SHENZHEN, China — The tech company SiCarrier is hardly a household name. The government-backed Chinese firm makes things most people have probably never heard of, like epitaxy equipment and atomic layer deposition tools used in microchip fabrication.
Wonky stuff.
But at a chip industry expo in the southern Chinese city of Shenzhen this fall, a crowd thronged its display booth, snapping pictures, doing livestreams and marveling at its wares.
"Their products are very good — they're excellent," said Zhang Hengming, who showed up at the SiCarrier booth waving the blue flag of an AI electronics alliance that he chairs. "We support made-in-China products so they can get stronger and reach the world."
The U.S. has been using ever-tightening trade restrictions to limit the flow of high-end microchips to China, along with the gear to make them. The goal is to try to keep Beijing behind in artificial intelligence and to prevent China's military from acquiring the best chips.
But this chip "blockade," as some see it, has been a rallying cry in China — even as Trump appears poised to loosen it.
Zhang said that overall, the friction with the United States over technology is a good thing for China's chip industry. "Chinese chips will be able to compete in the world. No problem," he said.
SiCarrier, known as Xinkailai in Chinese, is just four years old. But it's hailed as a key player on China's "national team" — a group of state-backed companies that are hustling to try to overcome Western tech restrictions. (SiCarrier did not make any executives available for interviews with NPR at the expo.)
In the shadow of U.S. export controls, the Chinese government has been doubling down on its pursuit of tech self-sufficiency, pouring upward of $200 billion into efforts to create a modern and self-reliant chip industry.
Chinese chips lag the best in class, especially for AI. And while China can design and prototype relatively advanced chips, it has not mastered mass production.
In April, according to state media, Chinese leader Xi Jinping said China needs scientific breakthroughs if it is to gain superiority in AI, and he pledged significant policy support.
At the expo in Shenzhen, hundreds of companies were on hand, and thousands of people showed up for a gander at the best that China has to offer.
Jarod Wang, who works for a Shenzhen-based chip design firm, says building a competitive and advanced chip industry won't be easy.
In a speech at the expo, Wang ticked through the challenges: China can't buy the best graphics processing unit (GPU) chips for AI data centers; it's blocked from acquiring high-end chipmaking equipment; it faces cloud computing restrictions; state-of-the-art AI software is geared for foreign chips. The list went on.
"So for every company, in terms of development, they actually need to innovate their pathways," Wang told NPR after his talk. In other words, he said, they need to move beyond a copy-paste mentality.
"They can no longer just continue following their previous development mindset," he added.
Wang is optimistic about China's ability to innovate its way forward alone, he said, because there are no major technical barriers when it comes to chipmaking. "It simply requires time and resources," he said.
This is familiar territory for China. The ruling Communist Party regularly projects the country as an underdog that, led by the party, has time and again relied on itself to forge technological breakthroughs. That is how it became a nuclear and space power.
Zeng Yaoguang works for a company that makes precision pneumatic tools, which were on display at the chip expo in Shenzhen. The equipment is used in chip fabrication plants and other factories.
"These were high tech 15 years ago in China," he said. "Now they're very common here. We think chips are so advanced, but in 50 years, or 30 years' time, they could be like this stuff."
Chinese people work hard — and work smart, he said. And when it comes to the semiconductor industry, he thinks, China may be able to surpass the U.S. very quickly.
Experts can't rule that out. And some have been puzzling over the Trump administration's decision this month to roll back some chip restrictions and allow California-based Nvidia to sell one of its best GPUs — the H200 — to China. Nvidia is the world's leader, by far, in such chips, which are used in AI data centers for model training and operation.
On social media, Trump said that H200s would be sold to select buyers in China in a way that allows for "continued strong" U.S. national security and that the U.S. would get a 25% cut of sales. Nvidia would still be prohibited from selling its top-of-the-line products to China.
Administration officials have argued that allowing China access to good — but not cutting-edge — chips from the United States would keep it hooked on U.S. technology and dampen China's drive for chip independence.
Chris Miller, author of the book Chip War: The Fight for the World's Most Critical Technology and a professor at Tufts University, is skeptical. Private companies in China might like to buy American chips, but the government has maintained its focus on its longer-term priority. "The Chinese government is dead set on trying to build out its own chip ecosystem, which it has been trying to do now for over a decade," he said.
"We've seen the Chinese government over the last six months tell Chinese tech companies that they can't buy certain types of U.S. chips because they want these companies to build out the Chinese ecosystem," Miller continued. "I think this shows that even if we want China to stay addicted to our chips, the reality is that China is trying to find its own sources and it doesn't trust U.S. supply chains."
David Sacks, a China specialist at the Council on Foreign Relations, says whatever Trump's rationale may be for offering China H200 chips, it carries a strategic risk.
"If you believe that AI is going to change economies and potentially even change the way of warfare," he said, "why give your chief rival the tools that it needs to potentially beat you in this race?"
Jasmine Ling contributed to this story.
Copyright 2025 NPR
