By W. Patrick Wilson 

Evan Swarztrauber’s September 12 essay in The Capitol Forum asks Congress to protect the companies that depend on Arm’s instruction set architecture, the basic command language that lets software run on a processor. I share that goal. The companies I worry about most, though, are the ones that do not exist yet: small design teams and venture-backed startups that need an affordable, predictable path onto the Arm architecture so they can take on the incumbents. Measured by how it would treat them, his case does not hold up. 

His argument runs like this. Arm, which is owned by Japan’s SoftBank, built its business as a neutral supplier of chip designs. Now it has begun selling chips of its own, which puts it in competition with the companies that license its technology. Because those companies cannot easily switch to another architecture, he warns that Arm could favor its own products and squeeze its customers, and he cites Arm’s lawsuit against Qualcomm as evidence. His remedy is a bill from Sens. Katie Britt (R-Ala.) and Adam Schiff (D-Calif.) that would require foreign-owned instruction set providers to license U.S. companies on fair, reasonable and nondiscriminatory terms. This is designed to hit only Arm; in the midst of a big lawsuit. 

Swarztrauber’s case overlooks four things. A nondiscrimination mandate would put at risk the special terms Arm offers startups. Staying at the frontier of chip design is expensive, and royalties pay for it. Qualcomm has its own record as a licensor, and it is not a good one. And unlike a startup, Qualcomm already has somewhere else to go. Most of all, he overlooks who would gain. In its own lawsuit against Arm, Qualcomm complains that Arm refused to offer it commercially reasonable prices. A federal rate mandate would hand Qualcomm the leverage over Arm’s pricing that it has been seeking in court. 

Start with how new entrants get access today. Arm Flexible Access for Startups gives companies with less than $50 million in funding and less than $5 million in annual revenue “$0 access” to Arm IP, tools, training and support, with “license fees due only at manufacture.” That program exists because Arm can offer different terms to different classes of customers. A licensor under a legal duty to treat every customer alike has every reason to retire its cheapest terms rather than extend them to Qualcomm. 

Staying at the frontier costs Arm more each year. In the fiscal year that ended March 31, Arm spent $2.8 billion on research and development, 56% of its $4.9 billion in revenue. Much of that money goes to the engineers who do the work. Arm’s average workforce grew from about 7,700 to about 9,000 in a single year, in a market where the Semiconductor Industry Association projects 67,000 U.S. chip jobs will go unfilled by 2030. Arm’s gross margins are the envy of the industry, but after paying for that research, the company kept only about 18% of its revenue as net income. Consider another chip company, the memory maker SK hynix, which kept 44% of its revenue in 2025. Qualcomm wants a tailor-made statute to lower Arm’s royalties, yet those same royalties fund the designs a rival startup can use at no upfront cost. 

The op-ed leans heavily on Arm’s lawsuit against Qualcomm. That case turned on whether licenses held by Nuvia, a chip design startup Qualcomm bought for $1.4 billion in 2021, carried over to Qualcomm. A jury and then a federal judge sided with Qualcomm, and Arm has said it will appeal. That is a contract dispute between two large firms, and the courts are handling it. 

What the op-ed leaves out is Qualcomm’s own record with new entrants. In 2019 the European Commission fined Qualcomm €242 million for selling baseband chips below cost to Huawei and ZTE “with the intention of eliminating a competitor,” in Commissioner Margrethe Vestager’s words. That competitor was Icera, a British modem startup that was out of the market by 2015. In April 2023, South Korea’s Supreme Court upheld a fine of about 1 trillion won ($760 million) after finding that Qualcomm refused to share key standard-essential patents with modem chipset makers, including Intel and Samsung. Qualcomm licensed its cellular patents to phone makers and declined to license the chip companies trying to compete with it. Arm’s model works the other way: it licenses its architecture directly to chip designers, from the largest firms down to startups with no revenue. 

Swarztrauber’s op-ed also faults Arm for considering royalties based on the value of finished devices. That is odd, because this is exactly Qualcomm’s own business model. Qualcomm’s published 5G licensing program charges 3.25% of the net selling price of branded 5G handsets. The royalty is set by the price of the phone, and the program’s terms say nothing about whose modem is inside, whether Qualcomm’s or a rival’s such as MediaTek’s. A company that has collected royalties on the price of the whole phone for years is an odd messenger for the view that this very practice threatens competition. 

The author also leans heavily on a railroad analogy and wrongly asserts that licensees have nowhere else to go. Qualcomm knows better. In December 2025 it acquired Ventana Micro Systems, a RISC-V processor designer, and said the RISC-V architecture “has the potential to advance the frontier on CPU technology.” Qualcomm is fortunate that it can afford to design for two rival architectures at once. A startup with 20 engineers cannot, which is why stable, standard terms matter far more to a startup than new leverage in a renegotiation matters to Qualcomm.  It’s an arms race for talent right now, and Qualcomm wants to leverage its huge numerical advantage against any disruptive new players.  

Swarztrauber also paints Arm’s new data center chip as a move against its customers. Yet Arm’s own CEO, Rene Haas, has said publicly that Meta asked Arm to build it. In addition, there is another reason to jump into the market that anyone who has recruited engineers will recognize. The best chip designers want to see their work reach silicon as a finished, manufactured chip. A company that licenses designs but never tapes-out a chip of its own (sending a completed design to the factory) will have a hard time hiring and keeping talent, and these engineers are critical to every Arm licensee, large or small. Engineers doing meaningful work are the lifeblood of the semiconductor business. 

That brings me to the Britt-Schiff bill. As the op-ed describes it, the bill would impose fair, reasonable and nondiscriminatory obligations only on foreign-owned ISA providers. In practice that means one company. Rate disputes under FRAND commitments in the patent world are slow, expensive fights over expert testimony, and the firms that win them are the ones with the largest legal budgets. Under this bill, the biggest licensees would gain a new forum to relitigate their rates. Small licensees would gain a process they cannot afford to use. When I was senior counsel to the House Committee on Small Business, I saw several similar instances of powerful incumbents trying to use regulation to pull up the ladder behind them. This bill fits that pattern. 

Rest assured though, even without this boutique legislative fix, regulators are already on the case. Qualcomm has taken its complaints to several venues, including the European Commission, the Federal Trade Commission and the Korea Fair Trade Commission. Korean officials raided Arm’s Seoul office in November 2025, and in May the FTC reportedly opened an investigation into whether Arm restricts access to its architecture as it begins selling its own chips. If Arm uses its chip business to disadvantage licensees, existing antitrust law already reaches that conduct. Congress should let those reviews run before enacting a legislative work-around for one company’s contractual fight.  It’s surprising these senators even want to.  

There is a simple test for any proposal in this area: Does it lower the cost for the next startup to get started? Does it help the disruptors? In the AI-powered economy, a special mandate built for just the largest customer in the room fails that test. More than anything else, protecting the entry path for small and new chip companies is how the United States will keep this market competitive and keep customers at the heart of innovation.

Patrick Wilson is Founding Principal of Semiconductor & Innovation Group (SIG). Prior to founding SIG, he worked for global chipmaker MediaTek for more than five years, served in the Office of the Secretary of Commerce as Director of the Office of Business Liaison, and was Senior Counsel to the House Committee on Small Business. He also worked with Arm while at the Semiconductor Industry Association.