Hackers Hunt for AI’s Most Valuable Commodity: Computing Power

A facts-only brief, written by a local AI model from this article in The Wall Street Journal, November 10, 2026. It may contain mistakes, and it is not the article: the full story is at wsj.com.

Hackers breached cloud-computing platform Railway in mid-October 2026, stealing over $1.1 million in assets including $371,000 in cryptocurrency and $750,000 in fraudulent AI service charges. Railway stated that attackers compromised up to 100,000 customer accounts to steal API keys, which were then used to access models from Amazon, OpenAI, Anthropic, and Google. This incident is part of a broader wave of cyberattacks targeting AI developer platforms, where criminals exploit stolen credentials to generate illicit compute revenue.

Background

• Researchers at Google’s cloud-security division, Wiz, report that stolen compute is often sold to “transfer stations” that provide access to US AI models for users in restricted countries like China and Russia.
• Chainalysis, a blockchain investigations firm, identified more than 700 such services online, noting that the top five have collectively generated nearly $6 million in cryptocurrency.
• Zeabur, another AI infrastructure company, permanently shut down following a similar hack that resulted in $100,000 in unauthorized charges, with its founder comparing the theft of API keys to hacking a bank.
• Google stated that total losses from “LLM jacking” can reach $1 million per organization, while Vectoral’s founder noted that these illicit sites receive tens of millions of monthly visitors.

Read the full article at wsj.com

More from Tuesday, November 10: