How US Chip Sanctions Are Building a Fortress for Huawei

In a rare interview last week, Huawei founder Ren Zhengfei made a startling admission: his company’s advanced AI chips “still lag behind the US by a generation.” Yet in the same breath, he insisted there was “no need to worry” about US sanctions. This apparent contradiction gets to the heart of China’s high-stakes AI strategy, a game that isn’t about building the single best chip in the world, but about creating an ecosystem that is good enough to win on its home turf. For China's tech giants and the global semiconductor industry, this marks a critical tipping point.

What's behind Ren Zhengfei's mixed message?

On a chip-for-chip basis, Huawei's Ascend processors are not on par with Nvidia's top-tier offerings. Chinese tech giants know this. There’s a reason companies like Alibaba, ByteDance, and Tencent have reportedly spent billions of dollars stockpiling Nvidia's export-compliant chips. According to some reports, Huawei’s chips have a tendency to overheat, and their software ecosystem is nowhere near as mature as Nvidia's CUDA platform, which has been the industry standard for nearly two decades.

The concern, however, isn't about a single chip. It's about what happens when thousands of them are "clustered" together. As Ren explained, by bundling its Ascend chips into massive, interconnected systems, Huawei can achieve performance that is competitive for many large-scale AI tasks. As detailed in an WSJ report, Huawei's CloudMatrix 384 system connects 384 Ascend 910C chips together. Analysts noted that while the individual Chinese chips were less powerful and the system consumed more power, the sheer scale of the cluster could offset those deficiencies and in some cases outperform an equivalent Nvidia rack system.

Why is Nvidia’s CEO praising Huawei?

Counterintuitively, one of Huawei's biggest cheerleaders has been Nvidia CEO Jensen Huang, who recently called the company "one of the most formidable technology companies in the world." This praise, however, is best understood through the lens of business and geopolitics. US export controls, designed to slow China's progress, have effectively locked Nvidia out of what Huang estimates will be a $50 billion AI market in the coming years.

The restrictions have cost Nvidia billions in lost sales and forced it to create a series of progressively weaker, export-compliant chips for the Chinese market. Huang’s praise for Huawei is a strategic message to Washington: by restricting US firms, the government is not stopping AI in China, but is instead creating a protected market where a formidable domestic competitor can flourish without facing its strongest global rival.

How are US sanctions creating a market for Huawei?

US export controls have created a unique competitive dynamic. Huawei’s Ascend 910B/C chips don't have to compete with Nvidia’s best-in-the-world Blackwell GPUs. Instead, they only need to be a viable alternative to Nvidia's downgraded H20 chips, or the even more restricted Blackwell variants planned for China.

With Washington tightening the screws—even warning that simply using Huawei's Ascend chips anywhere in the world could violate export controls—Chinese companies face a clear choice. They can rely on a constrained supply of watered-down US chips that could be further restricted at any moment, or they can invest in a homegrown platform from Huawei that has the full backing of the Chinese government. For state-owned enterprises and government data centers, the choice is increasingly clear, providing Huawei with a foundational customer base to fund its R&D and scale production.

What's next for China's AI self-sufficiency?

The real battle for China's AI future is about the software ecosystem. Currently, most of China’s leading AI models, including those from the highly efficient startup DeepSeek, were built using Nvidia's hardware. Developers are trained on it, and the tools are mature.

The "tipping point" will be the moment a major Chinese AI lab, pushed by political incentives or supply realities, decides to build its next flagship model exclusively on Huawei’s platform. Such a move would have a cascading effect, forcing other developers, toolmakers, and startups to align with the domestic standard to remain competitive. It would mark the true birth of a self-sufficient Chinese AI stack, decoupled from Silicon Valley. While Huawei's chips may still be a "generation behind" today, the combination of clustering, a protected market, and immense political will means they are on the cusp of becoming good enough to dominate their own digital destiny.

The Reference Shelf

Why Is Huawei Downplaying Its Chips? (Bloomberg)

Why Huawei Can Only Make 200,000 AI Chips a Year (ARPU)

What Is at Stake in Taiwan’s Ban on Huawei and SMIC? (ARPU)