NVIDIA CEO Jensen Huang has publicly warned that if DeepSeek optimizes its new AI models to run on Huawei chips rather than American hardware, it would represent "a horrible outcome" for the United States. The warning signals a fundamental shift in how AI development could unfold globally, moving away from the American technology stack that has underpinned U.S. dominance in artificial intelligence for the past decade .

What Is CUDA and Why Does It Matter So Much?

CUDA is NVIDIA's software framework that has become the default development environment for artificial intelligence researchers and companies worldwide. When AI engineers write code, they typically write it for CUDA. When startups build AI products, they build them on CUDA. When governments invest in AI infrastructure, they purchase NVIDIA GPUs because that is what the software requires .

This software-hardware relationship creates what experts call a "moat," or competitive advantage. Even if other countries develop chips that could theoretically compete with NVIDIA's hardware, they still need NVIDIA's software ecosystem to make those chips useful. NVIDIA's dominance rests not just on making the fastest chips, but on CUDA's position as the foundation of AI development globally .

How Is DeepSeek Breaking Free From NVIDIA's Ecosystem?

DeepSeek, China's most capable AI laboratory, is preparing to launch V4, a multimodal foundation model expected later this month that will run on Huawei's latest Ascend 950PR processor. More significantly, the company has spent months rewriting its core code to work with Huawei's CANN framework, moving away from the CUDA ecosystem entirely .

This migration is the key threat Huang is warning about. DeepSeek's V3 model, launched in late 2024, was trained on 2,048 NVIDIA H800 GPUs and demonstrated that the company can produce frontier-competitive AI models with fewer resources than American rivals. Its R1 reasoning model matched or exceeded the performance of models that cost orders of magnitude more to train. If V4 performs well on Huawei silicon without any NVIDIA involvement, it validates an alternative path for AI development that does not depend on NVIDIA at any point in the supply chain .

What Are the Key Factors Behind Huang's Concern?

Huang's warning is not primarily about current hardware performance gaps. On raw performance, Huawei's chips are not competitive with NVIDIA's best. The Ascend 910C, the predecessor to the 950PR, delivers roughly 60% of the inference performance of NVIDIA's H100, a chip that is itself two generations behind NVIDIA's current best. American chips are approximately five times more powerful than their Chinese equivalents today, and that gap is projected to widen to 17 times by 2027 .

However, Huang acknowledged on the Dwarkesh Podcast that even if China had inferior chips, it could still catch up with the United States in AI development given its abundant energy resources and large pool of AI researchers. The implication is that raw hardware performance is only one variable in the equation. Software optimization, researcher talent, and energy availability can compensate for silicon disadvantages .

• Hardware Performance Gap: American chips are currently five times more powerful than Chinese equivalents, with projections showing this gap widening to 17 times by 2027
• Software Ecosystem Migration: DeepSeek's move from CUDA to Huawei's CANN framework breaks the dependency on NVIDIA's software that has kept Chinese labs tied to American technology
• Geopolitical Implications: If DeepSeek proves competitive models can be built without NVIDIA, the argument for maintaining export controls weakens and the assumptions underlying U.S. AI policy come under pressure
• Inference Versus Training: While Huawei chips struggled with training AI models reliably, they appear adequate for inference, the phase where commercial value is generated and where users interact with models

How to Understand the Export Control Paradox?

The situation exposes a tension at the center of American chip export policy. NVIDIA restarted production of the H200, a more powerful chip, for sale in China, as CEO Jensen Huang confirmed in March. However, China has been blocking H200 imports to protect Huawei's domestic chip business, and NVIDIA's Chief Financial Officer has stated the company has recorded no revenue from China H200 sales .

This creates a paradox: the controls designed to limit China's AI capabilities are instead accelerating the development of a Chinese alternative. DeepSeek's experience with its R2 model illustrates both the promise and the limits of the Huawei path. R2 was repeatedly delayed because of training failures on Huawei hardware. Chinese authorities had urged DeepSeek to train on domestic chips, but the company encountered stability issues that forced it to revert to NVIDIA GPUs for training while using Huawei chips only for inference .

"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 US," warned Jensen Huang.

Jensen Huang, CEO at NVIDIA

What Does This Mean for NVIDIA's Future?

The stakes are concrete when considering NVIDIA's scale. The company's market capitalization exceeds 3 trillion dollars. Its data center revenue grew 93% year over year in its most recent quarter. Its chips power the training runs for virtually every major AI model outside China .

If the most capable Chinese AI lab demonstrates that competitive models can be built without NVIDIA, several consequences could follow. The argument for maintaining export controls weakens. The argument for buying NVIDIA weakens. The geopolitical assumptions that have shaped AI policy for the past three years come under pressure. However, this does not mean Huawei is about to overtake NVIDIA. The performance gap is large and growing. The R2 training failures demonstrate that Chinese hardware is not yet ready for the most demanding AI workloads .

Huang is not warning about today's capabilities but about a trajectory in which DeepSeek proves the concept, other labs follow, and the CUDA moat that has made NVIDIA the most valuable company in the AI supply chain begins to erode. The fact that the CEO of NVIDIA is making this argument publicly suggests he believes the risk is no longer theoretical. DeepSeek's V4 launch will be the first major test of whether a multimodal foundation model can run competitively on Huawei silicon without NVIDIA involvement .