Every year, the startup graveyard gets new headstones. Some are quiet acqui-hires buried in press releases. Others are spectacular implosions that make you wonder how anyone thought this was a good idea in the first place.
2025 has been particularly brutal. Startup shutdowns have increased significantly, with 966 U.S.-based startups closing in 2024 compared to 769 in 2023 - a 25.6% increase TechCrunch - and that trend is continuing into 2025.
The easy money era is over, AI hype is colliding with reality, and investors are finally asking uncomfortable questions like "but how do you make money?"
As we approach the end of 2025, let's examine the startups that didn't make it and extract some hard-won lessons from the wreckage.
1. Builder.ai: The "AI" That Was Actually Just Humans
The Promise: Build custom apps using AI, as easy as ordering pizza. Revolutionary no-code platform backed by Microsoft and Qatar Investment Authority.
The Reality: Builder.ai filed for bankruptcy in May 2025 after burning through $445 million, with revelations that much of the "AI-powered" development was actually performed by hundreds of offshore human developers Rest of WorldRest of World.
What Went Wrong: This is the startup equivalent of someone wearing a fake mustache and calling it "AI innovation." The company had been operating without a CFO since July 2023 and inflated sales figures by massive amounts - internal audits slashed 2023-2024 projections by 75% CTOL Digital Solutions. When lenders seized $37 million of the company's $42 million in cash DEV Community, it was game over.
Founder Sachin Dev Duggal styled himself as "chief wizard," which in retrospect should have been a red flag. The company raised at a $1.3 billion valuation, but 2023 revenues were revised down to just $140 million CTOL Digital Solutions, and the business model never actually worked.
The Lesson: You can't fake it till you make it when "it" requires actual technology you don't have. Also, financial governance matters - maybe hire a CFO?
2. Pandion: When the Pandemic Boom Goes Bust
The Promise: Revolutionize package delivery with a tech-enabled logistics network combining sortation centers and a million-driver gig economy fleet.
The Reality: Pandion shut down abruptly in January 2025 after raising $125 million, informing its 63 employees they were being let go with no severance, paid only through January 15 GeekWire.
What Went Wrong: Pandion was founded in 2020 by Scott Ruffin, former head of Amazon Air, during the pandemic e-commerce explosion. Perfect timing, right? Except the company launched just as a wider freight recession began in mid-2022, facing thin margins in e-commerce fulfillment, tough pricing competition, and a vanished venture capital funding environment Sourcing Journal.
Despite raising $41.5 million in Series B funding in March 2024 with projected sales of $220 million Sourcing Journal, the company couldn't make it past Q4 2024. Talks with potential acquirers dragged through the holidays but ultimately went nowhere.
The Lesson: Market timing is everything. Building during a boom is easy; surviving the inevitable correction is what separates real businesses from opportunistic bets. Also, competing directly with UPS, FedEx, and Amazon when you're burning cash is... ambitious.
3. Lilium: The Electric Dream That Couldn't Fly
The Promise: Electric vertical takeoff and landing (eVTOL) jets for commercial air travel. A team of ex-Airbus, Boeing, and NASA engineers building the future of transportation.
The Reality: Lilium declared insolvency in 2025 after years of struggle, unable to achieve satisfactory performance or obtain necessary regulatory approvals for flight, and failing to raise the investment needed to continue Tech Funding News.
What Went Wrong: Building aircraft is hard. Building electric aircraft that take off vertically is harder. Getting them certified to carry passengers? Nearly impossible without infinite funding. Despite raising over $1 billion and partnerships with Hyundai Tech Funding News, the company couldn't overcome the twin challenges of technical performance and regulatory hurdles.
The aviation industry is littered with promising eVTOL companies that discovered physics, regulators, and certification timelines don't care about your pitch deck.
The Lesson: Some markets have such high barriers to entry that even a billion dollars and brilliant engineers aren't enough. Sometimes the tech just isn't ready, no matter how badly we want flying taxis.
