Rakesh Kumar says US chip controls backfired as China gains ground

Rakesh Kumar has a gracious way of gloating.

“I’m not going to say ‘I told you so,'” Kumar said. “But it really is ‘I told you so.'”

The professor of electrical and computer engineering at the University of Illinois Urbana-Champaign has spent four years as one of Washington’s most persistent academic critics. His target: the US effort to freeze China’s chip industry in place through export controls.

The numbers behind the told-you-so

Start with Nvidia. In the first half of 2026, the company’s share of China’s AI accelerator market fell to 34%, a decline of 44% year-over-year.

That translates to 678,000 units shipped. Still a big number, but a much smaller slice of a growing pie.

Huawei, meanwhile, climbed to a 28% share over the same period, shipping 572,000 AI accelerators in China. Local companies are increasingly swapping in domestic hardware where foreign products used to sit.

Total AI accelerator shipments in China surged to approximately 2 million units in the first half of 2026.

The mix is also changing. The data points to a shift toward non-GPU chips, which matters because GPUs are Nvidia’s home turf.

What the controls did, and didn’t, do

None of this means export controls were toothless. Restrictions from the Biden administration and allied nations like the Netherlands have constrained China’s ability to produce the most advanced chips.

The key choke point is EUV lithography. These machines use extreme ultraviolet light to etch the tiniest circuit patterns onto silicon, and they are essential for leading-edge manufacturing.

Cut off access to EUV, and you cap how advanced a country’s homegrown chips can get. China has poured tens of billions of dollars into its semiconductor supply chain over the past decade, yet it still relies on foreign equipment for advanced nodes.

A shifting policy landscape

In 2026, export controls on advanced computing items moved toward case-by-case review rather than blanket rejection. That includes case-by-case licensing and tariff adjustments under the Trump administration’s renewed focus on the issue.

The book, and the India angle

Kumar’s arguments are laid out at length in his 2026 book, Chip Age: How Semiconductors Became the World’s Most Powerful Resource.

The book treats chips as a strategic resource on par with the commodities that shaped earlier geopolitical eras. It examines supply chain fragility, great-power rivalry and the surging demand created by AI.

In a recent interview, Kumar also pointed to China as both a template and a warning for countries like India. China’s tens of billions in investment show what heavy state commitment can build, but money alone did not buy China access to the most advanced manufacturing tools, and dependence on foreign equipment remained a vulnerability that rivals could exploit.

What this means

For investors in US chip companies, the first-half data is a clear signal. Access to China cannot be treated as a stable revenue stream when policy can tighten or loosen with each administration.

The growth in total shipments, to approximately 2 million units, confirms that Chinese demand for AI hardware is enormous, and that a growing share of it is being met by local suppliers and non-GPU designs.

For policymakers, the lesson Kumar has been pressing is about second-order effects. Export controls can protect a technological lead at the cutting edge while simultaneously funding, in effect, the rival’s domestic alternative everywhere below it.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.