Nvidia CEO Jensen Huang Criticizes US Chip Export Controls as "Loser's Mentality"

Gate News message, April 20 — Nvidia CEO Jensen Huang recently appeared on a prominent U.S. podcast to discuss American chip export restrictions, delivering sharp criticism of the policy during a heated exchange with host Dwarkesh Patel.

When asked whether the U.S. should relax chip export controls, Huang responded with unusual emotion, calling the rationale behind the restrictions “extremely stupid” and a classic “loser’s mentality.” He argued that China already possesses substantial computing resources and, despite Nvidia’s chips maintaining advantages in performance and efficiency, Beijing can develop advanced AI models through scaling compute capacity. Therefore, Huang contended, excluding Nvidia from the Chinese market would not prevent China’s AI development.

Huang emphasized the importance of a unified global technology ecosystem: “We want to ensure that all global AI developers build on the American technology stack, so that AI contributions and progress, especially open-source developments, flow back to the American ecosystem. Creating two separate ecosystems—one running only on foreign technology stacks and another running on American stacks as a closed system—would be extremely unwise and detrimental to the U.S.” He rejected the notion that competing in China would lead to market loss, stating, “Our market share is growing, not declining. You cannot claim that participating in the Chinese market means losing it; that logic doesn’t hold… you’re not talking to a loser.”

Huang further explained that AI chip ecosystems are not easily disrupted like automotive or consumer electronics markets: “Computing ecosystems are complex. This is why x86 architecture has endured and why ARM maintains high stickiness. These ecosystems are difficult to replace, and migration costs are prohibitively high. Most enterprises won’t bear them.” He stressed that Nvidia’s responsibility is to continuously cultivate its ecosystem and drive technological progress to win in market competition.

When asked about Anthropic’s use of Google’s TPU and Amazon’s Trainium chips instead of Nvidia hardware, Huang acknowledged a strategic misstep: during Anthropic’s early days, Nvidia lacked the financial capacity to offer multi-billion-dollar equity packages like Google and AWS. He noted that venture capital firms would not invest $5-10 billion betting on a lab to become Anthropic, leaving startups with few options beyond aligning with tech giants. To avoid repeating this mistake, Nvidia subsequently invested in both Anthropic and OpenAI. Huang characterized Anthropic’s chip choices as “an exception, not a trend,” adding, “Without Anthropic, TPU and Trainium would have zero growth—100% of their adoption comes from Anthropic.”