Nvidia CEO: China Already Possesses Compute Power for Advanced AI Training—Raising Global Security Stakes

Nvidia CEO Jensen Huang has sounded an alarm about China's existing computational capacity to develop advanced AI models comparable to Anthropic's Claude Mythos, potentially threatening global cybersecurity infrastructure.

During an appearance on the Dwarkesh Patel podcast Wednesday, Huang was directly asked whether China's access to chips capable of training a model like Claude Mythos—which exhibits cyberoffensive capabilities—poses a national security risk to the US. His answer was unambiguous: China already has what it needs.

"The amount of capacity and the type of compute it was trained on is abundantly available in China," Huang explained. "You just have to first realize that chips exist in China."

The Scale of Chinese Computing Infrastructure

Huang emphasized just how vast China's computational footprint is. "They have datacenters that are sitting completely empty, fully powered. They have ghost cities, they have ghost datacenters too. They have so much infrastructure capacity. If they wanted to, they could just gang up more chips."

The Nvidia CEO pointed out that China manufactures 60% of the world's mainstream chips, employs some of the planet's best computer scientists, and hosts approximately 50% of the world's AI researchers—all backed by abundant energy resources. These factors combine to create an environment where developing models with Mythos-level capabilities becomes a matter of allocation, not limitation.

Why Mythos Matters as a Threat Vector

Anthropic restricted access to Claude Mythos in April after discovering thousands of software vulnerabilities across major operating systems and browsers. The AI Security Institute (AISI) subsequently evaluated Mythos on April 13, confirming the model could "execute multi-stage attacks on vulnerable networks and discover and exploit vulnerabilities autonomously"—tasks requiring days of human professional effort.

Anthropic's research revealed that 99% of vulnerabilities discovered by Mythos remain unpatched, creating a dangerous attack surface. Financial institutions face particular risk, as many banks still operate on decades-old software infrastructure vulnerable to AI-enhanced exploitation.

In November, Anthropic disclosed that a Chinese state-sponsored group had previously manipulated its Claude Code tool, attempting to infiltrate approximately 30 global targets with limited success. A production-grade Mythos equivalent would significantly amplify such capabilities.

A Call for Strategic Engagement Over Antagonism

Rather than advocating for punitive measures, Huang surprisingly advocated for dialogue. "Victimizing them, turning them into an enemy, likely isn't the best answer," he stated. "They are an adversary. We want the United States to win. But I think having a dialogue and having research dialogue is probably the safest thing to do."

This perspective contrasts with statements from US Treasury Secretary Scott Bessent, who on Tuesday called Claude Mythos a "step function change in abilities and learning capabilities" positioning America ahead in the AI race against China.

The crypto and blockchain ecosystem faces indirect but significant exposure here—advanced AI models capable of discovering zero-day exploits threaten the infrastructure securing digital assets, exchange platforms, and cryptocurrency custody solutions.