Qualcomm Sells Chips. The Profit Is in the Patents Behind Them.

Open up almost any 5G phone and you will find a Qualcomm chip doing the hard work of turning radio waves into the bars in your status bar. That chip is the part of Qualcomm everyone can see and name. But the chip is not where the company makes its money. Even if a rival had built the modem inside your phone, Qualcomm would still get paid — because a slice of the standard itself, the math that lets any phone talk to any tower, is patented, and Qualcomm owns a large stack of those patents.⁸ It collects on the standard whether or not it sells you the silicon. The chip is the storefront. The patents are the vault.

The official story is that Qualcomm is a chipmaker that happens to license some patents on the side. The truth runs the other way: Qualcomm is a patent-royalty machine that happens to run a giant, low-margin chip business — and the licensing unit quietly pays for the silicon, not the reverse.

The 14% of revenue that earns 30% of the profit

Qualcomm splits into two engines. QCT designs and sells chips: handsets, automotive, IoT. QTL does nothing but license patents and collect royalties. In fiscal 2024, QCT's handset chips alone pulled in $24.9 billion, automotive $2.9 billion, IoT $5.4 billion — a roaring, capital-hungry semiconductor operation against $39.0 billion in total company revenue.¹² Next to it, QTL looks tiny: about $5.6 billion, roughly 14% of sales.³ But that framing is the trap. QTL's pre-tax margin runs near 72% — roughly $4.0 billion of pre-tax earnings on roughly $5.6 billion of revenue — which means a unit that produces one-seventh of the revenue produces close to a third of the entire company's profit.³ The chip business sweats; the patent business prints.

Here is why the margins diverge so violently. The patents are a fixed cost paid years ago in R&D and never paid again. Once a Qualcomm invention is written into a global cellular standard, every handset that touches that standard owes a royalty — whether the maker buys Qualcomm chips or a competitor's. WIPO's PCT Yearly Review 2025 ranked Qualcomm third in the world for PCT applications published in 2024, with 3,848 filings,⁹ and the portfolio it has accumulated over decades is the kind of asset that costs nothing more to license to the next billion phones. A chip has to be fabricated, shipped, and supported. A royalty has to be invoiced. That is the whole difference between a 72% margin and a thin one.

Why you can't just split the chips off from the patents

Every few years someone proposes the tidy fix: spin QTL out as a clean, high-margin licensing company and let QCT be a normal chip business. It sounds obvious and it is financially illiterate. The two are co-dependent. The licensing engine throws off the high-margin cash that funds QCT's brutal R&D and fab commitments; the chip business keeps Qualcomm at the bleeding edge of the very standards QTL licenses, which is what makes the patents essential rather than expirable trivia. Separate them and you get one company that licenses a slowly aging patent stack with no fresh inventions feeding it, and another that competes in a low-margin chip market without the war chest that let it lead. The thinness of QCT is not a weakness to be amputated. It is the thing QTL exists to subsidize — and QTL is the thing QCT exists to keep relevant.

But didn't the FTC prove this whole thing is an illegal racket?

The fair objection is the strongest one against this model: that the royalty isn't earned, it's extracted — that Qualcomm's 'no license, no chips' policy coerces handset makers into paying for patents they'd otherwise dispute, and that a federal court said so. The first half is a real critique. The second half is wrong. The FTC did win at the district court in May 2019, but on August 11, 2020 the Ninth Circuit unanimously reversed that judgment and vacated the worldwide injunction.⁴ The FTC then declined to take it to the Supreme Court, and its request for rehearing was denied.⁵ The legally operative outcome is the opposite of the headline most people remember: Qualcomm's core licensing practices were found not to violate antitrust law.

The reasoning matters more than the result. The appeals court drew a line the press narratives kept blurring: handset makers are Qualcomm's customers in the chip market, not its competitors in it. Charging customers a lot may be hard on them, but in antitrust terms that's 'hypercompetitive,' not 'anticompetitive' — economic harm to a buyer is not the same as harm to competition. That distinction is the load-bearing wall under Qualcomm's whole model. It's why the royalty survives the obvious attack: a buyer paying more is not, by itself, a crime.

Apple was the one licensee with the leverage to test the model, and it did — through years of litigation, the FTC trial, and testimony from Apple's COO that it paid roughly $7.50 per handset,¹⁰ roughly four times below the ~$30 figure that circulates online with no primary source behind it. Then, three months before the Ninth Circuit even ruled, Apple settled, signed a six-year license, and a multiyear chip supply deal on top of it.⁶⁷ Both companies confirmed only that Apple paid Qualcomm; neither disclosed how much, and any specific figure you've seen is unverified. The signal in that settlement is louder than any number: the company best positioned to break the royalty looked at its odds and chose to keep paying it.

So strip away the silicon and the courtroom drama, and what's left is a machine that was built decades ago and costs almost nothing to run: a stack of inventions written into the standard that every phone must speak, collecting a toll on the entire industry — its own chip customers and its competitors' alike. The chips are the part you can hold. The genius is the part you can't: Qualcomm didn't just build a better modem. It got its ideas written into the rules, and then charged the world for following them.