Qualcomm booked around forty-four billion dollars in revenue last fiscal year. A little under six billion of that came from a division most people holding a Snapdragon phone have never heard of, a division that runs at margins the chip business can only dream about and that, for most of the last decade, generated the majority of Qualcomm’s pre-tax profit outright. The chips get the headlines; the royalties quietly pay for everything.
When I wrote about Qualcomm trying to sell Wall Street on a future in data-center silicon, I set this part aside on purpose and said it had earned its own post. So here it is: the part of Qualcomm called QTL, the licensing business, and the reason the whole diversification story exists in the first place.
The mechanics are simpler and stranger than the chips they protect. To buy a Qualcomm modem, you first have to license Qualcomm’s patent portfolio. No license, no chips. That policy is the spine of the company, and it has been for decades.
The clever part is how the royalty gets calculated. It isn’t a fee on the modem. It’s a percentage of the wholesale price of the entire handset. Two phones can carry the identical Qualcomm modem, and the more expensive one pays Qualcomm more because the camera, display, and storage all count toward the number the royalty is skimmed from. Analysts have pegged the historical rate on a 4G phone at a little over 3% of wholesale, inside a general band of 3 to 5%. After a decade of regulators leaning on the company, there’s now a cap on the device price the royalty can run against, somewhere around $400, so a thousand-dollar flagship isn’t taxed on the full grand. You license the whole bundle too, roughly 140,000 patents across tens of thousands of families, handed over as one portfolio rather than a menu you pick from. Qualcomm gets paid on the value of the phone, not the value of its slice of the phone. That is the thing regulators spent ten years trying to break, and mostly couldn’t.
Start in China. In February 2015, the NDRC fined Qualcomm about $975 million, calculated as 8% of the company’s 2013 sales in the country, and Qualcomm chose not to appeal. It agreed to license standard-essential patents separately from the rest, and to drop the royalty base in China to 65% of a device’s price. A concession, but a local one.
South Korea went harder. The KFTC hit Qualcomm with 1.03 trillion won, around $850 million at the time, in December 2016, specifically for the “no license, no chips” coercion and for refusing to license its standard-essential patents to rival chipmakers like Intel and Samsung on fair terms. Qualcomm fought it for six years. Korea’s Supreme Court upheld the decision in 2023, and that one stuck.
Europe split the difference across two separate cases, and the split is instructive. The Commission fined Qualcomm 997 million euros in 2018 for paying Apple to stay exclusive, and the EU’s General Court threw that decision out entirely in 2022, censuring the Commission for procedural failures serious enough that it declined to appeal. A second case, a 242 million euro fine in 2019 for selling 3G chips below cost to strangle a British rival called Icera, later absorbed into Nvidia, went the other way: the General Court largely upheld it in 2024, trimming the number slightly to 238.7 million euros. One overturned, one confirmed, from the same regulator against the same company.
The one that mattered most happened at home. The US FTC sued in 2017, and in 2019 Judge Lucy Koh ruled that Qualcomm’s practices had strangled competition in the modem market for years. Then, in August 2020, the Ninth Circuit reversed her completely. It held that “no license, no chips” was neutral across chip suppliers, and that Qualcomm had no antitrust duty to hand its patents to competitors. That reversal is the load-bearing beam under the entire business. Everything QTL does today sits on top of it.
Step back, and the pattern is almost funny. Regulators kept winning the argument. Qualcomm kept winning the business. These standard-essential patent fights are a permanent feature of the industry, not a Qualcomm quirk; OPPO spent two years locked out of Germany and the UK over Nokia’s SEPs during the same stretch. What makes Qualcomm’s version singular is the scale, and how little the fines actually changed the machine.
Apple is the cleanest example. It sued in 2017, calling the fees extortionate, dragged Qualcomm through courts on several continents, and then in April 2019 simply paid: a one-time settlement reported at roughly $4.5 billion, a six-year patent license, and a multi-year deal to keep buying Qualcomm modems. The war ended in an afternoon with a signature, the way these things usually do.
Here’s what a decade of headlines got wrong. The fines were never going to kill QTL, and they didn’t. The thing that actually threatens the royalty machine is much quieter, and it’s already happening.
Apple shipped its own modem, the C1, in early 2025 in the iPhone 16e. Six years of work and a bought-out Intel modem division went into it. The chip is competent: good power efficiency, lower peak speeds than Qualcomm’s part, no millimeter-wave yet. A second-generation version is expected around 2026 to close some of that gap. Qualcomm’s own stated planning assumption has been that it supplies only about 20% of the modems in the 2026 iPhone. So the chip revenue from its largest customer is draining away on a schedule everyone can read.
This is where QTL stops looking like the chip business and starts looking sturdier than it. Apple building its own modem does not end Apple’s royalty payments. That modem still has to implement 5G, and Qualcomm holds standard-essential patents on 5G, so Apple can stop buying the silicon and still owe the license. The current agreement runs to March 2027. Which means the real cliff was never the C1. It’s the 2027 renegotiation, when Apple sits down to re-price a license covering patents it is now practicing in its own chip, represented by the same lawyers who sued the first time.
The base underneath all of this is smartphones, and smartphones stopped growing years ago. Patents expire on a rolling basis. The Huawei license lapsed, and Huawei dropped out of the numbers entirely. New signings keep the total roughly flat, around $5.6 billion a year, with Transsion and a couple of Chinese makers re-upping, but “roughly flat” is the whole story. This is a mature royalty stream doing an excellent impression of a stable one. And a mature royalty stream throwing off billions in high-margin profit is exactly the thing you use to fund an escape from your own dependence on it. The pivot into data centers, laptops, and automotive that I wrote about earlier reads less like a confident company reaching for more and more like QTL’s cash buying Qualcomm a second act while the first one is still paying out.