The artificial intelligence boom has produced an unmistakable paradox: record-breaking private valuations alongside a rapidly growing graveyard of startups, products, and initiatives that simply could not survive. Two recent casualties illustrate the pattern with brutal clarity. Relay, an AI-powered workflow automation tool that positioned itself as a Zapier alternative, shut down entirely in August after five years of operation. Just weeks later, chatbot platform Character.ai filed for bankruptcy in early October, only five years after its founding. These are not isolated incidents. They are signals of a fundamental market reality: the window for standalone AI startups is narrowing fast.
The scale of AI failure: 42 percent of initiatives abandoned
According to S&P Global Market Intelligence, roughly 42 percent of all AI initiatives are ultimately abandoned by their corporate parents. The reasons span insufficient funding, technical challenges, intensifying competition, difficulty scaling, and weak user demand. This attrition is not confined to small startups. Some of the most well-funded and hyped AI projects—from hardware devices to flagship apps—have collapsed, pivoted, or significantly missed expectations. What follows is not a scorecard of flops but a record of what these attempts can teach the rest of the industry about the brutal economics of building AI products today.
The era of easy growth and generous venture capital is giving way to a more demanding phase.
OpenAI’s own missteps: a super app redesign that backfired
Even OpenAI, one of the most valuable private companies in the world, has not been immune to failure. In July, the company attempted to turn ChatGPT into a broader “super app” by combining separate modes—Chat, Codex, and Work—while renaming the traditional interface “ChatGPT Classic.” The redesign drew immediate criticism from users who found the new interface confusing and cluttered. OpenAI rolled back the change within days, restoring the familiar single-interface layout. The episode revealed the risk of overcomplicating a product that had already achieved mass adoption through simplicity.
Atlas, Operator, DALL-E, and Sora: features folded or shut down
The super app attempt was not OpenAI’s only consolidation play. ChatGPT Atlas, a stand-alone AI-powered web browser launched less than a year earlier, was discontinued on August 9, with its most useful features absorbed into ChatGPT. The company followed the same pattern with Operator, an AI agent designed to browse the web and complete tasks on a user’s behalf. Image generation capabilities have increasingly been built directly into ChatGPT, shrinking DALL-E’s role as a separate destination. Most notably, OpenAI’s video-sharing platform Sora shut down in March 2026 after struggling with high operating costs and poor user retention. For OpenAI, folding features into ChatGPT has worked better than maintaining separate products—but even the mother ship cannot afford to alienate its core user base with unnecessary complexity.
Apple’s Siri AI: delayed, sued, and finally released
Apple’s long-promised AI overhaul of Siri provides another cautionary tale about the gap between marketing and execution. When Apple unveiled Apple Intelligence in 2024, an improved Siri was center stage—a voice assistant that could understand context, work across apps, and complete tasks effectively. But the release was repeatedly delayed due to engineering issues and bugs. Those delays ultimately contributed to a $250 million settlement over claims regarding how Apple marketed the AI capabilities of the iPhone 16. The new Siri AI finally appeared in the iOS 27 beta in July, rolling out to English-language users this month, with other languages to follow. The episode underscores a hard truth: even the world’s most valuable company cannot rush AI reliability.
Microsoft’s Recall: privacy firestorm and lingering doubts
Microsoft’s Recall feature, announced at its 2024 Build developer conference, was pitched as an AI-powered “photographic memory” for Windows PCs. The tool would periodically take screenshots of a user’s activity so they could later search through their digital history. The announcement triggered an immediate backlash over privacy and security, with critics warning that Recall could create a searchable archive of passwords, private messages, and financial details. Microsoft delayed the feature for nearly a year to redesign its security and privacy protections. Yet the controversy has not subsided. A cybersecurity researcher recently built a tool that could extract and display data captured by Recall, reigniting questions about whether the underlying problem has truly been solved. Microsoft’s experience shows that AI features that touch sensitive user data face an especially steep trust barrier.
The hardware graveyard: Humane AI Pin and Rabbit R1
Humane AI Pin: $230 million raised, then a battery fire warning
Perhaps the most famous AI hardware failure is the Humane AI Pin. Designed as a wearable device to replace the smartphone, Humane raised $230 million from investors and generated enormous pre-launch buzz. But the product struggled with performance issues from the start. The situation turned critical when Humane warned customers to stop using the charging case due to a potential battery fire risk. In February 2025, the company shut down its AI Pin business, and most of its assets were acquired by HP for $116 million—less than half the amount raised. The lesson: ambitious hardware concepts demand not just AI capability but also flawless industrial design, battery safety, and everyday reliability.
