US Commerce Secretary Gina Raimondo, the Biden administration’s leading light behind efforts to limit China’s access to advanced chips and related technologies, now says that export controls are merely “speed bumps” and that “trying to hold China back is a fool’s errand.”
In Raimondo’s view, the CHIPS and Science Act – a US$52.7 billion industrial policy aimed at reviving US semiconductor production and high-tech R&D and signed into law by President Joe Biden in 2022 – is more important than sanctions against China.
“The only way to beat China is to stay ahead of them,” Raimondo told The Wall Street Journal in an article published on December 22. “We have to run faster, out innovate them. That’s the way to win,” she said.
President Biden, speaking at the Brookings Institution earlier in December, said that the CHIPS and Science Act, together with the Bipartisan Infrastructure Law and the Inflation Reduction Act, “mark the most significant investment in America since the New Deal.”
This assessment is not wrong, but it is incomplete in leaving out the success of the biggest speed bump, problems with the CHIPS Act and the unintended consequences of sanctions.
Trying to hold back China may be a fool’s errand in the long run but it has had one notable success: In 2019, the US government persuaded the Netherlands to ban the export of ASML’s EUV lithography machines to China.
This limited China’s ability to make chips beyond 7nm, and, at an expensive stretch, 5nm design rules, while Taiwan’s TSMC is now in commercial production at 3nm and is planning to introduce 2nm in 2025.
As a result, Nvidia, AMD, Apple and other non-Chinese integrated circuit design companies have access to mass production at 5nm, 4nm and 3nm, while Huawei and other Chinese tech companies do not.
Samsung is close behind TSMC, although at a smaller scale; Intel is outsourcing to TSMC while working on its 3nm yields; and Samsung, Intel and Japan’s Rapidus are all aiming at 2nm.
ASML CEO Christophe Fouquet recently said that without EUV lithography, Chinese semiconductor makers would lag the global industry by 10 to 15 years. At the leading edge of miniaturization, that may be true.
His predecessor, Peter Wennick, said, “If they cannot get those machines, they will develop them themselves. That will take time but ultimately they will get there.” But how much time? Five years have already passed.
At present, Chinese engineers seem to be having enough trouble developing their own ArF immersion DUV lithography, the chip-making technology just behind EUV, although they have imported a lot of the former equipment.
They have also turned to open-source RISC-V architecture, chiplets and creative thinking to circumvent export controls on EUV lithography, other sophisticated equipment and advanced ICs such as Nvidia’s A100 and Blackwell AI processors, which are produced by TSMC.
According to Jack Clark, former policy director at OpenAI and co-founder of California AI developer Anthropic, “One way China will get around export controls [is] building extremely good software and hardware training stacks using the hardware it can access.”
He also wrote that “Made in China will be a thing for AI models, same as electric cars, drones, and other technologies.”
The US Semiconductor Industry Association (SIA) reports that, as of December 17, the CHIPS Program Office had announced $42.4 billion in grants and loans to 27 companies, catalyzing 40 semiconductor-related projects in 21 American states.
“These projects include total investment of more than $386 billion over two decades,” the SIA wrote, “with the vast majority invested by 2030.”
Then, on December 20, the Department of Commerce announced $4.7 billion in funding for Samsung Electronics, bringing the total to more than $47 billion and perhaps alleviating concerns that the incoming Trump administration might move to stop CHIPS Act funding, which the president-elect has called a “bad deal.”
The biggest CHIPS Act subsidies have gone to Intel, TSMC and Micron Technology. So far, only one company, Microchip, has abandoned its application for CHIPS funding – because it is closing factories, not building new ones.
But Intel, having dropped into the red and seen its share price collapse, is cutting its capital spending by more than 20% and laying off more than 15% of its workforce. CEO Pat Gelsinger, who lobbied hard for CHIPS Act funding, has also been forced out.
On balance, the CHIPS Act is a success but the crisis at Intel was an unpleasant surprise and the Act itself undermines US criticism of semiconductor subsidies in other countries, which have grown by leaps and bounds in China, Taiwan, Japan, South Korea, India and Europe. As a result, the US share of the global semiconductor industry might not rise by as much as some had hoped.
A study conducted by the SIA and Boston Consulting Group, released in May 2024, concluded that “The US share of the world’s chip manufacturing capacity will increase from 10% in 2022 – when the CHIPS and Science Act was enacted – to 14% by 2032, marking the first time in decades the US has grown its domestic chip manufacturing footprint relative to the rest of the world. In the absence of CHIPS enactment, the US share would have slipped further to 8% by 2032.”
Commerce Secretary Raimondo expects the US to account for about 20% of advanced logic IC production by 2030. Part of that will be made by TSMC, which currently accounts for 64% of advanced logic production, according to SemiWiki. TSMC is also building fabs in Japan, but at the end of the decade, most of its production will probably still be in Taiwan.
Getting to even 14% won’t be easy. Industry association semi expects North America to account for only 9% of worldwide semiconductor production capacity in 2025. That would put the US in fifth place after China (30%), Taiwan (17%), South Korea (16%) and Japan (14%).
It has become somewhat fashionable to predict that Chinese semiconductor investment will slow down, but with the Biden administration now launching a Section 301 investigation into China’s alleged “targeting of foundational semiconductors (also known as legacy or mature node chips) for dominance,” this seems unlikely.
The Office of the US Trade Representative (USTR) will also look into the incorporation of these semiconductors into electronic equipment used in “critical industries like defense, automotive, medical devices, aerospace, telecommunications, and power generation and the electrical grid,” as well as “materials critical to chip manufacturing such as silicon carbide and wafers.”
Silicon carbide is used to make power semiconductors used in electric vehicles.
As noted by Asia Times journalist Yong Jian, Chinese commentators point out the hypocrisy of the creator of the CHIPS Act accusing China of “non-market practices,” that China’s semiconductor production is intended primarily for domestic consumption, and that as the US continues to tighten its sanctions, China keeps increasing its investments.
By now it should be clear that the US wants to contain the entire Chinese semiconductor industry, not just the advanced chips it says are key to national security. For its part, China is trying to reduce its large semiconductor trade deficit, which not only costs it financially but also makes it vulnerable to sanctions.
But those sanctions continue to incentivize the Chinese innovation Biden’s outgoing administration had hoped to suppress, the “fool’s errand” Raimondo has belatedly acknowledged on her way out the door.
Follow this writer on X: @ScottFo83517667