The AI hype cycle has already buried hundreds of companies. According to CB Insights, more than 500 AI startups shut down or went dark between 2023 and 2025. The money they burned tops $10 billion in disclosed funding alone. Most of it is gone forever.
Why the Graveyard Is Growing Right Now
Everyone told you AI was the next internet. Maybe it is. But the first wave of internet companies also produced a graveyard that stretched from 1999 to 2002. We called it the dot-com bust. History rhymes.
The difference this time is speed. Startups raised enormous rounds in 2022 and 2023 on the back of ChatGPT excitement. Valuations had nothing to do with revenue. They were bets on attention. When attention moved on, the money ran out fast.
Humane raised $240 million, according to Crunchbase, to build the AI Pin. The device launched in April 2024 at $699 and got destroyed in reviews. Battery life was poor. Responses were slow. The camera overheated. By late 2024, the company was reportedly trying to sell itself for less than it had raised. According to The Verge, it eventually sold its software assets to HP for roughly $116 million. Investors lost more than half their money.
Rabbit raised money on the back of its R1 AI gadget, which shipped in 2024 to similarly lukewarm reception. The device was ly a phone app wrapped in orange plastic. According to Android Authority, users could replicate most of its features with a $0 app download. The hardware play was dead on arrival.
These are not edge cases. They are symptoms of a pattern.
The Pattern Behind Every Failed AI Bet
I have watched money move through tech cycles for years. The AI graveyard follows a simple formula. A real breakthrough happens. Venture capital floods the space. Founders build for demos, not deployment. Costs are ignored. Revenue is promised but not built. Then reality arrives.
Inflection AI raised $1.5 billion, according to Bloomberg, to build a personal AI assistant called Pi. The product was thoughtful and well built. But in March 2024, Microsoft hired the entire leadership team including CEO Mustafa Suleyman. The company was effectively gutted. Its remaining assets were licensed, not sold. That is what a soft failure looks like at the top of the market.
Amazon spent an estimated $10 billion or more on Alexa over a decade, according to reporting from Business Insider. The product had 100 million devices in homes. Yet it never found a business model that covered its costs. Amazon began laying off Alexa teams and pivoting the product toward paid subscriptions in 2023. A hundred million units and billions spent. Still not a real business.
The common thread is unit economics. Nobody asked whether the product could ever make more money than it cost to run. When you are running large language model inference at scale, the compute costs are brutal. Without a clear path to covering those costs, you are burning investor money with a timer attached.
That is the rich versus poor divide in AI investing. Rich thinkers ask: what does this cost to deliver at scale, and who pays for it? Poor thinkers ask: is the demo impressive?
If you are comparing financing options to fund a business or project during a period like this, SuperMoney loan comparison lets you see rates from multiple lenders side by side before you commit to anything.
What This Means for You
You are not a venture capitalist. But this still affects you directly.
First, if you work in tech or AI, the companies you work for are not immune. Evaluate your employer the same way a smart investor would. How much runway do they have? What does their burn rate look like? Are customers paying or are they on free trials? Those questions protect your income.
Second, the data your accounts are sitting in at failed AI companies does not disappear cleanly when a startup shuts down. Bankruptcy sales often include customer databases. Your email, your usage patterns, maybe your payment history, can transfer to a buyer you never agreed to work with. If you have used a string of AI apps over the past two years, your data footprint is wider than you think. Monitoring your credit and identity through a service like IdentityIQ credit monitoring can catch early signs that your data has moved somewhere it should not have.
Third, and this is the move I would actually make: the graveyard creates opportunity. Failed AI companies leave behind open source models, abandoned research, and former employees who know exactly what did not work. The people who understand failure in AI are more valuable right now than the people who only know the hype. Hire them. Learn from them. Build on what they left behind.
The best time to shop is after a fire sale. We are in one right now.
The Bottom Line
The AI graveyard is not evidence that AI is fake. It is evidence that money moves faster than markets. The technology is real. The hype around it was not. We are now in the period where the serious builders separate from the pitch deck founders. The companies that survive 2026 will own the next decade. The ones that die will fund the lesson. Know which side you are on.
Frequently Asked Questions
How many AI startups have failed so far?
According to CB Insights, more than 500 AI startups shut down or went dormant between 2023 and 2025. That figure only counts companies with disclosed funding. The actual number including bootstrapped projects is likely far higher.
Why did Humane AI fail if it raised so much money?
Humane raised $240 million but built hardware in a category with no proven demand. The AI Pin required a subscription on top of the purchase price and delivered a worse experience than a smartphone already in every buyer’s pocket. High capital plus weak product-market fit is a reliable path to the AI graveyard.
What happens to user data when an AI startup shuts down?
Data often transfers to buyers in bankruptcy or asset sales. Users rarely get notified in a meaningful way. Reading the privacy policy of any AI app before signing up, and monitoring your credit and identity afterward, is a reasonable precaution.
Is the AI graveyard a sign that AI itself is failing?
No. It is a sign that capital was misallocated during a hype peak, which happens in every major technology cycle. The dot-com bust killed thousands of companies and produced Amazon, Google, and eBay. The AI graveyard will do the same thing. The technology survives. The weak implementations do not.
What should investors learn from failed AI companies?
The primary lesson is to evaluate unit economics before the demo impresses you. If a company cannot explain what it costs to serve one customer and what that customer pays, it is not a business yet. It is a bet on future pricing power that may never arrive.
