What’s behind the US move to limit Huawei’s AI chip production in 2025?

The United States has claimed that Chinese tech giant Huawei will produce no more than 200,000 advanced AI chips in 2025, significantly below the country’s risingdemand.

But officials warn this could lull Washington into a false sense of security, as China accelerates its semiconductor ambitions and chips away at American dominance.

As quoted by Reuters, US Commerce Under Secretary Jeffrey Kessler told a congressional subcommittee on Thursday that Huawei’s Ascend AI processors—designed to rival Nvidia’s chips—will mostly stay within China due to sweeping export restrictions. But he cautioned lawmakers against complacency.

“We shouldn’t take too much comfort in the fact that China’s production of these advanced chips is relatively small,” Kessler told the House Foreign Affairs Committee. “China is catching up quickly.”

Huawei’s AI chip ambitions face a production ceiling

The US has spent the past five years building a tech barrier around Chinese firms—blocking access to advanced lithography equipment, semiconductor IP, and chipmaking software. Huawei, under American sanctions since 2019, remains the centrepiece of this strategy.

At the hearing, Kessler said most of Huawei’s projected 200,000-chip output this year would remain within China’s borders.

As quoted by Reuters, he added that the figure is a “ceiling, not a floor”—reflecting Huawei’s constrained access to chip fabrication tools and high-end packaging equipment.

China’s AI chip gap: Scale vs sophistication

According to TechInsights, China’s domestic AI accelerator demand crossed 1.5 million units in 2024—meaning Huawei’s 2025 output would meet less than 15 per cent of market needs. In contrast, OpenAI’s Stargate facility in Texas is expected to house over 500,000 Nvidia H100-class chips, more than double Huawei’s annual production.

Huawei’s founder and CEO Ren Zhengfei admitted in an interview with China Daily this week that its chips remain “at least one generation behind” US rivals. Still, he noted that the company is investing over $25 billion annually in R&D to close the gap.

Meanwhile, White House AI advisor David Sacks told Bloomberg that while Chinese AI models are 3–6 months behind American ones, the real differentiator is scale—not innovation. “If they close the manufacturing gap, that’s when the global balance shifts,” Sacks said.

“Smuggling is happening”: US contradicts Nvidia

In a sharp rebuke of Nvidia CEO Jensen Huang, who previously told Bloomberg there was “no evidence” of advanced US chips being diverted to China, Kessler offered a blunt counterpoint, “It’s happening. It’s a fact,” he said during testimony, as quoted by Reuters. “We do see significant efforts by wrongdoers to divert AI chips.”

Kessler urged Congress to fund 200 new enforcement officers and expand overseas teams to crack down on diversion networks using third-party nations to circumvent export bans.

Backdrop: A tech war born from tariffs

The AI chip dispute is merely the latest front in a broader US–China rivalry that has grown into a full-blown economic cold war over the past seven years. The roots of this conflict trace back to President Donald Trump’s first term, when the US began imposing sweeping tariffs on Chinese goods.

The standoff escalated dramatically in February 2025, when Trump—newly returned to the White House—signed an executive order imposing a 10 per cent blanket tariff on all Chinese imports. He also revoked the de minimis exemption for low-value parcels, targeting e-commerce shipments that he said were “fentanyl pipelines.”

Beijing hit back with tariffs on coal, liquefied natural gas, crude oil, and vehicle components. It also introduced export controls on rare earths—specifically terbium, dysprosium, and gadolinium—strategic minerals critical to high-performance electronics and military hardware.

By March 2025, the confrontation deepened. Trump raised tariffs to 20 per cent, then up to 54 per cent under a so-called “Liberation Day” order. In response, China escalated duties on key American goods to as high as 125 per cent and filed a formal complaint against the US at the World Trade Organization.

Both nations also hardened export controls on semiconductors, AI software, and chipmaking materials, sending shockwaves through global tech supply chains. The US tightened licensing requirements for chip design tools, while China restricted access to gallium, graphite, and other rare minerals.

A fragile ceasefire was reached in May 2025, when negotiators from both sides met in Geneva and agreed to a 90-day moratorium on new tariffs. But the calm didn’t last.

On June 5, a pivotal 90-minute phone call between Trump and Chinese President Xi Jinping helped restore momentum. That led to high-level trade talks in London from June 9–10, where officials from both countries finalised a framework agreement to resume rare earth exports and stabilise the tech trade.

Trump publicly declared the deal “done” on June 11, pending formal sign-off from Xi. But analysts remain cautious.

What lies ahead?

Although Huawei’s chip capacity remains capped for now, US officials are under no illusion that this advantage will last.

As quoted by Bloomberg, Sacks warned that “the AI landscape is changing month by month.” If China achieves chip self-sufficiency at scale, it could reshape global power structures—not just in tech, but geopolitics too.

For Washington, that makes enforcement, alliances, and R&D investment more urgent than ever.

“We are ahead today,” Kessler said. “But tomorrow is still up for grabs.”

(With inputs from the agencies)