US Scraps Biden Rule That Limited AI Chip Exports

On Tuesday the US Department of Commerce withdrew the AI Diffusion Rule, due to start on 15 May. The regulation, issued five days before the Biden team left office, would have capped sales of American AI chips to most nations. According to the Department of Commerce, the measure would have raised costs for US firms and harmed ties with dozens of partners.

Under-Secretary Jeffery Kessler told staff to drop enforcement and promised a new plan later. He said the Trump team wants an “inclusive strategy” that shares AI technology with trusted countries while blocking rivals. The statement claims the step keeps US innovation moving and cuts paperwork for developers. Kessler framed the withdrawal as part of a larger drive to clear outdated regulations.

The reversal followed months of lobbying from chip designers and cloud platforms. According to The Register, firms warned that tight limits would drive customers toward Chinese suppliers if US parts sat in red tape. A few security voices backed the rule, but they lost the debate.

 

What New Controls Did Commerce Announce?

 

Commerce will file a Federal Register notice that scrubs the old policy. Details of the next framework are scarce, but the Bureau of Industry and Security (BIS) set out extra shields against unauthorised chip flows in the same statement.

One notice warns that anyone using Huawei Ascend processors anywhere in the world breaches US export rules, as the devices likely rely on American know-how without a licence, according to the Department of Commerce. The warning piles pressure on suppliers still linked to the Shenzhen firm.

BIS also reminded cloud suppliers that they risk penalties if they let Chinese military projects rent AI hardware. According to The Register, chip makers must now seek licences before selling accelerators to foreign infrastructure-as-a-service firms when those parts might train hostile models. The department says the rule will close what it calls the “IaaS loophole” that let foreign users sidestep chip bans by renting power in distant clouds.

The update even covers chips already inside overseas data centres. Moving existing graphics processors within a foreign operator’s network now needs approval if the hardware could aid adversary weapons work.

 

 

How Could The Changes Affect Global AI Work?

 

Washington argues that cancelling the diffusion rule while adding targeted chip curbs will keep American leadership in AI while still letting allies buy hardware. Supporters call the mix more flexible than the Biden blueprint and say it cuts the chance of a blanket embargo that might invite retaliation. Commerce officials contend that strong research links with Europe, Japan and Australia depend on this mixed stance.

Opponents fear loopholes. They warn that rival governments could buy mid-range chips, chain them together and still train advanced models. According to The Register, Commerce tried to plug this gap by stating that even “in-country” transfers of older gear will face scrutiny. They also note that software tricks now let developers squeeze more performance from mid-grade silicon, making guardrails on every chip central, in their view.

Chip vendors such as Nvidia have stayed quiet since the announcement. The companies had fought against the diffusion rule earlier in the year, arguing that US suppliers could lose billions if barred from large parts of the world market. Their task now is to examine the replacement once it surfaces.

Trade lawyers say the new guidance also matters for large cloud firms. US data centre operators already block direct work with Chinese defence labs, but they must now ask tougher questions about resale and third-party access to their hardware.

A replacement rule is promised, though no draft has appeared. Until then, chip sellers and cloud hosts must work with a patchwork of advisories, licence forms and case-by-case rulings.