Founders are the risk-takers and visionaries who drive our economy forward. They take bold bets, build teams, and pursue ideas that could reshape entire industries.
But the reality is that not all startups make it. The statistics remain stark: 93% of startups ultimately shut down. What has changed dramatically is how and when they wind down.
This was not a year of collapse. It was a year of maturation.
The companies shutting down now are not early experiments. They are the mid-stage, post-ZIRP (zero interest-rate policy) generation that raised capital, built product, hired teams, and still found the model could not sustain a next round. The correction has moved from "failed ideas" to "failed models."
Here's the truth: Founders don't shut down because they're done building. They shut down to clear the path for what comes next.
At SimpleClosure, our mission is to remove unnecessary friction from that process. This year, we analyzed hundreds of structured shutdowns to understand what's really happening in the startup ecosystem.
Let's dig in.
Companies Are Shutting Down Later in Their Lifecycle
One of the clearest year-over-year shifts is in stage composition. In 2024, most shutdowns were early attempts: pre-seed and seed companies that never fully broke out. In 2025, the mix has moved meaningfully up-market.
The biggest shift is at Series A, which jumped from ~6% to ~14% of all shutdowns. A 2.5x increase. More companies that made it through early validation, secured institutional investment, and built real product are now reaching the end of their runway.
Pre-seed and seed make up a smaller share of the total. This doesn't mean fewer early companies are failing; it means the overall mix has shifted upward as more mature venture-backed teams hit their limit.
Companies Shutting Down in 2025 Have Raised More Capital (and Are Older)
While early-stage companies continue to churn, the companies shutting down later in the funnel have raised more capital and been operating for far longer. Companies shutting down in 2025 are not new experiments. Many are 7 to 10 years old, dating back to fintech, insurtech, mobility, logistics, and marketplace waves founded before the AI boom.
The youngest shutdowns, pre-seed and seed companies averaging three to five years old, mostly stem from the 2021 to 2022 formation wave: lightweight SaaS, early automation tools, creator platforms, fintech experiments, and first-generation AI wrappers. Fast-formation companies built in an environment of cheap capital and rapid experimentation.
The companies shutting down at Series A and beyond are notably older. Series A failures average around 7 years old and were largely founded in 2017 to 2019, when fintech, proptech, logistics, HR tech, and B2B SaaS were dominant. Series B and Series C+ shutdowns are closer to a decade old, reflecting the 2014 to 2017 wave of marketplace models, on-demand services, and operational platforms that raised heavily, survived multiple pivots or bridge rounds, and are only now reaching the end of their runway.
The Geography of Shutdowns
The geography of 2025 shutdowns mirrors where the ZIRP-era startup boom was most concentrated. The states that saw the greatest influx of venture-backed companies between 2020 and 2022 are now seeing the largest share of closures.
California and New York dominate because they generated the largest numbers of venture-backed companies in the prior cycle. The sharp upward movement in Pennsylvania and Colorado signals local ecosystems whose 2020 to 2021 vintages are now confronting the same funding constraints seen nationally. States like Massachusetts, Florida, and Illinois show moderate declines, consistent with more stable formation patterns.
The correction is heaviest in the same places where startup formation was heaviest during ZIRP.
Industry Mix Is Normalizing
The industry mix of 2025 shutdowns looks less like a sector-specific collapse and more like a normalization after the ZIRP-era boom. AI remains one of the largest categories, but a broader set of sectors is now appearing as their boom-year vintages reach the "hard decisions" stage.
AI stays on top, but others are catching up. AI's share edges down from 17.7% to 15.9% as other sectors start to show more closures. This isn't an AI collapse; it's other "boom" sectors finally hitting their own reality checks.
B2B SaaS and developer tools are seeing more late-cycle shutdowns. B2B SaaS rises from 5.2% to 7.7%. Developer tools inch up from 6.0% to 6.4%. These are often 2020 to 2022 vintage companies that got to real product and some revenue, but not to the efficiency or retention needed for today's funding bar.
Fintech and CPG are cooling as a share of closures. Fintech falls from 5.6% to 4.3%. CPG drops from 8.5% to 6.1%. That doesn't mean these categories are "healthy" so much as they've already absorbed an earlier wave of shutdowns.
