Nvidia CEO Jensen Huang has issued a stark warning that China possesses the massive computing power and data center infrastructure necessary to train an artificial intelligence model rivaling Anthropic's advanced Claude Mythos, raising significant concerns about global cybersecurity. Speaking on the Dwarkesh Patel podcast on Wednesday, Huang stated that the hardware capacity used to train the Mythos model is "abundantly available in China."
Huang revealed that China has "enormous" compute resources, including "ghost datacenters" that sit completely empty but fully powered. "They have datacenters that are sitting completely empty, fully powered. You know, they have ghost cities, they have ghost datacenters too," Huang said. "They have so much infrastructure capacity. If they wanted to, they [could] just gang up more chips."
The urgency of this warning stems from the specific cyberoffensive capabilities discovered in Anthropic's Claude Mythos model. In April, Anthropic limited access to Mythos after it identified thousands of software vulnerabilities across major operating systems and browsers. The AI Security Institute (AISI) evaluated Mythos on April 13, finding the model could "execute multi-stage attacks on vulnerable networks and discover and exploit vulnerabilities autonomously," tasks that would take human professionals days of work.
Huang emphasized China's technological advantages, noting the country manufactures 60% of the world's mainstream chips, has some of the best computer scientists, hosts 50% of the world's AI researchers, and has abundant energy resources. Despite identifying China as an adversary and expressing a desire for the United States to win the technological race, Huang advocated for dialogue rather than conflict. "Victimizing them, turning them into an enemy, likely isn't the best answer," he said. "I think having a dialogue and having research dialogue is probably the safest thing to do."
This warning comes as U.S. Treasury Secretary Scott Bessent hailed Mythos as a "step function change" in abilities that would keep America ahead of China in the AI race. Meanwhile, Reuters reported that financial institutions remain particularly vulnerable to AI-boosted hacks because they rely on decades-old software. The concerns are grounded in recent history—Anthropic reported in November that a "Chinese state-sponsored group" manipulated its Claude Code tool in attempts to infiltrate about 30 global targets, succeeding in a small number of cases.
