Examining the broader impact of changing chip export controls

Though US President Donald Trump has suggested blocking exports of NVIDIA Corp.'s Blackwell-generation AI chips outside of the US, industry and policy experts remain unclear on how or if the limits will be enforced, raising significant questions for US allies and supply-chain partners across Asia.

Earlier this month, Trump said in a "60 Minutes" interview that Blackwell chips would be reserved for the US, saying, "The most advanced [chips], we will not let anybody have them other than the United States." The remark raised questions across the industry about whether it signaled a shift toward hard geographic restrictions — keeping Blackwell on US soil — or simply reiterated the longstanding end-user and entity-based controls that could permit US companies to deploy Blackwell in overseas data centers in Japan, South Korea, Singapore, Taiwan, Australia or India, so long as the chips do not reach restricted entities.

While there are early indications the White House is open to Blackwell exports — the Commerce Department this month authorized the export of up to 35,000 Blackwell chips to G42, based in the United Arab Emirates, and Humain, based in Saudi Arabia — there are also signs that export restrictions are still being figured out and subject to change. If the US prohibits exports of NVIDIA's Blackwell AI chips, Asian allies and US supply-chain partners would face significant adjustments to cloud infrastructure, data governance and AI competitiveness, technology and policy experts told S&P Global Market Intelligence.

"There's a lot of confusion and anxiety, but there's no actual policy changes yet, just a suspicion that the US could conceivably tighten controls, including in Korea in line with the more 'America first' type of agenda," said Jennifer Hendrixson White, a distinguished visiting fellow at UPenn who held senior roles across the US government from 2010 to 2025, including in the Senate and United States Department of State.

Spokespersons for the Office of Science and Technology Policy and the United States Department of Commerce did not respond to requests for clarification.

China in the spotlight

In particular, policy and industry leaders are closely watching which restrictions will apply to China. Currently, the Commerce Department relies on performance thresholds that block the sale of the most powerful graphics processing units — including NVIDIA's H100, H200 and Blackwell B200 — to China.

NVIDIA has advocated for a loosening of these restrictions. While the company is approved to sell its H20 chips in China, the Chinese government has objected to the use of H20 chips in the country, citing security concerns.

"While we were disappointed in the current state that prevents us from shipping more competitive data center compute products to China, we are committed to continued engagement with the US and China governments and will continue to advocate for America's ability to compete around the world," NVIDIA CFO Colette Kress said during the company's Nov. 19 earnings call. "To establish a sustainable leadership position in AI computing, America must win the support of every developer and be the platform of choice for every commercial business, including those in China."

On Nov. 24, however, Commerce Secretary Howard Lutnick told Bloomberg News that Trump is considering whether to allow NVIDIA to sell its H200 chips to China. The H200 is estimated to be twice as powerful as the H20.

As of Nov. 24, Visible Alpha consensus estimates for NVIDIA's China revenues for the current quarter are up 14.4% from a month ago. Analysts now expect $12.93 billion in revenues from the country in the fiscal fourth quarter, up from the $11.31 billion estimate from one month earlier.

Allied AI ambitions at stake

The impact of the ban on exporting advanced chips to China is not limited to that country, experts told S&P Global Market Intelligence.

Carmen Li, founder of Silicon Data, said a tighter supply of Blackwell chips could also spill into adjacent markets. When access to H100s was curtailed in China, regional cloud prices rose as demand concentrated on a smaller pool of available GPUs. A similar squeeze on Blackwell, Li said, could redirect more workloads to H100 clusters and introduce new volatility across Asia's cloud markets.

Smaller regional providers could be hit particularly hard, said Charles Wong, CEO of Bifrost AI. From the operator perspective, access to the newest GPUs is a primary competitive lever. Large US players can often adapt around supply-chain constraints, Wong added, but many smaller firms cannot.

"That is literally how they get that competitive edge, because if they can train bigger models faster than their competitors, then it's going to be everything for them," Wong said. "There are plenty of founders and businesses in the space which don't have those protections, don't have those fail-safes, and they're going to be disproportionately affected."

Wong said customers are already adapting to constraints by diversifying hardware stacks.

"There's simply a forcing function of needing to find different suppliers, different vendors for everything," he said.

A more restrictive trade environment, he warned, could carry long-term costs for innovation.

"I think the way for the global business environment to work is by open innovation, open trade … for us to have that free-market motivation pushing us along," Wong said.

Supply chain challenges across Asia

Korea's semiconductor sector illustrates the cross-pressures facing regional suppliers. Korea had previously been afforded "tier one" access under the now-rescinded Biden-era AI Diffusion Rule, which granted it the highest level of access to US advanced chip exports.

Companies such as Samsung Electronics Co. Ltd. and SK hynix Inc. play central roles in the global memory supply chain and continue expanding their US manufacturing footprints.

Korean semiconductor companies not only serve as major suppliers to NVIDIA — they also maintain NAND flash memory and Dynamic Random-Access Memory facilities in China.

"NVIDIA can't really move forward without [Samsung and SK]," said Don Southerton, a Korean business consultant and strategist. "There's so much of a symbiotic relationship between them."

Tighter US controls on Blackwell chips "will have a secondary effect on Korean producers, who may lose out from the China market gap" while navigating more complex compliance burdens, said Tom Ramage of the Korea Economic Institute Of America Inc. The restrictions could also accelerate China's pace of development of its own domestic industry and create more competition for Korean firms, Ramage added.

"If anything, the restrictions on the Blackwell AI chips represent a US policy of picking the winners and the losers when it comes to the AI race," he said. "Carve-outs for allies will come with conditions that align firms more closely with US economic security priorities."