The growing graveyard of failed AI startups and corporate bets

The corporate graveyard

The rapid pace of AI development has created a significant number of projects that never make it to long-term viability. As TechCrunch reported, this 'AI graveyard' includes everything from independent startups to internal corporate experiments. Many of these projects fail because they cannot compete with the native tools being built by tech giants.

For example, Relay was an AI-powered automation tool designed as an alternative to Zapier. It allowed users to automate emails and tasks using AI agents, but it struggled to survive as a standalone product. This happened because companies like Google and OpenAI began building similar automation features directly into their own existing platforms.

This trend is visible at the enterprise level as well. S&P Global Market Intelligence notes that roughly 42% of AI initiatives are eventually abandoned by their corporate parents. These failures are often driven by a mix of insufficient funding, technical hurdles, competition, and a lack of clear user demand.

Lessons from the giants

Even the most valuable companies in the world are not immune to project failures. OpenAI has had to discontinue several stand-alone products to streamline its offerings. For instance, ChatGPT Atlas, a dedicated AI web browser, lasted less than a year before its features were absorbed into the main ChatGPT app.

The company also faced a significant public relations hurdle with a July 9 redesign of ChatGPT. They attempted to create a 'super app' with different modes like Codex and Work, but users found the new interface cluttered and confusing. OpenAI was forced to roll back the changes quickly to return to the familiar layout.

Other high-profile projects from major players have also hit roadblocks. Sora, OpenAI's video-sharing platform, was shut down in March 2026. The company cited high operating costs and difficulties with user retention as the primary reasons for ending the project.

Hardware and privacy hurdles

The move toward dedicated AI hardware has seen notable failures in both performance and safety. The Humane AI Pin, which raised $230 million, struggled with basic functionality and eventually faced a battery fire risk. Humane shut down its AI Pin business in February 2025, and HP acquired most of the company's assets for $116 million.

Similarly, the Rabbit R1 was marketed as an AI companion that could perform tasks for its owner. While it sold 100,000 units early on, reviews described the device as unfinished with unreliable performance. The company is now trying to pivot the R1 into a computer controller for 'vibe-coding' and developing a new project called Cyberdeck.

Privacy concerns have also derailed major software launches. Microsoft's Recall feature, intended to act as a 'photographic memory' for Windows PCs, faced immediate backlash for potentially capturing sensitive data like passwords. Although Microsoft delayed the feature for a year to address security, researchers have already found ways to extract the captured data.

The collapse of Character.ai

One of the most prominent startup failures is Character.ai, which filed for bankruptcy in early October 2025. The platform became famous for letting users chat with bots modeled on celebrities and fictional characters. However, it faced a perfect storm of legal, regulatory, and operational challenges that it could not overcome.

The company faced lawsuits from families alleging the chatbots contributed to self-harm and received a cease-and-desist letter from Disney over unauthorized character use. While the company settled several suits and barred users under 18 from open-ended chats, the financial and legal pressure remained too high. The startup was left to struggle after Google paid $2.7 billion to license its technology and bring its founders away.

The current status of the company is in a state of transition. While the app is still running during the bankruptcy process, Disney has emerged as the opening bidder. They are offering $20 million for a license to some of its intellectual property, with the sale expected to wrap up by late 2025.

Market saturation and pivots

Many smaller AI products are disappearing because they cannot find a unique product-market fit. Yupp was a playground where users could compare over 800 different AI models and earn cryptocurrency. Despite its reach, the founders stated it never achieved enough traction to survive and shut down in March 2026.

Other niche products are being squeezed out by larger ecosystem updates. Huxe, an AI audio app that turned text into podcast-style audio, shut down in May 2026 as platforms like Spotify began offering similar tools to larger audiences. This highlights how quickly a unique feature can become a standard offering in a major app.

Finally, some companies are choosing to shut down products simply because they are too expensive to maintain for free. Figgs AI, which allowed users to create customizable characters for role-play, drew over 1 million users but could not sustain the costs of the service. These examples show that the 'AI graveyard' is a result of high costs meeting fierce competition from established tech leaders.