Qualcomm unveiled the Snapdragon 835 in November as the first mobile processor built on a 10nm process, and that is the smallest part of what is going on. The bigger story is that two of the most consequential Qualcomm announcements of the year have almost nothing to do with the phone in your hand. One is a 5G modem that exists before 5G networks do. The other is a deal with Microsoft to put Snapdragon inside Windows laptops (WinHEC). Put the three together and the pattern is unmistakable: Snapdragon is trying to leave the smartphone it conquered.
Take the 835 first. It is built on Samsung’s 10nm FinFET process, the first chip anywhere to ship commercially on that node, and Qualcomm’s pitch is a package that is meaningfully smaller and more power-efficient than the 820 it replaces, with an integrated X16 modem that does Gigabit LTE, Bluetooth 5, and the usual generational bumps to the Kryo CPU and Adreno GPU. Qualcomm is holding the granular spec breakdown for CES in January, which is its own tell about how confident it feels, but the two facts it led with, first to 10nm and Gigabit-class LTE on the die, are the ones that matter. Smaller and cooler means a phone maker can spend the saved millimeters on battery or thinness, and the modem keeps doing the integration work nobody else matches.
Then there is the X50, which Qualcomm announced back in October as the world’s first 5G modem. There is no 5G network to plug it into. The standard is not even finished. Announcing a 5G modem in 2016 is Qualcomm planting a flag years before the territory exists, and it is exactly the move that built the company: own the modem, own the transition, make every carrier and OEM come through you when the new generation finally lands. I am skeptical of any 5G timeline anyone gives me right now, because the gap between “we have a modem” and “you can buy a phone that uses it” is measured in years. But the X50 is Qualcomm doing the thing it is structurally best at, getting to the next radio generation first.
The one that actually surprised me is Windows. At Microsoft’s WinHEC event in Shenzhen this month, Microsoft and Qualcomm announced that full Windows 10, the real desktop one, will run on Snapdragon, with an emulation layer so existing x86 apps work without being rewritten. We have seen Windows on ARM before and it was a disaster: Windows RT could not run normal desktop software and died for it. The difference now is the emulation. If Win32 apps actually run on a Snapdragon laptop, then the pitch writes itself, a fanless, always-connected PC with multi-day battery and a SIM card, built on phone-chip efficiency. That is a real product category that does not exist yet, and Qualcomm wants to invent it. I want to see the emulation perform before I believe it, because emulating x86 on ARM is the kind of thing that demos beautifully and stutters in real life. But the ambition is correct.
For context on where Qualcomm sits today, the current flagships, the Galaxy S7 and Google’s Pixel and the OnePlus 3 and 3T, all run the 820 or its tweaked 821 sibling on Samsung’s 14nm node, and they run it well, the comeback from the 810 worked. Apple’s A10 is the chip to beat on raw single-thread, Huawei’s Kirin 960 is closing fast, and Samsung’s Exynos keeps the 835 honest. The phone race is still the day job, the one that filled the CES floor back in January. I am leaving Qualcomm’s automotive cockpit ambitions for another post, because between 10nm, a 5G modem, and a Windows laptop play, this chip already has too many storylines.
What I keep thinking is that the 835 itself is almost the least interesting announcement in its own launch window. It is a very good phone chip on a leading node. But the X50 and the Windows deal are Qualcomm saying out loud that the smartphone is a solved problem and the growth is somewhere else, in PCs, in cars, in whatever 5G turns out to be. Whether any of that pays off is a question for years from now. The flag is planted today.