DeepSeek V4 model to run on Huawei chips, testing China's AI self-sufficiency ambitions

DeepSeek Shifts to Huawei Chips for Next-Generation AI Model

The global tech industry has been waiting for weeks to see DeepSeek's next major launch, and reports now confirm that the Chinese AI startup's V4 model will run on Huawei chips rather than US-made processors. According to The Information, Chinese tech giants including Alibaba, ByteDance, and Tencent have placed bulk orders for Huawei's latest chips totaling hundreds of thousands of units in preparation for the DeepSeek V4 model launch1

1

5

5

. This decision represents a decisive pivot away from Nvidia hardware and marks a critical test of China's AI self-sufficiency trajectory.

DeepSeek first captured global attention in January 2025 with its R1 deep-reasoning chatbot, which President Donald Trump called a "wake-up call" for American firms. Built on the V3 model released in December 2024, R1's low-cost, high-performance capabilities triggered a tech stock selloff that wiped out $600 billion from Nvidia's market value in a single day3

3

. Now, the upcoming V4 model—expected to be multimodal, generating text, pictures, and video—could deliver another shock to US tech valuations.

Breaking Industry Norms to Embrace Domestic Hardware

In a departure from standard industry practice, DeepSeek did not provide early versions of its flagship model to US chipmakers like Nvidia and AMD for performance optimization3

3

. Instead, the company has spent recent months working directly with Huawei and Cambricon Technologies to rewrite pieces of the model's underlying code and conduct testing5

5

. This collaboration ensures that V4 runs efficiently on domestically produced chips, reducing reliance on foreign hardware that has become increasingly difficult to access.

The shift comes as US export restrictions, citing national security concerns, have blocked China's access to top-tier AI processors from Nvidia. Wei Sun, principal AI analyst at Counterpoint Research, emphasizes the significance: "It's important to know because at one level, it is a signal of China's AI self-sufficiency trajectory"1

1

. However, transitioning from Nvidia chips to Huawei alternatives requires "substantial re-engineering" that can slow development cycles and introduce performance trade-offs, especially for a model expected to be state-of-the-art4

4

.

Geopolitical Implications and Strategic Independence

The decision behind the DeepSeek V4 model to use Huawei chips extends beyond technical considerations into deeply strategic territory. By building a foundational AI model optimized for domestic semiconductors, DeepSeek is effectively future-proofing itself against further US export restrictions2

2

. The surge in chip orders from Chinese tech giants signals strong industry confidence in Huawei's AI hardware capabilities and reflects a broader alignment with domestically produced technology across China's AI ecosystem.

Stephen Wu, founder of Carthage Capital fund, predicts that V4 "will not just be a software update; it will be a highly capable, open-source model that handles massive context windows at a fraction of the cost"1

1

. If DeepSeek has successfully trained V4 entirely on Huawei silicon, it would signal a material shift in the geopolitical tech landscape4

4

. This development could accelerate semiconductor independence for China while gradually eroding Nvidia's market share in the region.

Questions About Training Methods and Chip Smuggling

Speculation surrounds not just which chips will run V4, but which processors were used to train it. AI training requires substantially more computing power than inference—the process of generating responses to user queries. Some reports allege that DeepSeek may have skirted the export ban to train V4 using thousands of Nvidia's top-end Blackwell chips, dismantled in third countries and smuggled to China1

1

4

4

. Nvidia told The Information it had not seen evidence of this and called such smuggling "farfetched."

Meanwhile, other Chinese AI companies are demonstrating progress with domestic chips. Zhipu unveiled an image generator in January that was entirely trained on Huawei chips, while Alibaba announced plans to open a new data center for AI training and inference in southern China, powered by 10,000 of its own chips and operated by China Telecom1

1

. These developments suggest that China's AI ecosystem is actively building infrastructure independent of US AI processors.

What's at Stake for US Chipmakers and Global AI Competition

The implications for US chipmakers are significant. Nvidia and AMD face losing not just market access but also the early optimization opportunities that are critical for fine-tuning chip performance with next-generation AI models2

2

. If Chinese firms successfully optimize their models for local chips, they may no longer need US hardware at scale, fundamentally altering the competitive balance in artificial intelligence.

DeepSeek is reportedly developing two additional V4 variants, each optimized for different capabilities and designed specifically for Chinese chips5

5

. This multi-model strategy could give China's AI a diversified portfolio capable of competing globally. The V4 launch, expected within weeks, will test whether Huawei can truly match Nvidia in power and scale—and whether DeepSeek can maintain its reputation at the frontier of AI technology while navigating the constraints of the export ban. For investors questioning whether US AI firms need to spend billions on computing power, V4's performance could provide answers that reshape market expectations and trigger another wave of valuation reassessments across the tech sector.