
Artificial intelligence has become one of the biggest technology businesses in the world. Companies are investing billions of dollars in AI, startups are attracting large amounts of funding, and new tools are launching almost every week. But behind the excitement, many AI products are struggling to survive, and some have already shut down despite attracting millions of users or raising millions of dollars.
From AI-powered gadgets that failed to impress customers to software products that could not compete with larger companies, the industry has already seen several painful setbacks. Some companies ran out of money, while others discovered that people were not willing to pay for their products. In other cases, bigger technology companies added similar features to products people already used, leaving smaller startups with little reason to exist.
A report published by TechCrunch on October 24, 2026, highlighted several of these cases in what it calls the AI graveyard. The list includes failed startups, discontinued products, delayed features and projects that did not meet expectations. Here are some of the biggest examples, why they struggled and what business owners can learn from their experiences.
Why Are AI Startups and Projects Failing?
The rapid growth of artificial intelligence has created the impression that almost any AI business can become successful. However, building a useful AI product is only one part of running a successful company. A business must also attract customers, earn enough money to cover its costs and give people a good reason to keep using its product.
According to S&P Global Market Intelligence, about 42% of AI initiatives are eventually abandoned by the companies that started them. The reasons include a lack of funding, technical problems, strong competition, difficulty growing the business and weak customer demand. This suggests that even large companies with money and technical experts can struggle to turn AI projects into successful products.
One major problem is that AI technology is changing quickly. A startup may spend months building a tool, only for a larger company to introduce a similar feature inside an existing product. Customers may then choose the new feature because it is easier to access or already included in a service they pay for.
Another challenge is the cost of running AI products. Some tools require expensive computer systems to generate answers, process images or perform tasks. If customers are not paying enough to cover these costs, a company can attract many users and still lose money.
1. OpenAI’s Sora Video Platform
Sora was OpenAI’s AI video generation platform. It allowed users to create videos from written instructions, making it possible to produce visual content without traditional filming equipment or advanced video editing skills.
The product attracted attention because AI video generation offered new opportunities for filmmakers, marketers, content creators and ordinary users. However, operating a video generation service can be expensive because it requires significant computing power to create and process videos.
OpenAI shut down Sora in March 2026 after facing high operating costs and challenges keeping users engaged. Its closure showed that a product can attract enormous attention without becoming a sustainable business.
The lesson is that a popular launch does not guarantee long-term success. Companies must find a way to keep customers interested and generate enough income to support the cost of providing their services.
2. ChatGPT Atlas
OpenAI also discontinued ChatGPT Atlas, its standalone AI-powered web browser, on August 9, 2026, less than a year after its launch. The browser was designed to bring AI assistance directly into the browsing experience, allowing users to work with information on websites in a different way.
However, OpenAI was also adding more features directly to ChatGPT. As those features became part of its main product, there was less need for users to rely on a separate browser just to access certain AI functions.
Instead of keeping Atlas as an independent product, OpenAI moved useful features into ChatGPT. This is an important distinction because Atlas’s discontinuation does not mean all its technology or features disappeared.
The case shows why standalone AI products can struggle when the companies behind them already have a larger platform. If a feature works better as part of an existing product, maintaining a separate app may no longer make business sense.
3. OpenAI’s Operator
Operator was an AI agent designed to browse websites and complete tasks on behalf of users. Rather than simply explaining how to do something, it could interact with websites and attempt to carry out actions for the user.
This type of technology is part of the wider shift from AI chatbots to AI agents. Instead of only answering questions, agents are designed to take steps towards completing a task, such as searching for information or interacting with an online service.
OpenAI eventually folded Operator’s capabilities into ChatGPT rather than maintaining it as a separate destination. The company has been moving towards making more of its AI features available through one main product.
This does not mean AI agents have failed. Instead, Operator’s story shows that a useful technology does not always need to remain a separate product. Sometimes, a company can reach more users by making the feature part of a service people already know.
4. OpenAI’s ChatGPT Super App Redesign
Not every setback in the AI industry involves a product shutting down. Sometimes, a company launches a major change and then has to reverse it because customers do not like it.
OpenAI attempted to turn ChatGPT into a broader super app by combining different experiences, including Chat, Codex and Work modes. The traditional version was renamed ChatGPT Classic as part of the redesign.
However, users criticised the new interface for being confusing and cluttered. OpenAI later rolled back the change and restored the familiar ChatGPT experience.
This case highlights a simple business lesson. Adding more features does not automatically make a product better. If customers find a new design difficult to understand, a company may lose some of the simplicity that made the product popular in the first place.
Businesses should therefore test major changes carefully and listen to customer feedback before forcing a new experience on everyone.