Harder, capital-intensive sectors are now showing up more. Biotech (including diagnostics) jumps from 1.2% to 3.6%. Healthcare climbs from 0.8% to 2.8%. Climate & Energy rises from 2.0% to 2.8%. These sectors involve long R&D cycles, regulatory drag, and significant capital needs. Their larger share of shutdowns suggests the 2020 to 2021 enthusiasm for deep tech is meeting the current fundraising environment.
The broader theme: 2025 isn't about a single sector collapsing. It's about multiple 2020 to 2022 "boom" cohorts (AI, SaaS, climate, biotech) hitting their respective filter moments at roughly the same time.
The First Wave of AI Shutdowns Has Arrived
Given AI's prominence in both formation and shutdown data, we took a deeper look at what's actually failing and why.
"Wrappers & Apps" include copilots and assistants layered on foundation models, AI-powered productivity tools, content and design generators, and vertical SaaS products where the core differentiation is an LLM front-end rather than proprietary data or infrastructure.
These companies found early usage but struggled to translate that into durable economics because margins compressed as API/model costs shifted, switching costs for customers were low, and differentiation eroded as the same capabilities became widely available.
Infra and dev-tool companies fail less often, but when they do, they've raised more
Infrastructure and dev-tool AI companies (LLM hosting platforms, evaluation tooling, MLOps/data infra, security/governance layers) are a smaller share of closures, but the ones that shut down have generally raised more capital.
These companies sell into more complex enterprise environments, have longer integration cycles, and require more up-front capital to build infrastructure, tooling, and go-to-market motion.
AI shutdowns are spread across stages, not just pre-seed experiments
AI closures aren't confined to ultra-early companies. AI companies represent roughly 12 to 20% of shutdowns within most stages, from pre-seed through Series B. The median AI company that shuts down has raised around $2.4M, versus $2.8M for the overall dataset.
These aren't just weekend projects. Many AI shutdowns raised institutional seed rounds, survived at least one iteration cycle, built real product, and still could not find enough differentiation or efficiency to justify another round.
What this first wave tells us
Putting it together:
AI is not over-represented in shutdowns, but it is over-represented in a certain type of shutdown. The dominant pattern is AI wrappers and application-layer tools built quickly on top of commoditized models, without deep defensive moats.
AI infrastructure/dev tools are harder to kill, but the stakes are higher. Fewer infra companies shut down, but those that do have raised roughly 2x the capital of wrapper/app peers, and often face direct competition from hyperscalers and cloud-native offerings.
The AI ecosystem is moving from novelty to selection. The 2023 to 2024 cycle rewarded speed and UX, leading to a long tail of thin GPT-wrapper products that raised early capital simply by being first to market. But the 2025 shutdown data shows that the market has shifted and is now filtering aggressively for companies with proprietary data advantage (not just API access), real unit economics (not usage arbitrage), and deep integration into enterprise workflows, not tools that simply sit on top of them.
From SimpleClosure's perspective, AI shutdowns look less like the end of a fad and more like the first real selection event in a still-young market. The founders winding down these companies are often the same ones who will found the next generation of AI-native startups, with clearer differentiation, tighter burn, and a more realistic view of where value actually accrues in the stack.
Conclusion
The 2025 shutdown data tells a consistent story: the market is maturing, not collapsing. Companies are shutting down later, older, with more capital raised. The 2021 to 2022 vintage is working through its filter moment.
Founders don't shut down because they're done building. They shut down to clear the path for what comes next.
SimpleClosure partners with leading VCs and law firms to reduce the complexity, cost, and time of startup dissolution. Our expert team handles everything from asset distribution to regulatory filings, helping you:
Save 85% on shutdown costs
Start dissolving in under 48 hours
Return remaining capital to investors efficiently
Maintain compliance throughout the process
We want to make shutting down a company as easy as starting one. Join the growing network of VCs and law firms using SimpleClosure. Contact partners@simpleclosure.com to learn more about our partnership program.
Get your free founder's guide to shutting down
A comprehensive breakdown of how founders can quickly and easily execute a fully compliant shutdown
