut burn, which slows the work that was supposed to produce the revenue.
- Raise revenue faster, which depends on a market response nobody controls.
- Shorten the path by narrowing the product or the customer to what's closest to paying.
None of these is free, and none is guaranteed. What the framing does is make the trade-off visible early, when there are more options, instead of late, when there's one.
A few signals tend to show where a company sits on the curve. Retention that holds suggests the dip may be bottoming out. A sales cycle that keeps lengthening suggests the opposite. If the team can't say its break-even month, it doesn't yet have a plan for the valley.
Where the valley sits in the stage model
The valley doesn't map to one stage. It begins after the first outside funding and runs through whatever stage the company is in when revenue finally catches up. For many software companies that means the seed stage and the early part of the next one. The stages of a startup run in that sequence from idea to exit. Companies that cross the valley often move into a scale-up phase.
Key Facts: The Startup Valley of Death
- The valley of death has two meanings: a research-to-product funding gap in innovation policy, and a stretch of negative cumulative cash flow between launch and sustainable revenue in startup finance.
- A 1999 National Research Council volume describes the gap between basic research and industry-driven product development as the "valley of death", in a speech about Representative Vernon Ehlers' science policy report.
- Auerswald and Branscomb's 2003 paper, Valleys of Death and Darwinian Seas, argues that information and motivation asymmetries can deter private investment in early-stage technology.
- The J-curve of cumulative cash dips as spending outpaces revenue, bottoms out, then rises if the business works. The valley is fatal when runway ends before the curve turns.
- Among US private-sector establishments that opened in the year ended March 2013, 50.6% were still operating after five years and 34.7% after ten. These are establishments, not startups.
- Crunchbase found that only 24% of US startups that raised a $1 million-plus seed round in 2023 had progressed further, and that the wait for a Series A now exceeds two years.
Related reading
