By Evan Swarztrauber

For decades, Arm Holdings, the British semiconductor design firm owned by Japan’s SoftBank, has licensed its proprietary technological language that connects processors to software on fair and neutral terms. Arm’s instruction set architecture (ISA) underpins nearly every smartphone and is widespread throughout cars, cloud data centers, and countless connected devices.

Until recently, its approach has been a case study in the benefits of openness and interoperability.  Arm profits handsomely from its intellectual property, and its neutrality created a foundation for chipmakers across the world to build on its technology. Indeed, when Japan’s SoftBank acquired Arm in 2016, CEO Masayoshi Son said its basic business model would not change and promised to “provide neutrality to all the partners.”

Now Arm and its affiliates are moving to compete directly with ISA licensees. That poses a policy challenge: When a foundational technology controlled by a foreign company is indispensable to U.S. industries, what guardrails are needed to preserve the fair access on which future innovation—and entire product roadmaps—depend?

Imagine if a nationwide railroad spent decades urging manufacturers to build factories along its tracks, offering fair terms and promising not to compete directly with those factories. If the railroad then began building its own factories along those tracks and manufacturing competing products, it could now favor its own shipments, delay rivals’ access, or demand harsher terms from companies who now had nowhere else to go.

This is the concern now playing out with Arm. Companies have invested decades of engineering and billions of dollars in products that depend on Arm’s ISAs. For many firms, switching ISAs would be costly, lengthy, or impractical. And it’s a particular concern for those developing semiconductors the U.S. government considers critical to economic and national security.

After SoftBank acquired Arm, both companies began looking for ways to leverage Arm’s dominance, including plans to hike some royalty rates by as much as 300% and charging royalties based on the value of finished devices—which would have been a stark departure from its longtime business model of calculating royalties based on the value of chips.

Concerns grew when SoftBank moved into finished chips on two tracks. In 2025, it acquired Ampere, a maker of Arm-based data-center processors. And in March, Arm itself moved beyond licensing processor designs and launched its own chip that competes with the Arm-based data center processors developed by Amazon, Google, Microsoft, and NVIDIA.

A recent dispute with Qualcomm illustrates the structural conflict. Arm sued its longtime customer in 2022, arguing that processor designs developed by Nuvia—which Qualcomm had acquired—were not covered by Qualcomm’s existing license and should be destroyed. Arm also threatened to terminate the architecture license on which much of Qualcomm’s chip business depends.

Qualcomm prevailed on the central issues in that first case, but the conflict has continued. In separate litigation, Qualcomm alleges that Arm failed to negotiate in good faith over access to its next-generation Armv10 architecture and demanded unreasonable terms for related technology.

Arm and SoftBank deny they are engaging in anticompetitive conduct, and Arm argues that its incentive to innovate depends on enforcing its contracts and protecting its intellectual property. But the question is not whether Arm should profit and exercise those rights. It is whether an indispensable firm should be permitted to discriminate against its customers for anticompetitive reasons.

The question isn’t limited to Arm. The same issue arises wherever one company controls a chokepoint—an app store, operating system, cloud platform, search index, or online marketplace—while competing with dependent businesses. The recurring danger is that control of one layer can be used to suppress competition in another.

Markets cannot function properly when the owner of an indispensable gateway can weaponize it against competing, dependent businesses. Targeted remedies can often solve for competitive bottlenecks without the morass of price setting and government controls that tend to accompany public-utility status.

Thankfully, Congress is already stepping up to the plate. The bipartisan Fair Access to Critical Computing Technologies Act offers a narrow fix. Proposed by Sens. Katie Britt, R-AL, and Adam Schiff, D-CA, it would require foreign-owned ISA providers to offer U.S. companies access on fair, reasonable, and nondiscriminatory terms. Given the strategic role of semiconductors in commercial and defense systems, it deserves serious consideration.

Arm is already familiar with fair, reasonable, and nondiscriminatory terms. The company promised to maintain a neutral licensing framework when NVIDIA sought to buy Arm in 2020. While the acquisition failed, the framework offers a useful template (and if Congress adopts the Britt-Schiff proposal, a test case) for defining fairness and applying it throughout the tech stack: what qualifies as an essential input, when differentiated contract terms are discriminatory rather than commercially justified, and how policymakers can protect interoperability while respecting legitimate intellectual-property rights.

Congress should consider applying these principles more broadly. “Full-stack neutrality” is a light-touch framework that prohibits companies controlling digital chokepoints from unreasonably discriminating against rivals that depend on their technology or using one layer to foreclose competition in another.

A targeted nondiscrimination requirement like the Britt-Schiff proposal could address the immediate concern, but full-stack neutrality could enhance competition across the tech ecosystem.

Arm should remain free to build factories along its tracks. It should not be free to preference its own freight while sidelining longtime customers. Ensuring that foundational technologies remain platforms for competition rather than instruments of exclusion is the essence of full-stack neutrality.

Evan Swarztrauber is a senior fellow at the Digital Progress Institute and a former policy advisor at the Federal Communications Commission.