Rabbit R1: 100,000 units sold but an unfinished product
The Rabbit R1, unveiled at CES in January 2024, was another heavily hyped AI companion device designed to perform tasks on behalf of its owner. Rabbit said it sold 100,000 units shortly after launch. But early reviews described the device as unfinished, with unreliable performance and a limited number of useful integrations. Unlike Humane, Rabbit has not given up. The company continues to update the R1, positioning it as a computer controller for agentic tasks, and recently announced a new hardware project called “Project Cyberdeck” aimed at portable vibe-coding. The R1’s trajectory illustrates how strong initial sales can mask deep product shortcomings—and how difficult it is to recover from a half-baked launch.
Consumer AI apps that could not find their footing
Character.ai: from billion-dollar valuation to bankruptcy
Character.ai was founded in 2021 by former Google researchers Noam Shazeer and Daniel De Freitas. The app let people chat with customizable bots modeled on celebrities and fictional characters, and it quickly became a buzzed-about consumer AI destination. The company raised $150 million at a $1 billion valuation in 2023. The following year, Google paid a reported $2.7 billion to license its technology and bring its founders back, leaving the startup to continue without them. Then the trouble began. Families sued, alleging that the chatbots contributed to teen suicides and self-harm. Disney sent a cease-and-desist letter over unauthorized use of its characters. Character.ai barred under-18 users from open-ended chats and agreed to settle several suits in early 2026. Despite these efforts, the company filed for bankruptcy in early October, citing legal, regulatory, and operational headwinds. In an ironic twist, Disney is now the opening bidder, offering $20 million for a license to some of its intellectual property. The app remains running during the bankruptcy process, with the sale expected to conclude by late November or early December. Character.ai’s story highlights the unique legal and ethical risks that come with allowing users to interact with AI personalities—especially when those personalities are beloved (and trademarked) characters.
Figgs AI: free became too expensive
Figgs AI operated from 2023 to 2024, letting users create and interact with customizable AI characters for role-play and storytelling. The platform drew more than 1 million users, but its developers said keeping the service free became too expensive to sustain, and they shut it down. Figgs AI’s fate is a stark reminder that user growth without a viable monetization model is a ticking clock.
Notion Mail: users preferred separate agents
Notion Mail launched in April 2025 as an AI-focused email product designed to help users organize and automate their inboxes. But Notion noticed that its users were turning to separate AI agents to handle their email instead. As a result, Notion Mail is scheduled to shut down on September 22. The product’s failure reveals how quickly user behavior can shift when more flexible AI tools become available—and how hard it is for a single-purpose app to compete with general-purpose AI agents.
Yupp: 800 models, but no product-market fit
Yupp was an intriguing AI experiment: a free playground that let users compare responses from hundreds of AI models side by side, vote on responses, and earn cryptocurrency. At its peak, the platform offered more than 800 models from companies including OpenAI, Google, and Anthropic. But according to its founders, Yupp never achieved strong enough product-market fit to survive. The platform shut down in March 2026. Yupp’s case shows that even a technically impressive aggregation tool can fail if it doesn’t solve a clear, urgent problem for a willing-to-pay audience.
Huxe: audio generation crushed by platform competition
Huxe was an AI audio app built by former developers of Google’s NotebookLM. It turned written information into conversational, podcast-style audio. The company shut down in May 2026, in part because larger platforms began offering similar experiences to large existing audiences. Spotify, for example, expanded its own AI-powered audio tools around the same time. Huxe’s shutdown illustrates a recurring theme: when platform giants add a feature that mimics a startup’s core value, the startup rarely survives.
What market realities mean for AI startups
What do these stories tell us? First, consolidation is accelerating. OpenAI, Google, Microsoft, Apple, Spotify, and other platform players are absorbing the most valuable AI features directly into their ecosystems, leaving little room for standalone products. Second, hardware is an especially unforgiving category for AI startups: the combination of massive capital requirements, manufacturing complexity, and user expectations for reliability creates a nearly insurmountable barrier. Third, consumer AI apps that rely on free usage and viral growth face a precarious future unless they find a path to revenue before investor patience runs out. Fourth, legal and regulatory risks—particularly around AI-generated content, privacy, and harm to minors—are growing and can capsize even well-funded companies like Character.ai.
The AI graveyard is not a sign that artificial intelligence is overhyped. It is a sign that the market is maturing. The era of easy growth and generous venture capital is giving way to a more demanding phase. Future AI startups will need to offer deep, defensible value in niches that large platforms cannot easily replicate, or they will need to accept that their most likely exit is an acquisition by one of the giants. The graveyard is growing, but the lessons it offers are invaluable for anyone building the next generation of AI products.
- What percentage of AI initiatives are abandoned?Approximately 42 percent of all AI initiatives are ultimately abandoned by their corporate parents, according to S&P Global Market Intelligence.
- Which AI startups recently shut down?Relay, an AI-powered workflow automation tool, shut down in August after five years, and Character.ai filed for bankruptcy in early October.
- Why did OpenAI's super app redesign fail?OpenAI's attempt to turn ChatGPT into a super app drew criticism for confusing complexity, forcing a rollback within days.