4. Canoo: The EV That Couldn't
The Promise: Modular electric vehicles for fleets and ride-sharing, with distinctive futuristic design and skateboard platform technology.
The Reality: The California-based EV startup filed for bankruptcy in January 2025 despite raising over $1 billion and achieving a multi-billion dollar valuation Tech Funding News.
What Went Wrong: Disappointing earnings projections in 2024 marked the beginning of the end Tech Funding News, followed by layoffs that couldn't stem the bleeding. Canoo joins a graveyard of EV startups that discovered building cars at scale requires more than slick renderings and pre-orders.
The EV market has proven to be a bloodbath for startups. Tesla survived by the skin of its teeth; almost everyone else is finding out that competing with established automakers with deep pockets and decades of manufacturing expertise is nearly impossible.
The Lesson: Hardware is hard. Automotive hardware where you compete with Ford, GM, and Tesla while also fighting every other well-funded EV startup? That's hell mode.
5. EasyKnock: The Landlord Nobody Wanted
The Promise: Help cash-strapped homeowners unlock equity through sale-leaseback deals - sell your house to us, stay as a tenant, buy it back later.
The Reality: EasyKnock abruptly shut down in December 2024 after facing over two dozen lawsuits and investigations by multiple state attorneys general for alleged deceptive business practices RealEstateNews.com.
What Went Wrong: This is what happens when your business model is "legally questionable" meets "ethically dubious." Homeowners often received far less equity than expected when properties sold, many couldn't afford to buy their homes back, and some were evicted from houses they once owned NPR.
EasyKnock reached a $200,000 settlement with Massachusetts in 2023 and faced cease-and-desist orders from Michigan, with Connecticut's AG filing suit in November 2024 RealEstateNews.comInman. The company raised $455 million, but when your customers are suing you and regulators are investigating you, that's not a sustainable business.
The Lesson: If your business model requires evading consumer protection regulations and your customers keep suing you, you don't have a business - you have a lawsuit factory. Also, targeting vulnerable populations is bad business and bad karma.
6. Rain AI: The Chip Dream That Melted
The Promise: Build revolutionary AI chips that would "power the future of AI," with backing from Sam Altman and deals with OpenAI.
The Reality: Rain AI struggled with difficulties in prototyping and fabrication that dampened investor enthusiasm, with leadership changes failing to turn around performance, now surviving on bridge financing while seeking a buyer Tech Funding News.
What Went Wrong: Hardware startups face a brutal reality: you can't iterate your way to success like software. Despite high-profile clients and personal investment from Sam Altman Tech Funding News, actually manufacturing chips that work at scale is extraordinarily difficult and capital-intensive.
The company is technically still alive, but in startup terms, "limping along on bridge financing" is the walking dead. They're now touting their technology IP rather than their chip-making capabilities - never a good sign.
The Lesson: Semiconductor manufacturing is one of the hardest things humans do. Having a famous investor and ambitious promises doesn't overcome the physics and economics of chip production.
7. Spotlight Therapeutics: When Cutting-Edge Isn't Enough
The Promise: CRISPR-based gene-editing therapies for ophthalmology, a UC Berkeley and UCSF spinout backed by Google.
The Reality: The healthtech startup focused on delivering CRISPR gene-editing therapies shut down after initial results were underwhelming, despite Google funding Tech Funding News.
What Went Wrong: Gene therapy is the future - just not yet, apparently. Even with prestigious academic roots, Google's checkbook, and genuine unmet medical need in ophthalmology, the science just wasn't ready for commercialization.
Biotech is brutal: 9 out of 10 drug candidates fail, timelines stretch for decades, and regulatory approval is a marathon. Having cutting-edge technology doesn't guarantee it'll work in actual human patients.
The Lesson: Deep tech is really deep. Sometimes the science needs another decade of research before it's ready for a commercial company. Being early is indistinguishable from being wrong.
