TechCrunch's AI Graveyard Tracks Shuttered Projects, From Siri Delays to OpenAI's Super App

TechCrunch has compiled a running list of AI projects and startups that shut down or failed to meet expectations, a record that now spans delayed features at Apple, a messy product launch at OpenAI, and dozens of quieter failures across the industry. The list, published this week, is meant to serve as a counterweight to the steady stream of funding announcements and model releases that dominate AI coverage. It arrives as investors and enterprise buyers grow more skeptical of timelines that slip and products that ship half-finished.

Apple's Siri overhaul is among the most prominent entries. The company has repeatedly pushed back the release of a more conversational, AI-powered Siri, missing deadlines it set publicly and leaving users with a voice assistant that trails rivals from Google and Amazon. Apple has said the work is ongoing, but the delays have become a case study in how hard it is to retrofit large language models onto a shipping product used by hundreds of millions of people.

OpenAI's rollout of what it called a "super app" also made the list. The launch drew criticism for a confusing feature set and technical stumbles, a rare public misstep for a company that has otherwise set the pace in generative AI. The episode showed that even the best-funded labs can fumble execution when they try to bundle too many capabilities into a single product. OpenAI has since adjusted the offering, though the early reception remains a cautionary tale.

Beyond those two names, the graveyard includes a long tail of startups that raised money on AI promises and then folded. Some built thin wrappers around existing models and could not defend against price cuts from foundation labs; others misjudged demand for niche applications. The pattern echoes earlier tech cycles, where a wave of companies formed around a new platform and most disappeared once the platform owners absorbed their features.

The list matters because it gives a clearer picture of where AI is actually working and where it is not. Failure data is scarce in a field that prefers to highlight benchmarks and funding rounds, so a public tally of shutdowns and missed deadlines fills a gap for founders, investors, and engineers deciding what to build next. It also puts pressure on companies to be more honest about what their products can do today rather than what they might do later.

What to watch next is whether the graveyard grows faster than the list of successes. If foundation models keep improving and platform owners keep bundling features, more application-layer startups will likely face the same fate as the ones already recorded. The open question is whether any of the delayed projects, Siri included, can eventually ship in a form that justifies the wait.