US-China rivalry over AI and chips intensifies ahead of Trump-Xi summit
Sign up now: Get insights on Asia's fast-moving developments
The US wants to maintain its lead in the AI race as China continues its relentless drive for self-sufficiency and builds up its retaliatory toolkit.
PHOTO: REUTERS
BEIJING – In April, the US Congress began considering a Bill that would further choke off China’s access to high-tech tools used to make semiconductors, which are critical to the artificial intelligence race between the two countries.
The MATCH Act, if passed and enforced, will close loopholes in the US export controls and significantly hamstring China’s AI development, which is already constrained by limited computing power.
The latest push reflects Washington’s broader efforts to tighten restrictions amid intensifying technological competition between the two countries.
Washington wants to maintain its lead in the AI race as Beijing continues its relentless drive for self-sufficiency and builds up its retaliatory toolkit.
China’s Ministry of Commerce warned on April 25 that it would monitor the US legislative process and “resolutely take necessary measures”, adding that Beijing opposed any behaviour that “abuses export controls”.
This comes ahead of a high-stakes summit in Beijing between presidents Donald Trump and Xi Jinping from May 13 to 15 in Beijing.
During Mr Trump’s visit to China – the first by a US president in almost nine years – both leaders are expected to display a level of bonhomie. But any rapport may do little to diminish the tech competition.
Mr Kyle Chan, an expert with the US think-tank Brookings Institution, said on April 16 that Chinese AI companies, particularly start-ups, do not have access to the computing scale available to their American competitors.
This is due to US export controls on cutting-edge AI chips, the lower performance and availability of Chinese domestic chip alternatives, and far less access to capital compared with their American peers.
This underscores China’s continued reliance on leading-edge semiconductors and the tools needed to manufacture them – an area that remains a key vulnerability and sticking point for Beijing.
Professor Nicholas Wright, a technology expert at the US think-tank Center for Strategic and International Studies, said that in AI, the US builds the best racing cars, while China builds the best reasonably priced family saloons.
“Which is better depends on what you are trying to do. America’s edge in cutting-edge AI research and huge spending on AI infrastructure makes it a bit stronger overall right now,” he told The Straits Times.
Hardware remains a major limitation for China, although it is catching up slowly but steadily, noted Prof Wright. “China may even surge ahead if some new technology renders GPUs, in which American designers excel, less central to AI,” he said.
GPUs, or graphics processing units, are essential for training advanced AI models. Although Mr Trump approved Nvidia’s H200 AI chips for sale to China in late 2025, US Commerce Secretary Howard Lutnick said in April that no Chinese firms have bought them yet.
This is because the Chinese government was “trying to keep their investment focused on their own domestic industry”, he told a Senate hearing.
In late April, Chinese start-up DeepSeek released its latest AI model – its first optimised for AI chips and software from Huawei – as part of efforts to develop an alternative to Nvidia’s technology.
With Chinese AI models such as DeepSeek’s just months behind the leading ones from the US, the White House in April accused Beijing of mounting an “industrial-scale” campaign to exploit technology developed by US firms.
As Washington tightens export controls, Beijing has responded on two fronts: accelerating domestic alternatives and expanding the legal tools it can use against foreign restrictions.
On April 7, China announced new regulations to ensure “industrial and supply chain security”, which allow countermeasures against “discriminatory restrictions” by foreign governments, especially in critical technologies.
Ms Hu Xinyue, a senior analyst at the S. Rajaratnam School of International Studies in Singapore, told ST that the rules are China’s first dedicated administrative regulations on industrial and supply chain security.
“This reflects China’s efforts to weaponise domestic law and regulations and legitimise its retaliatory measures in the US-China rivalry. Beyond export controls, the next arena of US-China rivalry may extend to economic lawfare,” she said.
In 2025, Chinese domestic AI chips accounted for nearly 41 per cent of China’s market, compared with Nvidia’s market share of more than 90 per cent before US export restrictions tightened in 2023.
At Mr Trump and Mr Xi’s last meeting in October 2025, China agreed not to restrict the export of rare earths, so long as the US does not raise tariffs to prohibitive levels and expand an export control list targeting Chinese firms.
Ms Hu, who researches US-China competition in semiconductors, said that the geopolitical significance of semiconductors or AI is unlikely to diminish in the foreseeable future given their dual-use nature and wide application scope.
She assessed that China’s measures are likely to remain largely defensive in nature.
“Powerful tools such as rare earth are more likely to be used as bargaining chips only rather than outright coercion, given China’s long-standing opposition to the unilateral use of sanctions and its position as a perceived victim of Western economic coercion.”
Prof Wright said that which country could pull ahead in the near future depends on what triggers the next big leap in AI.
“I think that is more likely to come from US science, but might not,” he said.
For example, China seems likely to build many more robots than the US, and if the physical data from those robots prove key to the next big breakthrough in AI, then China could leap ahead, he added.