8-10. The Honorable Mentions of Horror
Plenty Unlimited (AgTech)
The indoor vertical farming company filed for bankruptcy in 2025 despite promising increased yields and pesticide-free products to help feed a growing global population Tech Funding News. However, Plenty has since restructured and is now focusing on producing strawberries Tech Funding News - proof that failure doesn't always mean the end.
The Lesson: Vertical farming sounds great until you realize growing lettuce indoors costs more than traditional farming, and nobody wants to pay $12 for a head of lettuce.
Cushion (FinTech)
The fintech app that helped consumers negotiate bank fees shut down in late January 2025, despite raising $21.6 million and reaching an $82.4 million valuation by 2022 Jumpstart Magazine. Turns out, taking a commission on bank fee refunds wasn't a big enough market.
The Lesson: Being right about a problem (bank fees suck) doesn't mean there's a big enough market to build a venture-scale business.
The Unnamed AI Coding Startups
Several AI coding startups have discovered there isn't enough demand to support the AI hype, with Builder.ai being the most prominent example Tech Funding News. When ChatGPT can do 80% of what your product does for $20/month, your margin for differentiation is razor-thin.
The Lesson: The AI gold rush created a lot of picks-and-shovels businesses, but when the picks and shovels are commoditized by OpenAI, Anthropic, and Google, you need a really good moat.
The Macro Picture: Why 2025 is So Brutal
This isn't just bad luck. The surge in failures stems from the massive funding boom of 2020-2021, when VCs funded companies at heated valuations with thin due diligence TechCrunch. Those companies had 2-3 years of runway. Now? The bill is due.
Companies funded at unreasonably high valuations face increased risk because investors won't invest more unless business growth is extremely strong TechCrunch. When you raise at a $1 billion valuation but only have $50 million in revenue, your next round requires massive growth that most companies simply can't deliver.
Add to this:
- Interest rates that killed cheap debt
- Venture capital that dried up in 2023-2024
- Return to normalcy after pandemic boom times
- AI hype creating unrealistic expectations
The result? A bloodbath.
Common Themes in 2025 Failures
Looking across these failures, patterns emerge:
1. Pandemic Timing - Companies founded during 2020-2021 boom times are failing now as markets normalize and funding disappears.
2. Fake It Till You... Don't Make It - Builder.ai's fake AI is the extreme example, but many startups oversold capabilities they couldn't deliver.
3. Hardware is Still Hard - Canoo, Lilium, Rain AI: all discovered physical products are way harder than software.
4. Regulatory Risk Matters - EasyKnock, Spotlight Therapeutics: sometimes regulations and reality say "not yet."
5. Unit Economics Eventually Matter - You can't lose money on every transaction and make it up in volume. Eventually, someone asks "how does this actually make money?"
6. Market Timing is Everything - Pandion had a great business model... for 2020. By 2025, the market had moved on.
Experts predict 2025 will likely see continued high volumes of startup failures, with most companies that raised in 2020-2022 either finding a new path forward or shutting down by early 2025 TechCrunch.
But here's the thing: this is actually healthy. The 2020-2021 funding bubble created hundreds of companies that should never have raised $100 million. The market is correcting. The survivors will be leaner, meaner, and built on actual sustainable business models rather than VC-subsidized growth fantasies.
For founders heading into 2026, the message is clear:
- Revenue matters more than growth
- Unit economics matter more than TAM
- Real technology matters more than buzzwords
- Sustainable businesses beat blitz-scaling
- Cash is king, burn is death
The golden age of "raise money, figure it out later" is over. Welcome to the era of "figure it out, then raise money."
Silver Linings?
Not every failure is permanent. Plenty restructured and survived Tech Funding News. Convoy's technology was acquired by Flexport. Rain AI might sell its IP. Builder.ai's team will build new companies.
Failure in startups isn't death - it's education. These founders will come back smarter, more cautious, and better prepared. The ecosystem learns from each collapse.
And hey, at least we'll have plenty of content for "Top 10 Startup Failures of 2026."
Stay scrappy out there.
