The AI Graveyard: Tracking 100 Failed and Folded Tools

ToolDirectory.AI tracks 100 AI products that have shut down, been acquired, or abandoned their domains. The data highlights a significant spike in 2026, where 88 tools ceased independent operations. This 'graveyard' serves as a stark reminder of the high failure and consolidation rates within the burgeoning AI industry.

Key Points

• The AI industry saw a massive spike in product exits in 2026, accounting for 88% of the graveyard's entries.
• Acquisitions are a leading cause of tool disappearance, with products frequently being absorbed and redirected to parent company platforms.
• A high number of tools (34%) simply let their domains lapse, indicating many AI startups are abandoned without formal announcements.
• The failures and consolidations affect nearly every sub-niche of AI, from enterprise infrastructure to niche creative plugins.

Sentiment

The Hacker News community is overwhelmingly dismissive of the article itself, with near-universal agreement that the list is poorly researched, factually inaccurate, and likely AI-generated content. While many commenters sympathize with the general premise that AI products are failing at a high rate, they view this particular list as undermining that very narrative through its sloppiness. The irony of an AI-generated site tracking AI failures is not lost on the community.

In Agreement

• Many AI tools are indeed failing because they are just thin wrappers around someone else's AI with minimal added value, making them unsustainable businesses
• The AI market is experiencing a consolidation pattern similar to previous tech bubbles like crypto and the dot-com era
• AI product unreliability and liability concerns are legitimate reasons for product failures, as one commenter demonstrated by shutting down their own AI medical records product
• When OpenAI and Anthropic's addressable market is any software business, margins for wrapper products trend toward zero
• Getting attention for AI products has become harder even as building them got easier, contributing to high failure rates

Opposed

• The list is fundamentally flawed because it includes many products that are demonstrably still alive and active, including Streamlit, Langfuse, Weights & Biases, and Reclaim
• Counting acquisitions as deaths is misleading — acquisition is usually a success outcome, not a failure, and many acquired products continue operating
• The list itself appears to be AI-generated slop from a spam site with a history of low-quality submissions to HN
• Listing Bing AI as 'acquired by Microsoft' is absurd since Bing was always a Microsoft product, undermining the entire dataset's credibility
• The list is missing genuinely significant dead AI products like Sora, Bard, and the Dia Browser while including active ones
• Domain lapsing does not necessarily mean a product is dead — some companies simply moved to different domains