The AI graveyard grows as startups, features and chatbots disappear

TechCrunch says the AI boom is entering a harsher phase: a growing number of apps, products and startup bets are being shut down, rolled back or absorbed into larger platforms. The clearest example is Character.ai, which filed for bankruptcy in early October after years of legal, regulatory and operational pressure, but it is far from alone.

The pattern matters because it suggests the market is no longer rewarding novelty on its own. Consumer AI products now have to survive not just hype, but high compute bills, weak retention, safety scrutiny and fast-moving copycats from bigger companies with deeper distribution.

AI’s graveyard is getting crowded

TechCrunch framed its reporting as a roundup of AI products, startups and internal bets that have shut down, pivoted or failed to meet expectations. The list spans well-known brands, side projects inside major tech companies and smaller startups that briefly drew attention before running into economics, regulation or product-market fit problems.

That mix is important. This is not just a story about a few weak startups. It is also about what happens when a feature launches into a market where larger platforms can quickly build something similar, package it inside a broader product and make the standalone version redundant.

How the pattern shows up in the numbers

The clearest warning sign comes from S&P Global Market Intelligence, which TechCrunch cited as saying that about 42% of AI initiatives are eventually abandoned by corporate parents. That figure helps explain why so many consumer-facing AI tools are now disappearing almost as quickly as they arrived.

Here is a snapshot of the shutdowns, pivots and pressure points TechCrunch highlighted.

Why it matters

This wave of shutdowns matters because it shows where the AI market is breaking down: not in model quality alone, but in the business of turning AI into durable products that users keep paying for or keep using.

For users, the result is instability. A tool can feel essential one month and disappear the next, or be absorbed into a larger app with fewer standalone controls. For investors and operators, the message is sharper: distribution, retention and cost control may matter as much as the underlying technology.

For bigger platforms, the moment is a reminder that they can often wait for an emerging category to prove itself, then fold the useful parts into a much broader product line. That is especially true in chatbots, inbox tools, audio products and automation assistants, where the core experience can be replicated inside a dominant app.

What the AI graveyard says about the market

The underlying problem is economics. AI products can attract attention quickly, but the bills do not vanish when the novelty wears off. Compute is expensive. Safety work is ongoing. Regulators and courts can intervene. And if users do not return often enough, even a well-known product can become hard to sustain.

That combination explains why the most fragile products are often the ones that sit close to commodity use cases. If a larger company can add the same function to an existing product, the standalone startup may lose the very reason it exists.

Character.ai: from breakout startup to bankruptcy

Character.ai is the most dramatic case in TechCrunch’s roundup. Founded in 2021 by former Google researchers Noam Shazeer and Daniel De Freitas, the company quickly became one of the best-known consumer AI chatbots. It raised $150 million in 2023 at a $1 billion valuation, a figure that reflected the market’s belief that personality-driven AI chat could become a major consumer category.

That optimism did not survive the company’s legal and operational pressures. TechCrunch says Character.ai filed for bankruptcy in early October after years of strain, including regulatory and legal scrutiny. The company also barred under-18 users from open-ended chats late last year, and it later settled several lawsuits in early 2026 along with Google.

How did Google become part of the story?

Google is central because TechCrunch says the company reportedly paid $2.7 billion in 2024 to license Character.ai’s technology and bring its founders back. That deal underlines a recurring pattern in AI: even when an independent product struggles, its technology and talent can still be valuable to a larger platform.

In other words, the startup may not survive as a standalone company, but its ideas can be reabsorbed elsewhere. That may look like a rescue from the outside. It is also a sign that the market is consolidating around companies that can afford the technical and legal overhead of operating at scale.

What happens in the bankruptcy sale?

Bankruptcy does not necessarily mean disappearance. It can also become a mechanism for selling assets, including intellectual property and technology licenses. In Character.ai’s case, Disney is the opening bidder, offering $20 million for a license to some IP.

That bid raises the central question now hanging over the sale: what exactly is left to be sold, and who wants it most. The answer will shape whether Character.ai becomes a cautionary tale, a licensing shell or a set of assets broken up for parts.

OpenAI keeps folding products back into ChatGPT

OpenAI’s actions show another side of the same market pressure. Instead of disappearing entirely, some products are being collapsed into a broader flagship app, where they may be easier to manage and easier to explain to users.

On July 9, OpenAI rolled back a ChatGPT redesign after users complained the interface was confusing and cluttered. The reversal suggests that even popular products can push too hard on complexity and lose users’ trust or patience.

Then, on August 9, OpenAI discontinued ChatGPT Atlas and folded its features into ChatGPT. The company also shut down Operator as a standalone product and moved image generation further into ChatGPT.

In March 2026, OpenAI shut down Sora, citing high operating costs and retention problems. That decision is especially revealing because Sora had been among the most visible AI video products in the market. Even there, interest was not enough to offset the economics.

Why are bigger platforms consolidating AI features?

Bigger platforms consolidate AI features because one strong interface is often easier to support than several separate ones. It can reduce confusion, concentrate engagement and keep users inside a single product. But consolidation can also be read as a warning sign: if features cannot sustain themselves independently, the market may be forcing them into bundling as a survival strategy.

Apple and Microsoft show the costs of delay and backlash

The AI graveyard is not limited to startups. Large companies are also running into friction when features stall or draw criticism.