5. Apple’s Delayed Siri AI
Apple announced Apple Intelligence in 2024, with a more capable version of Siri presented as one of its major features. The company promised an assistant that could understand more context, work across apps and help users complete tasks more effectively.
However, Apple repeatedly delayed the release of the improved Siri because of engineering problems and bugs. The delays also contributed to a $250 million settlement over claims about how Apple marketed AI capabilities associated with the iPhone 16.
The updated Siri eventually appeared in the iOS 27 beta in July 2026, with an English-language rollout beginning in October. Support for additional languages is expected to follow.
Siri is not a dead product, so it should not be described as a complete shutdown. It is included here as an example of an ambitious AI project that struggled to deliver its promised features on time.
The lesson is that companies must be careful about promising capabilities before they are ready. In the AI market, customers expect rapid progress, but reliability and trust remain important.
6. Microsoft’s Recall Feature
Microsoft introduced Recall in 2024 as an AI-powered feature for Windows computers. It was designed to take periodic screenshots of a user’s activity, allowing them to search their digital history and find something they had previously viewed.
The idea was to give users a kind of photographic memory for their computer. For example, someone could search for a document, website or message they had seen earlier without remembering exactly where it was located.
However, the feature triggered serious concerns about privacy and security. Critics warned that storing a searchable record of computer activity could expose sensitive information, including private messages, financial details and other personal data.
Microsoft delayed Recall for nearly a year while it worked on security and privacy protections. Even after the redesign, researchers continued raising questions about whether the stored information could be accessed in unsafe ways.
Recall has not simply disappeared, but the controversy shows that a useful AI feature can struggle when people do not trust how it handles their information. Businesses developing AI products need to consider privacy and security from the beginning, not only after customers raise concerns.
7. Notion Mail
Notion Mail launched in April 2025 as an AI-focused email product. It was designed to help people organise their inboxes, label messages and automate parts of their email workflow.
However, the company found that users were increasingly turning to separate AI agents to manage email tasks. As AI assistants became more capable of working across different services, the need for a dedicated AI email product became less clear.
Notion Mail was scheduled to shut down on September 22, 2026. Its planned closure shows how quickly the market for AI productivity tools can change.
A startup may build a product around one task, such as organising email, only for another AI assistant to handle that task alongside many others. When customers can get the same result without adopting a separate product, the standalone service becomes harder to sell.
8. Humane AI Pin
The Humane AI Pin was one of the most widely discussed AI hardware products. It was designed as a wearable device that allowed users to access AI features without relying on a traditional smartphone screen.
Humane raised about $230 million from investors and attracted significant attention before the product reached customers. The idea was to create a new way for people to interact with AI throughout the day.
However, the device struggled with performance and did not deliver the smooth experience many users expected. Humane also warned customers to stop using its charging case because of a possible battery fire risk.
The company shut down its AI Pin business in February 2025, and HP acquired most of its assets for $116 million.
The AI Pin’s story shows the risks of building an entirely new type of hardware around a technology that is still developing. Customers need more than an interesting idea. They need a product that works reliably, offers clear benefits and justifies its price.
9. Rabbit R1
Rabbit launched the R1 in 2024 as a small AI-powered device designed to carry out tasks for users. The company said the device could help people interact with digital services through a simpler interface.
The launch attracted strong attention, and Rabbit reported selling 100,000 units shortly after its announcement. However, early reviews described the product as unfinished, with unreliable performance and a limited number of useful integrations.
The R1 has not been completely abandoned. Rabbit has continued updating the device and positioning it as a tool that can control computers and carry out tasks using AI. The company has also announced another hardware project called Project Cyberdeck.
The R1 is therefore better described as a product that struggled to meet expectations rather than a confirmed business failure. Its story highlights the difference between generating excitement before launch and delivering a product people find useful every day.
10. Huxe
Huxe was an AI audio app developed by former members of Google’s NotebookLM team. It turned written information into conversational audio, allowing users to listen to AI-generated discussions instead of reading long documents.
The idea was attractive because many people prefer listening to information while travelling, exercising or completing other tasks. Huxe offered a way to turn written content into something similar to a podcast.
However, larger platforms began offering similar experiences to their existing audiences. Spotify, for example, continued expanding its AI-powered audio tools.
Huxe shut down in May 2026. Its closure shows how difficult it can be for a smaller startup to compete when a large platform introduces similar features to a service that already has millions of users.
11. Yupp
Yupp offered a free platform where users could compare answers from hundreds of AI models side by side. At its peak, the service allowed people to test more than 800 models, including systems from OpenAI, Google and Anthropic.
The platform also allowed users to vote on responses and earn cryptocurrency. This gave people a way to compare different AI systems instead of relying on a single model.
