Biden Divides the World Into Three Tiers of AI Chip Access

The Biden-Harris Administration released a final interim rule on January 13 — the Interim Final Rule on Artificial Intelligence Diffusion — that reorganizes global access to advanced AI chips and to the weights of the most powerful models. The underlying idea, according to the White House statement, is simple yet ambitious: to ensure that "the world's AI runs on American rails" and that adversaries can't easily exploit the most capable systems.

The approach stems from a diagnosis the U.S. government has repeated for years: AI has become central to security and the economy, and in the wrong hands it can accelerate the development of weapons of mass destruction, powerful offensive cyber operations, or human rights abuses like mass surveillance. What's new isn't the goal — it's the mechanism. For the first time, Washington has drawn a complete map of the planet and assigned each region a tier of access.

A world moving at three speeds

The rule sets out six mechanisms that, in practice, sort countries into three broad categories.

At the top sit 18 key allies and partners, who face no restrictions on chip purchases. According to the White House, these are jurisdictions with robust technology-protection regimes and ecosystems aligned with U.S. national security and foreign policy interests. They can make large-scale purchases with no friction at all.

At the opposite end are the so-called "countries of concern," which the rule aims to shut out of the most advanced AI systems and the computing power needed to train them. Here, the message is one of direct containment.

Between these two poles sits the rest of the world, subject to caps and a verified end-user system that determines how much computing capacity each player can accumulate.

A "fast lane" for small orders

One important detail that shouldn't get lost in the geopolitics: the vast majority of chip orders aren't affected by the caps at all. Orders with a collective computing power of up to roughly 1,700 advanced GPUs require no license and don't count against national limits.

The White House notes that this category covers orders from universities, medical institutions, and research organizations for clearly benign purposes. The stated goal is to speed up low-risk shipments, framing the change as an improvement over the previous status quo.

Verified end users: the new passport for compute

The administrative core of the rule is its "verified end user" system, which allocates permissions based on where a company is headquartered and what security standards it meets.

• Universal Verified End User (UVEU). Reserved for entities headquartered in close allies and partners that meet high security and trust standards. With this status, they can deploy up to 7% of their global AI computing capacity in any country in the world — which, per the White House statement, likely amounts to hundreds of thousands of chips. It's a global, enduring status designed to let trusted companies scale quickly while keeping frontier training at home.
• National Verified End User (NVEU). For entities that meet the same security requirements but are headquartered anywhere that isn't a country of concern. They can purchase computing power equivalent to up to 320,000 advanced GPUs over the next two years, meant to serve local, government, and regional customers, always under safeguards against diversion.
• Non-verified entities outside close allies. Even without special status, these can still acquire large volumes of compute — up to the equivalent of 50,000 advanced GPUs per country. The rule presents this as a guarantee that U.S. technology remains available to foreign governments, healthcare providers, and local businesses.

On top of all this comes a diplomatic lever: governments that sign government-to-government agreements — aligning their export controls, clean energy policies, and technology security with those of the United States — can double their chip cap, up to 100,000 of today's advanced GPUs.

The logic is familiar: turning access to silicon into a bargaining chip for exporting not just hardware, but also standards and alliances.

Not just chips: model weights, too

One of the most novel aspects is that the regulation no longer deals solely with semiconductors. For the first time at this level of detail, it wades into model weights — the numerical parameters produced by training that constitute, in effect, the learned "intelligence" of an AI system.

The rule restricts the transfer to untrusted actors of the weights of the most advanced closed-weight models — those whose parameters aren't published. It also sets security standards to protect those weights so they can be stored and used safely around the world without adversaries gaining illicit access.

There's an explicit exception worth flagging for its implications: the rule does not in any way restrict the publication of open-weight model weights. That's a significant distinction for the entire community working with openly available code and weights, which falls outside the scope of this part of the regulation.

At the same time, the rule maintains the requirement that advanced semiconductors sold abroad not end up being used by countries of concern to train advanced AI systems, while still permitting access for general-purpose applications ranging from telecommunications to banking.

Continuity of a long-running strategy

The rule doesn't emerge in a regulatory vacuum. According to the White House, it builds on the chip controls from October 2022 and October 2023 and follows ten months of engagement with lawmakers from both parties, industry representatives, and foreign allies. The stated intent to build bipartisan support and consult with the industry is, in itself, an acknowledgment of how delicate this terrain is.

The central argument recurs throughout the text: don't offshore a critical technology, and make sure the global AI infrastructure depends on American technology. It's a bet on combining openness toward allies with closure toward rivals, calibrating each tier of access.

A legacy with a week left on the clock

The timing of the release is no small detail: the rule arrives with just days left before the change of administration. The fact that it's an Interim Final Rule — a format that takes effect without the usual prior public-comment process — reinforces the sense of a decision its author wants locked in before leaving office.

For industry, the implications are immediate. A GPU maker with customers spread across dozens of countries now has to operate within a map of caps, verifications, and categories that dictates who it can sell to and how much. For companies outside the 18 privileged allies, access to frontier compute now hinges on securing verified end-user status or on their government signing an agreement with Washington.

The big question the rule leaves open is its own survival. An interim regulation published in the final stretch of an administration is exposed to being revised, tightened, or loosened by the next one. What is fixed, though, is the conceptual framework: dividing global access to AI by tiers of trust, treating chips — and now also the weights of closed models — as strategic national security assets.