In what appears to be a case of diplomatic mind games in action, one day after the US government issued a regulation clearing the way for Nvidia to sell its H200 artificial intelligence processors to Chinese companies on a case-by-case basis, a published report has revealed Chinese custom officers have been told not to let them into the country.

The ruling announced Monday by the US commerce department’s Bureau of Industry and Security (BIS), the primary agency that oversees export controls, represents a revision to its licensing policy for semiconductor exports to China, it said in a release.

The BIS will now review license applications for Nvidia H200 and similar chips providing that certain security requirements are met, including the stipulation that exporting these products to China will not reduce global semiconductor production capacity currently available to US customers.

Nvidia expressed appreciation for the ruling. A spokesperson told NetworkWorld, “we applaud President Trump’s decision to allow America’s chip industry to compete to support high paying jobs and manufacturing in America. Offering H200 to approved commercial customers, vetted by the Department of Commerce, strikes a thoughtful balance that is great for America.”

But the script changed on Wednesday when Reuters reported, “Chinese customs authorities told customs agents this week that [the chips] are not permitted to enter China, according to three people briefed on the matter.”

More symbolic

“The impact of these flip-flop policies should be minimal to the enterprise,” said Scott Bickley, advisory fellow at Info-Tech Research Group. “The volumes of chips involved are not material to the Nvidia allocation from TSMC. This is more a symbol of the ongoing trade war between the US and China, with the latter indicating they don’t need US chips at scale for AI development.”

He noted, “the broader implications and potential impacts may signal to enterprise customers of Nvidia that perhaps they don’t need the latest and greatest GPUs from [them] either to achieve acceptable results across select AI workloads. It is doubtful that Nvidia would commission additional production issues for H200 without China as the customer willing to pay a premium price. Other customers will happily purchase this stock in lieu of China.”

And last month, Charlie Dai, VP and principal analyst at Forrester, said renewed H200 access is likely to have only a modest impact on global supply, as China is prioritizing domestic AI chips and the H200 remains inferior to Nvidia’s latest Blackwell-class systems in both performance and appeal.

He pointed out, “while some allocation pressure may emerge, most enterprise customers outside China will see minimal disruption in pricing or lead times over the next few quarters.”

H200 now pulled onto the ‘geopolitical chessboard’

Forrester senior analyst Alvin Nguyen said Wednesday that he agrees with Dai’s assessment, especially with the recent developments of the US now permitting and China moving to effectively ban the import of H200 chips.

“This is older AI technology; it is still useful, but adding a premium to it when the Chinese AI ecosystem is catching up or caught up to what is being offered will make it a target for capacity rather than a first choice for enterprises in China,” he said.

“For global enterprises with Nvidia in their AI tech stack, it makes sense to maintain standards across regions/locations if they are able to bring in H200s into China,” Nguyen said. “Outside of China, this could lead to longer lead times and costs not decreasing, but global enterprises are already plagued by uncertainty and will adjust.”

Sanchit Vir Gogia, chief analyst at Greyhound Research, had a different point of view, saying “the H200 situation matters because it has become a case study in how enterprise assumptions about infrastructure availability can be invalidated by policy volatility, not product lifecycle.”

A new category of risk

The H200, he said, “is not the most powerful chip in Nvidia’s portfolio, and wasn’t at the top of any CIO’s wish list. That’s precisely why it was attractive. It sat in the tier that enterprises could reasonably expect to access without bidding wars or global competition.”

He explained, “it had enough capability for mid-scale inference and internal fine-tuning, especially for firms not building frontier models. And yet, it has now been pulled onto the geopolitical chessboard, first through a highly conditional US approval and then a sudden, informal block at Chinese customs.”

The real story here, he said, “is not whether H200 itself makes or breaks enterprise AI plans. The story is that even legacy silicon is no longer safe from last-minute policy swings. Enterprises used to worry about whether chips were fast enough or cost-effective enough. Now they have to worry whether the rules will even allow those chips to ship, integrate, or support remote workloads in different geographies.”

This creates a new category of risk, Gogia said. “It is not technical. It is regulatory, interpretive, and highly political,” he said. “For enterprise CIOs and procurement heads, it means that AI infrastructure can no longer be built around static assumptions. What matters today is not just the specs of a chip, but the geopolitical narratives surrounding it.”

He added, “when something as structurally stable as a two-year-old GPU can be tossed into policy limbo, that sends a very clear message: infrastructure planning needs to be engineered for volatility, not just for scale or speed.”

This article originally appeared on NetworkWorld.