The Snapdragon history I care about does not start with a chip. It starts with a branding call in 2007, when Qualcomm took a pile of forgettable part numbers, MSM this and QSD that, and folded them under one name a marketing team could actually sell. I have spent most of two decades watching that name climb from spec-sheet footnote to the thing people ask for at the carrier counter, and lately to a chip trying to force its way into Windows laptops and XR headsets. This is the throughline, including the one year Qualcomm would rather nobody remembered and a bet that still has not paid off.

The name nobody asked for, until they did (2007 to 2009)

Qualcomm unveiled the Snapdragon name in late 2007 around the QSD8250, pairing a custom CPU core it called Scorpion with a GPU named Adreno. The Adreno detail is the one I keep coming back to: it is an anagram of Radeon, because that graphics block was AMD’s handheld GPU team, which Qualcomm bought outright and quietly turned into one of the most durable advantages it owns. The first real devices did not land until 2009, and you can read the announcement that put the whole lineup under one roof in the version written at the time. Nothing about it looked like the start of a twenty-year run. I still covered the devices equipped by Qualcomm, and not really the SoC itself. One of my mistake by then.

The decade I went back and filled in (2010 to 2018)

The S-series years run from the Adreno 205 quietly carrying the S2 generation through the 3.1 billion dollar Atheros buy and the first promise of a custom Krait core to the S4 putting an LTE modem on the die, the integration move that still defines the company. Then the Snapdragon 800 burned the S-numbers and reset the whole scheme in 2013. One chip here matters more than the rest, and not in a good way: Qualcomm dropped its own core to answer Apple’s 64-bit jump with the 810, whose Cortex-A57 ran so hot on early 20nm that phones throttled within minutes, reviewers cooked them on camera, and Samsung skipped it entirely for the Galaxy S6. That snub was the loudest vote of no confidence Qualcomm ever absorbed, and it taught the company a lesson about chasing a node too early that it would forget all over again later. The climb back ran through the 820 bringing the custom Kryo core home, the 835 going first to 10nm with 5G and Windows on Arm coming into view, and the 845 trading raw speed for an AI story, before 2018, the year Qualcomm fought off a 44 billion dollar hostile takeover and shipped anyway.

5G, and a lineup that would not stop multiplying (2019 to 2021)

This is where my coverage gets dense, because the lineup exploded. 2019 alone brought the 855 and then the 855 Plus for the gaming crowd, the 8cx aimed squarely at Windows laptops, the 730G I lined up against MediaTek’s gaming Helio, and the 735, the first Snapdragon to drop 5G into the mid-range. On the modem side, the X55 was quietly finishing its testing, and the discrete-5G approach it represented would shape phone design for the next two years.

Then the naming went sideways. The chip the whole internet called the 875 shipped as the 888. There was an 888 Plus with an AI-flavored pitch to match, an 870 that was an 865 wearing a new badge, a 778G and 780G for the middle, and a 480 that finally dragged 5G down to budget phones. Capping it, the X65 modem set a 10-gigabit record that mattered more for Qualcomm’s licensing leverage than for any phone you could actually buy. By the end of 2021 there was a Snapdragon for roughly every thirty-dollar band of the market, and keeping the names straight had become a part-time job.

The rebrand, and AI stops being a slide (2022 to 2024)

Qualcomm reset the whole scheme to “8 Gen 1” and immediately tripped over it: the first one ran hot on Samsung’s 4nm before a mid-cycle move to TSMC cooled things off. The recovery is the part worth your time. Everything we knew about the 8 Gen 2 before launch pointed at a chip that finally delivered, and it did. The 7 Gen 2 arrived at MWC 2023 to push the better silicon down a tier, and by that autumn the 8 Gen 3 and the on-device generative-AI sell at Snapdragon Summit made it clear where Qualcomm thought the next fight was. The pitch was no longer raw benchmarks. It was running a model on the phone without a round trip to a data center.

Snapdragon leaves the phone (2024 to 2026)

The most consequential thing Qualcomm did this decade was buy a CPU company. Nuvia, picked up in 2021, became the Oryon core, and Oryon is what finally let Snapdragon stop customizing Arm’s reference designs and start drawing its own. It showed up first in laptops as the Snapdragon X line, then in phones as the 8 Elite, which pointedly dropped the “Gen” from its name. Two years into the PC experiment, the question of whether the platform is actually ready is still open, and I lean skeptical on the software long before the silicon. Alongside it, Snapdragon Reality Elite stakes out the XR generation, and the two new mobile platforms from this spring keep pushing the same 5G-and-AI story down into cheaper devices. I am not touching the Apple modem divorce here, by the way. That is its own post, and a messy one.

The scorecard

Watch Qualcomm across these twenty years, and the throughline is not the phone; it is the refusal to stay inside it. The company that started by cramming a modem and a GPU onto the same die as the CPU spent the back half of that run carrying the same low-power trick everywhere it could reach. It went into cars, where the Snapdragon Digital Chassis turned an infotainment pitch into a whole platform, and it even wandered into the data center, with the Cloud AI 100 running inference nobody would associate with a phone company. The move I keep circling back to is Nuvia. Buying that team gave Qualcomm the Oryon core, and that acquisition still strikes me as the smartest thing the company did all decade: Oryon is the reason a Snapdragon laptop can finally trade blows with Apple Silicon instead of apologizing for itself. Somewhere in the same stretch, they quit treating AI as a cloud round-trip and bolted real NPUs into phones and laptops so the model runs on the thing in your hand. Then the part almost nobody clocks: Dragonwing and the Arduino tie-up, the VENTUNO Q, pushing that same efficient silicon down to makers soldering in a garage and out to factory floors running robots. I am not convinced “intelligent edge” is much more than a slide title yet, but the hardware under it is real, and it is the same bet Qualcomm has been making since 2006: put the compute as close to the work as possible, and make it sip power getting there.

The full list, chip by chip

I left out of this list all the devices and their launch posts because they were more focused on the device itself rather than the chip inside, as I mentioned. Plus too many posts, so the list would have been bloated at some point.

Two decades in, I keep landing on the same read: Qualcomm is at its best when nobody is watching the part it is actually good at. Now they are at full throttle in Data Servers, AI, Computing, and EVs, and over time they have become a fully horizontal company, covering different sectors with scalable products. I still think that is the best approach for generating revenue and showing their craft.