Qualcomm Doesn't Win by Building Better Chips. It Wins by Owning the Rules.

Somewhere right now, a handset rolls off a line in Shenzhen, gets boxed, and is sold half a world away. Qualcomm may not have made a single component inside it. It doesn't matter. A sliver of that phone's wholesale price still flows back to San Diego, because the phone speaks a language Qualcomm helped write into the standard, and Qualcomm collects a royalty calculated as a percentage of the whole device—not the modem, the whole phone.⁵ It is one of the strangest positions in technology: a company paid for what's invisible inside a product it never touched.

The official story is that Qualcomm is a chip company that also happens to hold some patents. The truth runs the other way. The chips are the visible, lower-margin business; the patents are the quiet machine that prints the profit. And the genius isn't any single invention. It's a loop, deliberately re-run across four generations of cellular technology, that turns every phone on earth into a recurring payment.

Seven engineers, one risky bet, and a tracking system for trucks

Qualcomm wasn't founded to license patents. It was founded in July 1985 by Irwin Jacobs, Andrew Viterbi, and five colleagues from Linkabit—seven people, not the two the legend usually names—and its first commercial product was OmniTRACS, a satellite system for tracking trucks.¹ The famous part came later, and it was anything but inevitable. CDMA—the spread-spectrum approach Qualcomm wanted to bring to civilian cellular—was widely seen as promising but risky, the kind of technology that needed money cobbled together from many sources, including government small-business research grants, just to prove it worked.² Jacobs has been honest about the lineage, too: spread-spectrum techniques came out of decades of prior work, not a single eureka in San Diego. Qualcomm's real contribution was narrower and harder—making the thing commercial.

And here the popular history gets the next beat wrong as well. The story goes that the standards body 'adopted CDMA' and swept away the rival TDMA approach. It didn't. TDMA-based D-AMPS was already the first North American digital interim standard, adopted in 1989. Qualcomm proposed its CDMA system in 1990, and in July 1993 it was approved as a second, parallel interim standard—IS-95, later called cdmaOne.³ CDMA didn't win the standards war by knockout. It won a seat at the table. The first commercial CDMA network didn't even launch in the US; it went live in Hong Kong in 1995, then South Korea, then America.⁴ What Qualcomm understood, earlier than almost anyone, was that a seat at the table was enough—if you owned the patents the table now depended on.

The licensing tail wags the chip dog

Strip away the narrative and look at the ledger, because the ledger states the thesis more bluntly than any phrase can. Qualcomm runs two main businesses: QCT, which designs and sells chips, and QTL, which licenses its patents.⁵ In fiscal 2024, QCT brought in $33.196 billion of revenue at a 29% pre-tax margin. QTL brought in $5.572 billion—about a sixth as much—at a 72% pre-tax margin.⁶ Sit with that contrast. The business that's a fraction of the size, that makes nothing you can hold, earns more than twice the margin of the one with fabs and engineers and the Snapdragon name on the box. That is not a chip company with a licensing sideline. It's a licensing engine with a chip business attached for reach.

How the same loop wins four times in a row

Here is the mechanism, and it's worth seeing as a loop rather than a list. Every new cellular generation—2G, 3G, 4G, 5G—requires a fresh standard, written by industry bodies so that phones and networks from different makers can talk. Qualcomm pours research into each generation early, gets its techniques baked into the standard, and ends up holding patents that are, by definition, unavoidable for anyone who wants to be compliant.⁵ Then it collects. The royalty is keyed to the wholesale price of the finished device, so as phones got more expensive and more numerous, the take grew with them—no extra invention required. The chip business amplifies the loop: by selling leading modems, Qualcomm stays at the frontier of the next standard, which feeds the next round of essential patents. Better silicon isn't the moat. The silicon is what keeps the company embedded where the next set of rules gets written.

Didn't the FTC prove this was illegal?

The fair objection is that this isn't strategy—it's coercion dressed up as a business model, and a federal court agreed once. The lever is a policy known as 'no license, no chips': you cannot buy Qualcomm's modems unless you also take a patent license. In May 2019, a district court ruled exactly that, finding Qualcomm held monopoly power in certain modem chips and that the policy forced handset makers into anticompetitive deals, and it issued a worldwide injunction.⁷ For a moment, the whole machine looked illegal.

Then it didn't. On August 11, 2020, the Ninth Circuit unanimously reversed the ruling and vacated the injunction, finding no antitrust violation.⁷ The honest reading is that the court drew a line many find uncomfortable: hard, even aggressive, isn't the same as illegal. Qualcomm's core practices survived intact, which is precisely what makes them a strategy worth studying rather than a cautionary tale. The model wasn't blessed as fair; it was found to be legal—and in a standards war, legal-and-unavoidable is the whole prize.

But surely 5G changes the math?

The sharper counter is forward-looking: the 3G and 4G generations were Qualcomm's to dominate, but 5G is crowded. That's true, and the data is humbling for anyone claiming Qualcomm 'owns' 5G. Independent rankings can't even agree who leads—Huawei, Qualcomm, Ericsson, Samsung, and Nokia all surface at the top depending on whose methodology you trust—and experts estimate only 10 to 20% of declared 5G essential patents are actually essential, which makes raw patent counts a poor proxy for licensing power.⁸ So the embellished version of the Qualcomm story—a clear majority of 5G patents—is simply false. But notice what the crowding does and doesn't threaten. It pressures Qualcomm's share of any single generation. It does not unwind the loop, because Qualcomm's position isn't one generation deep; it spans CDMA2000, WCDMA, TD-SCDMA, 4G and 5G at once.⁵ A rival who out-patents Qualcomm on 5G alone still has to license the rest of the stack a phone needs. Depth across generations is the moat that a single crowded generation can't drain.

Qualcomm makes its money the way a tollkeeper would if the toll were written into the road's blueprint. It doesn't need to make the best phone, or even sell you a chip. It needs only one thing: that the world keep using cellular standards it helped author, on devices whose price it gets a slice of. Spread-spectrum was someone else's idea. The chips can be matched. The court fight was nearly lost. None of it touched the real asset—the decision, made and re-made across forty years, to be in the room where the rules get written, and to want nothing from each phone but a sliver of its price.