BEIJING, Aug 11 — US semiconductor giants Nvidia and Advanced Micro Devices will pay the United States government 15 per cent of their revenue from selling artificial intelligence chips to China, according to media reports.
Here is what you need to know about the chips and their role in the US-China tech war:
Why are these chips so important?
The chips in question, according to the reports, are Nvidia’s “H20” chip and the “MI308” from Advanced Micro Devices (AMD).
The California-based companies developed these chips—less powerful than their flagship models—specifically for the Chinese market.
They are significant as they are AI “accelerators”, meaning that they are useful in carrying out demanding tasks and computation, said Dylan Loh, an assistant professor at Singapore’s Nanyang Technological University.
“This is obviously a key tool to expand the AI-hungry demands of countries and companies building out their AI tools,” Loh told AFP.
Previously, the United States had been restricting the export of such chips on national security grounds.
The two chipmakers had warned in April of big financial losses connected to these earlier constraints.
Nvidia expected the previous rules to cost it US$5.5 billion (RM23.3 billion), while AMD forecast it could sap as much as US$800 million from the company’s bottom line.
Why are sales to China controversial?
Nvidia and other US chip companies have lobbied against the tough restrictions in recent years on selling cutting-edge semiconductors to China.
That pressure appeared to have paid off last month, when Nvidia said Washington had pledged to let the company sell the H20 chips to China.
The policy reversal has “raised eyebrows”, Loh said.
“These chips and many other chips have a dual use nature which means it can be used to advance their military capabilities in various ways,” he told AFP.
This includes the possible use of these chips potentially by US forces in the event of a confrontation, Chong Ja Ian, an associate professor at the National University of Singapore, told AFP.
Defending the policy change last month, US President Donald Trump’s AI point person David Sacks told CNBC the H20 was a “deprecated chip” that is “not anywhere close to the state of the art”.
He said the reversal on the H20 came because Nvidia’s Chinese rival Huawei was making “huge strides” and could potentially threaten Nvidia’s market dominance.
Yesterday’s reported deal is “unprecedented” though, Chong said.
“They do suggest the cost of market access in an environment of more intense major power competition,” he added.
There could still be obstacles ahead for the chipmakers.
US lawmakers have proposed plans to require Nvidia and other manufacturers of advanced AI chips to include built-in location tracking capabilities.
On Thursday, Trump demanded that the new boss of US chip maker Intel resign “immediately”, after a Republican senator raised national security concerns over his links to firms in China.
What does China say?
China has meanwhile sought to increase its self-reliance in the field of semiconductors.
Nvidia is facing growing distrust in China, with state media on Sunday labelling the H20 chips as unsafe.
“The US government’s decision to allow Nvidia to export the H20 to China inevitably raises some troubling concerns,” said Yuyuan Tantian, an outlet affiliated with state broadcaster CCTV.
It is not a secure chip for China, not advanced, and not environmentally friendly, it added in an article on social media platform WeChat.
Beijing’s top internet regulator summoned Nvidia representatives in July to discuss “serious security issues” involving the H20.
The Cyberspace Administration of China said it had asked Nvidia to “explain the security risks of vulnerabilities and backdoors in its H20 chips sold to China and submit relevant supporting materials”.
The statement noted that, according to US experts, location tracking and remote shutdown technologies for Nvidia chips “are already matured”. — AFP
The semiconductors costing Nvidia, AMD dearly
