Tracking the AI Graveyard: What Recent Product Failures Teach the Tech Industry
From Character.ai's bankruptcy to failed hardware bets, a growing list of abandoned AI projects highlights the severe risks of platform overlap and high operational costs.
Building artificial intelligence products is proving to be a high-stakes enterprise where even well-funded projects frequently fail to survive. According to data from S&P Global Market Intelligence, roughly 42% of corporate AI initiatives are ultimately abandoned by their parent companies. The reasons for these failures span technical roadblocks, unsustainable operating costs, intense competition, and a lack of clear user demand. Recent high-profile exits illustrate the trend: consumer chatbot platform Character.ai filed for bankruptcy in early October after raising $150 million at a $1 billion valuation in 2023, while privacy-focused family assistant Ollie announced its shutdown effective November 6.
Character.ai, founded in 2021 by former Google researchers Noam Shazeer and Daniel De Freitas, faced legal headwinds including lawsuits alleging its chatbots contributed to teen self-harm, as well as a cease-and-desist letter from Disney. Although Google paid $2.7 billion in 2024 to license its technology and rehire its founders, the standalone company was unable to remain viable. Disney has since stepped in as the opening bidder with a $20 million offer for some of its intellectual property in a sale expected to wrap up by late November or early December.
Big tech missteps and corporate pivots
Even industry leaders have encountered friction when attempting to package or deploy new AI capabilities. OpenAI, despite its massive market valuation, rolled back an ambitious interface redesign introduced on July 9 that attempted to consolidate "Chat," "Codex," and "Work" modes into a single layout while renaming the standard version "ChatGPT Classic." Following user criticism over interface clutter, the company returned to the familiar layout. OpenAI has also retired several standalone ventures, including its AI-powered browser ChatGPT Atlas—discontinued on August 9 after less than a year—and folded tools like its Operator agent and DALL-E directly into ChatGPT. Meanwhile, its standalone video platform Sora shut down in March 2026 under the weight of high operating costs and user retention challenges.
Other major tech firms have faced their own hurdles. Apple repeatedly delayed its revamped, context-aware Siri feature due to engineering issues and software bugs, contributing to a $250 million legal settlement over iPhone 16 marketing claims before finally pushing Siri AI into the iOS 27 beta in July. Microsoft faced immediate privacy backlash over Recall, an AI "photographic memory" feature for Windows PCs announced at Build 2024. Despite a redesign following a year-long delay, cybersecurity researchers recently demonstrated tools capable of extracting captured sensitive data. Additionally, productivity platform Notion scheduled the shutdown of Notion Mail for September 22 after observing that users preferred using separate external AI agents for inbox management.
Hardware failures and niche tool consolidation
Hardware ventures attempting to move beyond traditional smartphones have seen particularly sharp turns. Humane raised $230 million for its AI Pin wearable, but poor performance and a battery fire hazard warning regarding its charging case led to the business shutting down in February 2025, with HP acquiring most of its assets for $116 million. Rabbit, which sold 100,000 units of its R1 device following its CES 2024 announcement, faced early criticism over unreliable performance and incomplete software, though it has since pivoted toward computer control tasks and announced a new hardware project called "Project Cyberdeck."
Specialized software products have similarly struggled to maintain independence as broader platforms absorb their functionality. Five-year-old workflow automation tool Relay shut down in August 2026 as tech giants built competing native features. Audio app Huxe, created by former Google NotebookLM developers to turn text into podcast-style audio, folded in May 2026 as platforms like Spotify expanded their native AI tools. Free testing sandbox Yupp, which allowed users to compare over 800 AI models side-by-side, closed in March 2026 after failing to find product-market fit. Similarly, storytelling platform Figgs AI shut down after finding its free model financially unsustainable despite attracting over 1 million users.
What it means for developers
The expanding list of abandoned AI projects highlights critical tactical lessons for software engineers and founders. First, relying entirely on a thin layer of functionality over existing foundation models carries severe platform risk. As companies like OpenAI, Google, and Apple integrate agentic tasks, web browsing, and workflow automation directly into their ecosystems, standalone utilities like Relay, Notion Mail, and Huxe quickly lose their core value proposition.
Second, unit economics and infrastructure expense remain primary failure points. High compute costs crushed standalone platforms like Sora and forced free services like Figgs AI out of operation. Developers must carefully manage API expenses, model selection, and token usage to ensure long-term sustainability. To test multi-model workflows without taking on massive infrastructure commitments, developers can try top AI models cheaply through one API at https://apixoai.online.
Finally, security compliance alone does not guarantee market traction. As demonstrated by Ollie's closure despite achieving SOC 2 compliance, and Microsoft's ongoing challenges with Recall, security standards must be matched with strong product-market fit and rigorous protection of user data. As the AI sector matures, building resilient applications will require balancing cost-efficiency with features that larger platforms cannot easily replicate.
Source: The AI graveyard: a running list of projects and startups that didn’t make it — TechCrunch AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
