Nvidia CEO Jensen Huang has sounded an alarm about a potential shift in the global artificial intelligence landscape. If Chinese AI startup DeepSeek optimizes its upcoming V4 foundation model on Huawei Technologies chips rather than American processors, Huang warns it would represent "a horrible outcome" for the United States. The concern centers on how standardizing around Chinese hardware and software could fundamentally alter which country leads in AI development .

What's at Stake in the DeepSeek-Huawei Partnership?

The stakes are remarkably high. Huang explained that if "future AI models are optimized in a very different way than the American tech stack," and as "AI diffuses out into the rest of the world" with Chinese standards and technology, China "will become superior to" the United States . This isn't merely about which company sells more chips; it's about establishing the technical foundation that the entire world's AI systems will build upon for years to come.

Huang

DeepSeek's V4 model, expected to launch later in April 2026, is rumored to run on Huawei's latest Ascend 950PR processor, according to reporting from The Information. This would mark a significant departure from the company's previous approach. DeepSeek's V3 model, launched in late 2024, was trained on 2,048 Nvidia H800 graphics processing units (GPUs), which are older processors based on Hopper technology . The shift to Huawei chips would represent the first major Chinese AI model optimized for homegrown hardware rather than American technology.

How Does the Export Control Puzzle Fit Into This?

The technical decision about which chips to use is inseparable from U.S. export controls on advanced semiconductors. The H800 GPU was specifically designed as a product for the China market to circumvent export restrictions that banned the sale of more powerful chips to China in 2023 . However, regulatory restrictions have eased recently under the Trump administration, with Nvidia restarting production of the H200, a more powerful chip, to be sold in China, as Huang noted in March .

This creates a complex situation. Even as American export controls have loosened, the incentive for Chinese companies to develop independence from American technology remains strong. A Reuters report last month suggested that DeepSeek's V4 model had been trained on Nvidia's Blackwell chips, which would violate U.S. export controls if true . Whether DeepSeek ultimately chooses Huawei or American chips, the company faces pressure from multiple directions: regulatory compliance, technological capability, and geopolitical considerations.

"If future AI models are optimized in a very different way than the American tech stack, and as AI diffuses out into the rest of the world with Chinese standards and technology, China will become superior to the United States," warned Jensen Huang.

Jensen Huang, CEO at Nvidia

Steps to Understanding the Broader Implications

• Technology Stack Standardization: When a major AI model optimizes around specific hardware, it creates a gravitational pull for other developers to use the same chips and software tools, establishing industry standards that are difficult to change later.
• Energy and Research Advantages: Huang noted that China benefits from "abundant" energy resources and a large pool of AI researchers, giving the country structural advantages if it can operate independently from American technology.
• Global AI Diffusion: As artificial intelligence spreads beyond the United States and China to other countries, those nations may adopt whichever technical standard is most accessible and cost-effective, potentially locking in Chinese technological influence worldwide.

The timing of Huang's warning is significant. His comments came on the Dwarkesh Podcast just days before DeepSeek's expected V4 launch, suggesting that the Nvidia CEO views this moment as a critical juncture for American technological leadership. The conversation reflects growing anxiety in Silicon Valley about whether the U.S. can maintain its dominance in AI infrastructure as Chinese competitors develop homegrown alternatives.

What makes this situation particularly complex is that it's not simply about corporate competition between Nvidia and Huawei. It's about which nation's technological ecosystem becomes the foundation for artificial intelligence systems used globally. If DeepSeek successfully optimizes advanced AI models on Huawei chips, it would demonstrate that Chinese hardware can compete with American processors for cutting-edge AI applications. That proof of concept could accelerate adoption of Chinese chips in other countries, gradually shifting the balance of technological power.

The outcome remains uncertain. DeepSeek's choice about which chips to use for V4 will send a powerful signal about the future direction of AI development. For now, Huang's warning serves as a reminder that the competition for AI leadership extends far beyond research papers and product announcements. It's fundamentally about controlling the hardware and software standards that will shape artificial intelligence for decades to come.