Apple repeatedly delayed the new Siri tied to Apple Intelligence because of engineering issues and bugs, according to TechCrunch. That kind of delay matters because it can slow down a platform company’s entire AI rollout strategy and weaken the sense that it is keeping pace with competitors.

Microsoft’s Recall feature faced a different problem. TechCrunch says it drew privacy and security backlash after its 2024 debut, and later faced renewed scrutiny from a researcher-built extraction tool. That episode shows how quickly trust can erode when an AI feature touches sensitive data or is perceived as too invasive.

Both cases point to a larger truth: for major platforms, the challenge is no longer just adding AI. It is adding it without creating bugs, privacy concerns or user resistance that undercut the feature before it matures.

Who is getting squeezed out?

The products most at risk are the ones that sit between novelty and necessity. They may be clever, useful or well-liked, but not indispensable enough to survive competition from larger platforms or the expense of running them alone.

TechCrunch’s roundup includes several examples of that squeeze:

• Notion Mail, scheduled to shut down on September 22 after users favored other AI inbox tools.
• Relay, shut down in August 2026 after larger platforms built similar automation features.
• Ollie, which said it will shut down on November 6 because it was unable to stand out, was failing to grow and had not raised enough capital.
• Huxe, which shut down in May 2026 as larger platforms offered similar AI audio experiences.
• Yupp, shut down in March 2026 after its founders said product-market fit was not strong enough.
• Figgs AI, which shut down after costs became too high to keep the service free.

Each closure has its own explanation, but the wider theme is consistent: users migrate quickly, differentiation is hard to defend, and the cost of being first is not always offset by staying power.

What Ollie’s shutdown says about startup pressure

Ollie’s explanation is blunt and useful. Co-founder and CEO Bill Lennon said the company was “unable to stand out, failing to grow, and not raising enough capital”. That line captures the basic problem facing many small AI firms: even a promising product can lose momentum if it cannot convert attention into durable scale or funding.

The startup’s failure is not unusual in isolation. What makes it noteworthy is how closely it mirrors the fate of other AI apps in this roundup. The result is a market where product quality alone is rarely enough.

Background: how we got here

The current wave of shutdowns fits a broader pattern in AI. Startups often launch quickly, ride a burst of interest and then confront a hard combination of compute costs, weak retention, regulatory scrutiny and rapid feature copying by larger platforms. TechCrunch notes that chatbots and agentic tools can be reabsorbed into bigger products when standalone demand proves weak.

That is why the boundary between “successful product” and “feature that disappears” has become so thin. A small team can demonstrate demand, but a much larger company can often outspend, out-distribute and outlast it.

Bankruptcy also adds another layer to the story. It can be used to sell assets, including intellectual property and technology licenses, which means a company’s end may still leave behind valuable pieces for someone else to acquire. In the AI sector, that can turn a collapse into a transfer of value rather than a clean shutdown.

By the numbers

The figures in TechCrunch’s roundup show just how quickly the AI market can move from expansion to retrenchment.

What happens next

The immediate next milestone is Character.ai’s bankruptcy sale, which TechCrunch says is expected to close in late November or early December. Whether that happens on schedule will shape the company’s next chapter and determine how much of the brand, technology or IP changes hands.

The other date to watch is September 22, when Notion Mail is scheduled to shut down. That closure, along with Ollie’s planned November 6 exit, will test whether users migrate smoothly to other AI inbox and automation tools or whether these categories remain vulnerable to consolidation.

More broadly, the next phase of the AI market will likely determine which products can survive as standalones and which will be absorbed into larger suites. The open questions are straightforward:

• Will Character.ai’s sale close on the expected timeline?
• What, if anything, will Disney ultimately obtain from the IP auction?
• How many of these products will be revived, copied or bundled into bigger platforms?

For now, the signal from TechCrunch’s roundup is clear: the AI boom has entered a sorting phase. The winners are no longer just the most talked-about products. They are the ones that can survive cost pressure, keep users engaged and avoid being swallowed by a larger platform that can do the same job inside a broader app.

FAQ

Why are so many AI apps shutting down?

Many AI apps are shutting down because they cannot overcome a mix of high compute costs, weak retention, regulatory pressure and direct competition from larger platforms that copy or absorb their features. TechCrunch’s roundup suggests that attention alone is not enough to keep a product alive.

What happened to Character.ai?

Character.ai filed for bankruptcy in early October after years of legal, regulatory and operational pressure. The company was founded in 2021, raised $150 million in 2023 at a $1 billion valuation, and later became part of a reported $2.7 billion Google licensing deal in 2024.

Is OpenAI shutting down products or just redesigning them?

Both. OpenAI rolled back a July 9 ChatGPT redesign after complaints, discontinued ChatGPT Atlas on August 9 and folded its features into ChatGPT, shut down Operator as a standalone product, and closed Sora in March 2026 because of high operating costs and retention issues.

What does this mean for smaller AI startups?

It means the bar is rising fast. Smaller AI startups must prove not only that people will try the product, but that they will keep using it and that the company can afford to run it. Without clear differentiation, strong growth and enough capital, many of them may not last.

Will these AI products come back under new owners?

Some may, but not all. Bankruptcy sales can transfer intellectual property and technology licenses, and larger platforms often absorb useful features into their own products. Whether a specific brand returns depends on who buys the assets and whether there is still a market for the standalone product.