TLDRocket
Sign in

The AI graveyard: a running list of projects and startups that didn’t make it

TechCrunch Lauren Forristal

Relay shut down as bigger AI tools swallowed its job. It’s part of a growing graveyard: even OpenAI, Apple, and Microsoft have missed hard.

Based on reporting by TechCrunch, Lauren Forristal — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Relay is gone. The AI workflow startup, built as a Zapier alternative, shut down on Monday after five years of trying to make email and task automation feel like a standalone product. That became a harder sell once OpenAI, Google, and other platform giants started baking similar automation into their own tools. Relay is a neat example of a broader problem: AI products don’t just die from competition. Sometimes they get absorbed by the very companies that inspired them.

The attrition rate inside big companies is ugly too. According to S&P Global Market Intelligence, about 42% of AI initiatives end up abandoned by their corporate parents. The reasons are usually familiar — funding dries up, the tech doesn’t cooperate, scaling gets messy, or users simply don’t care enough. That’s the backdrop for this growing AI graveyard, which now includes launches, pivots, shutdowns, and the occasional embarrassing retreat.

OpenAI, despite all its money and status, has had its own pile of misses. On July 9, it tried to turn ChatGPT into a broader super app by combining separate Chat, Codex, and Work modes and renaming the old version ChatGPT Classic. Users hated the cluttered new interface, and OpenAI rolled it back. The company has also been folding standalone products back into ChatGPT: Atlas, its AI browser, was discontinued on August 9 before it had even made it to a year; Operator’s web-browsing agent features were absorbed as well; and image generation has increasingly moved inside ChatGPT, shrinking DALL-E’s role. Sora, its video-sharing platform, later shut down in March 2026 after running into high operating costs and retention problems.

The pattern repeats elsewhere. Apple kept delaying the improved Siri promised under Apple Intelligence in 2024, eventually landing the company in a $250 million settlement over how it marketed iPhone 16’s AI capabilities. Microsoft’s Recall drew immediate privacy panic when it was announced at Build in 2024, then spent nearly a year in delay; even after a redesign, a security researcher recently found a way to extract captured data. And then there are the flashy hardware bets: Humane’s AI Pin raised $230 million and still ended with HP buying most of the assets for $116 million, while Rabbit’s R1 kept selling units but got dinged for feeling unfinished. The industry keeps treating AI like gravity doesn’t apply. It does.

My take — AI-written commentary, not fact-checked reporting

The lesson here is boring, which is usually how the real lessons arrive: bundling beats novelty. If a feature can be copied into ChatGPT, Apple Intelligence, or a browser, a cute startup around that feature is living on borrowed time. The AI market is starting to look less like a gold rush and more like a very expensive recycling program.

Read more about this at: TechCrunch

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.