The AI Graveyard: Scaling as a Euphemism for Burning Cash
AI-generated image · US National Wire
Opinion: From the collapse of Character.ai to the fire-sale of Humane, the industry's obsession with 'scaling' is masking a fundamental failure of unit economics.
For years, the Silicon Valley pitch has been the same: scale now, figure out the monetization later. But as TechCrunch first reported in its audit of the "AI graveyard," it is becoming clear that many of these darlings weren't scaling toward a solution—they were simply burning VC cash on products that never actually worked.
**Opinion: The Mirage of Growth**
The wreckage reveals a systemic failure to translate hype into sustainable business models. Character.ai, the chatbot platform, filed for bankruptcy just five years after launching. In the hardware sector, the delusion reached its peak with the Humane AI Pin. After raising $230 million to build a smartphone replacement, the product was plagued by performance issues and battery fire risks. By February 2025, the business was shut down, and HP acquired its assets for $116 million. Similarly, the Rabbit R1 arrived with massive hype and 100,000 units sold, yet early reviews described an unfinished device with unreliable performance.
**The 'Platform' Trap**
Many "AI-first" software companies discovered they were building features, not businesses. Relay, a five-year-old automation tool, shut down in August as giants like Google and OpenAI integrated similar automation into their own ecosystems. Even Notion struggled; Notion Mail, launched in April 2025, is scheduled for shutdown on September 22 after users preferred separate AI agents.
This is a corporate contagion as well. S&P Global Market Intelligence reports that approximately 42% of AI initiatives are abandoned by their corporate parents due to technical hurdles, funding gaps, or weak demand.
**Even the Titans are Stumbling**
OpenAI is not immune. TechCrunch notes the company shut down its video-sharing platform Sora in March 2026 due to high costs and retention issues, along with the AI-powered web browser ChatGPT Atlas and the web-agent Operator. Even a July 9 attempt to turn ChatGPT into a "super app" with "Chat," "Codex," and "Work" modes was rolled back almost immediately after users found the interface cluttered.
Apple and Microsoft have faced similar pitfalls. Apple's promise of a context-aware Siri led to repeated delays and a $250 million settlement regarding the marketing of the iPhone 16's AI. The new Siri only hit the iOS 27 beta in July. Meanwhile, Microsoft's Recall feature, pitched as a "photographic memory" for PCs, became a privacy nightmare; a cybersecurity researcher recently proved that data captured by Recall could still be extracted despite a redesign.
**The Bottom Line**
When we audit the AI graveyard, we see a recurring theme: a total disregard for the P&L in favor of the pitch deck. Scaling is only a virtue if the underlying unit economics are sound. If you are spending more to acquire and serve a customer than that customer is worth, you aren't scaling—you're just accelerating your descent into the graveyard.