Jensen Huang is worried: Tariff war will create a vaccum

Nvidia warns that new US restrictions on AI hardware exports may inadvertently strengthen Huawei, empowering China in the AI race. As US policy cuts off access to Nvidia’s ecosystem, Chinese companies may shift to Huawei's developing CANN platform. This could solidify China's position in the global AI arena, challenging US dominance.

It is evidently known in tech that power does not only belong to those who build the best hardware. It belongs to those who set the rules. Nvidia, the undisputed emperor of the AI age, understands this better than anyone. And today, it’s sounding an alarm: the new US restrictions on AI hardware exports could end up empowering China’s Huawei, not containing it.

At the heart of the drama is the AI Diffusion Rule, a complex framework that kicks in on May 15. It is supposed to control how the world’s most powerful AI chips like Nvidia’s prized H100s are distributed globally. The US strategy is clear: lock out rivals like China and Russia. But Nvidia sees a different battlefield emerging. Not one where America remains supreme by holding technology back, but one where it cedes the future to its competitors.

Jensen Huang, Nvidia’s charismatic CEO, minced no words. “America cannot lead by slowing down,” he warned US lawmakers. If the US retreats, others specifically China will rush into the vacuum. In technology, momentum is everything. If American companies lose their dominant ecosystems today, they may never regain them.

Right now, Nvidia’s CUDA ecosystem defines the world of AI. Six million developers across continents have built their code, businesses, and dreams around it. China’s tech titans like Alibaba, Tencent and ByteDance run on Nvidia’s hardware. Europe’s startups and researchers, too, rely on it. Nvidia’s sprawling dominance isn’t just about selling chips. It’s about shaping the very architecture of modern intelligence.

But behind China’s firewall, a different reality is taking shape.

Companies like Biren Technology, Moore Threads, and Innosilicon are racing to build domestic alternatives. None pose an immediate threat yet; their hardware is capable, but nowhere close to matching Nvidia’s platforms. The real wildcard is Huawei. Quietly but relentlessly, it has been building its own ecosystem, CANN, a CUDA rival that integrates deeply with its Ascend AI processors.

Huawei’s CANN is rough around the edges. Developers complain of bugs, documentation gaps, and frustrating performance inconsistencies. But ecosystems are stubborn things. Given enough pressure, and enough billions in R&D investment, they evolve. They mature. They capture mindshare.

And that pressure is now being applied, courtesy of US policy.

By cutting Chinese firms off from Nvidia’s ecosystem, the US is forcing them into Huawei’s arms. Companies that would have gladly stayed with CUDA will have no choice but to migrate, investing the engineering resources needed to make CANN viable. Over time, what is today a clumsy alternative could become a true competitor.

The analogy isn’t hypothetical. The world saw this once before with 5G. Faced with US sanctions, Huawei didn’t collapse; it dug in and built its own 5G empire, exporting it across the globe. American firms are still struggling to catch up.

In the AI race, the stakes are even higher.

Whoever sets the standards for AI hardware and software will wield immense geopolitical power, not just economic advantage. They will decide what principles govern AI development, who gets access to advanced models, and how future applications are regulated. It will be about far more than selling more GPUs or cloud hours.

Nvidia is not worried about losing today’s market. It’s worried about losing the architecture of tomorrow. As Jensen Huang put it bluntly, “If we lose that ecosystem to our competitors, it will be almost impossible to get it back.”

Already, countries outside the Tier 1 US allies list, in Europe, the Middle East, and Asia, are rethinking their dependencies. If Huawei’s hardware improves and comes cheaper (and less politically entangled), it could quickly become the go-to alternative. Especially for nations reluctant to bet their entire AI future on the whims of Washington’s export controls.

The irony, of course, is brutal: In trying to contain China, America may end up sparking its AI self-sufficiency at record speed.

Huawei, for its part, is pouring billions into its AI ambitions. And unlike Nvidia, it enjoys the full, unflinching support of the Chinese state. Every GPU it sells feeds a feedback loop of reinvestment, talent recruitment, and ecosystem hardening.

When the dust settles, Huawei and the Chinese tech sector at large could emerge not just as participants in the AI revolution, but as the ones setting its rules.

If Nvidia’s warning falls on deaf ears, the US might realize too late that it lost not because its technology was inferior, but because it tried to slow the game instead of playing faster and smarter.

In tech, leadership is never about who blocks innovation best. It’s about who enables it faster, wider, deeper.

And that is a lesson America cannot afford to forget.