AI Projects and Startups Fail, Expanding the AI Graveyard

Recent AI product failures highlight a widening gap between investment, expectations and commercial durability in the sector. While AI spending remains elevated, many standalone tools and corporate initiatives are being discontinued, absorbed into larger platforms, or abandoned because they cannot sustain demand, differentiation or trust. S&P Global Market Intelligence data suggests about 42 percent of AI initiatives are ultimately abandoned by their parent companies, citing insufficient funding, technical obstacles, competition, scaling difficulties and weak user demand. The most visible collapse came from Character.ai, a chatbot platform that filed for bankruptcy five years after launch, underscoring how quickly category leaders can falter under competition and cost pressure. Relay, a Zapier alternative built around AI-powered workflow automation, shut down in August despite five years of development, as major platforms including OpenAI and Google embedded similar capabilities directly into their own products. Even OpenAI has experienced high-profile reversals. On July 9, the company introduced a redesign that added separate Chat, Codex and Work modes and renamed the familiar interface ChatGPT Classic, but user backlash over confusion and clutter led it to roll back the change. The company has also consolidated stand-alone products into ChatGPT, discontinuing the ChatGPT Atlas browser on August 9 after less than a year and folding Operator into ChatGPT, while reducing DALL-E’s separate role. Its video-sharing platform Sora shut down in March 2026 after high operating costs and weak retention. Consumer device and feature launches have also struggled. Apple’s upgraded Siri AI, promoted with Apple Intelligence in 2024, repeatedly missed release targets and contributed to a $250 million settlement over marketing claims tied to the iPhone 16. A beta finally appeared in iOS 27 in July, with English-language rollout beginning this month. Microsoft’s Recall feature, unveiled at Build 2024 as an AI-powered search of a user’s PC history, drew immediate privacy objections and was delayed for nearly a year. Security concerns persisted after researchers demonstrated tools to extract captured data. Notion Mail, launched in April 2025 as an AI email product, is scheduled to shut down on September 22 after users preferred independent AI agents. Hardware remains especially risky. Humane’s AI Pin raised $230 million but collapsed under performance problems and a battery-safety warning, shutting down in February 2025, with HP acquiring most assets for $116 million. Rabbit’s R1 sold about 100,000 units after its January 2024 CES debut but faced criticism for unreliability and limited integrations. The company continues updating the device and has announced Project Cyberdeck. Software experiments such as Huxe, an AI audio app from former NotebookLM developers, shut down in May 2026, and model-comparison platform Yupp, which at peak offered more than 800 models for side-by-side testing, closed in March 2026 after failing to secure product-market fit. The pattern suggests that many AI products are failing not because demand for AI has vanished, but because the space has become crowded and increasingly dominated by major platforms with distribution, capital and integrated ecosystems.