Despite the interesting idea, Yupp’s founders said the company did not achieve strong enough product-market fit to survive. Product-market fit means a product has found a group of customers who need it and are willing to keep using it.
Yupp shut down in March 2026. Its experience shows that offering many features does not guarantee success if the company cannot build a lasting business around them.
12. Figgs AI
Figgs AI allowed users to create and interact with custom AI characters for role-play and storytelling. The service operated from 2023 to 2024 and reportedly attracted more than one million users.
People could create characters with different personalities and use them in conversations. The product appealed to users who wanted a more creative and personal experience with AI.
However, its developers said that keeping the platform free became too expensive to sustain. They eventually shut it down.
Figgs AI shows that a large user base does not automatically pay the bills. If a company spends money serving users but cannot generate enough revenue, it may struggle even when people enjoy the product.
13. Character.ai
Character.ai became popular by allowing users to chat with custom AI characters, including characters inspired by celebrities and fictional personalities. Founded in 2021 by former Google researchers, the company attracted significant attention and raised $150 million at a reported $1 billion valuation in 2023.
In 2024, Google paid a reported $2.7 billion to license Character.ai’s technology and bring its founders back into Google’s orbit. The startup continued operating, but it later faced serious legal and business challenges.
Families filed lawsuits alleging that interactions with its chatbots contributed to teen suicides and self-harm. Disney also sent a legal notice over the use of its characters without permission. Character.ai restricted open-ended chats for users under 18 and agreed to settle several lawsuits alongside Google in early 2026.
In October 2026, the company filed for bankruptcy, citing legal, regulatory and operational challenges. Disney became the opening bidder, offering $20 million for a licence to some of Character.ai’s intellectual property.
However, bankruptcy does not necessarily mean the service has already shut down. Character.ai’s app remains available while the sale process continues, with a deal expected to be completed by late November or early December.
The case shows how legal risk, user safety and the cost of operating a consumer AI platform can affect a company’s future. It also highlights why AI businesses must consider the possible harm their products could cause, particularly when users are young or vulnerable.
14. Relay
Relay was an AI-powered workflow automation tool that helped users automate tasks involving email, messages and other work applications. It was designed as an alternative to services such as Zapier, allowing people to build workflows with the help of AI agents.
The company operated for about five years before shutting down in September 2026. One major challenge was that larger technology companies began adding similar automation features directly into their own products.
This made it harder for Relay to give customers a strong reason to pay for a separate service. If a customer could automate tasks using tools they already had, switching to another platform became less attractive.
Relay’s closure is a clear example of how difficult it can be for a small AI company to compete against larger platforms with existing customers and more resources. A startup may build a useful product, but it still needs a lasting advantage that competitors cannot easily copy.
What Can Entrepreneurs Learn From These AI Failures?
The AI graveyard offers important lessons for entrepreneurs, investors and anyone planning to build a technology business. The main lesson is that a good idea and advanced technology are not enough to guarantee success.
First, businesses must solve a real problem. A product may attract attention because it uses AI, but customers will not keep paying for it if it does not make their lives easier or help them achieve something valuable.
Second, companies need a clear way to earn money. Free products can attract many users, but they still require funding to pay for development, servers, staff and customer support. Entrepreneurs need to understand how their business will cover those costs before growth becomes too expensive.
Third, businesses must understand their competition. A small startup can lose customers quickly when a large technology company adds a similar feature to an existing product. Companies should look for a clear advantage, such as specialist knowledge, better service, unique data or a strong connection with a particular group of customers.
Fourth, customer trust matters. Products that handle personal information, interact with children or make decisions on behalf of users must take privacy, security and safety seriously. A useful product can still fail if people do not trust the company behind it.
Finally, companies should be willing to change direction. Sometimes, a standalone product works better as part of a larger service. In other cases, the best decision may be to stop investing in a project that is not meeting its goals.
The Bottom Line
The AI graveyard is a reminder that the rapid growth of artificial intelligence does not guarantee success for every company or product. Some projects have shut down because they could not cover their costs, while others lost their place in the market when larger platforms introduced similar features.
The examples also show that not every setback is the same. Sora, Humane AI Pin, Huxe, Yupp and Figgs AI have shut down, while products such as Rabbit R1 and Apple’s Siri AI have continued to develop despite earlier problems. OpenAI has also discontinued standalone products while moving useful features into ChatGPT.
For entrepreneurs, the message is simple. Do not build a business only because a technology is popular. Start with a problem people genuinely want solved, create a product that works well, understand how you will earn money and keep listening to customers.
AI will continue to create new opportunities, but the companies that survive will need more than impressive technology. They will need good business decisions, strong customer trust and a clear reason for people to keep using their products.