Qualcomm buying Nuvia is a bigger deal than you think

Qualcomm logo during CES 2018
Qualcomm logo during CES 2018 (Image credit: Android Central)

Qualcomm recently announced that it intends to spend $1.4 billion to acquire Nuvia, a small start-up that specializes in building high-speed ARM server chips. Any deal of this nature would be subject to regulatory investigation, but this one will get a bit of extra scrutiny — Nuvia was founded by ex-Apple chip designer Gerard Williams who left the company under an agreement that he does not compete with Apple's chip business.

Building servers was fine with Apple because the company doesn't try to make any serious gains in that market. Instead, it spends resources on chips for consumer electronics, like phones or laptops. And it just so happens that Qualcomm has said the acquisition of Nuvia will provide tech for high-end chips for consumer electronics like the best Android phones as well as servers and self-driving cars.

The deal itself may be clouded by Apple's lawsuit against Williams and Nuvia, but I really don't think the sole purpose of trying to "disrupt Apple's business model" is in play here. Qualcomm does not want to make something just like Apple's A-series mobile chips. If it did, it would already be doing it. Qualcomm wants to win the fight when it comes to laptops.

Plenty of big-tech companies build servers now.

Name a tech company besides Apple and you'll likely name on who does build its own ARM server chips. Google, Samsung, Amazon, Facebook, you name it — they build them. That's because building and designing a high-performance ARM chipset isn't near the impossible side of the difficulty scale. I'm not taking away from the geniuses who are working to design ARM chips — they all know that things really get hard when you are stuck using battery power.

Qualcomm Snapdragon 888

Source: Qualcomm (Image credit: Source: Qualcomm)

That's where Apple has spent a good bit of R&D to build mobile ARM chips that are very powerful, even when compared to the latest chips from companies like Intel or AMD. But so has Qualcomm, though the two companies have very different business models.

Apple and Qualcomm have very different business models when it comes to making smartphone chips.

Apple would be ecstatic if every iPhone used the best and greatest A-series chip. To the company, the costs of adding a powerful ship to a "budget" class phone is outweighed by having a single model of SoC to build and maintain, as well as handle supply chain and production issues. This type of forward-thinking likely continues in it's new M-series ARM MacBook chips. One model, powerful enough to do everything actually saves a lot of money in the long run.

Qualcomm doesn't make the products that use its chips. It builds chips for mobile devices that its customers — companies like Samsung, Dell, HP, LG, and every other name in consumer electronics you can think of — want. And Samsung or Dell or whoever does not want to spend the money to buy something overpowered like an Apple A-series mobile chip. Companies building phones and other mobile products only want a chip that's strong enough to do whatever they need it to do today and that's why the Snapdragon line of mobile chips is so popular and successful.

Maybe Qualcomm wants to get into the server CPU market. It's surely pretty lucrative, especially if you are already known as a company that can build chips to order that are exactly what you need. Maybe.

Snapdragon 8cx Chip Front

Source: Qualcomm (Image credit: Source: Qualcomm)

What I think Qualcomm wants to do here is borrow some of the core design magic to boost its Snapdragon chips for phones and tablets, but mostly churn out a new breed of mobile chip designed for something like an ultralight Windows laptop or a high-end Chromebook.

In my eyes, this move is 100% about the future of mobile computing and 5G laptops powered by Snapdragon chips.

Grabbing up a company headed by an ex-Apple ARM chip designer would be a really good way to get started. Being able to partner with Microsoft, Google, Amazon (who would love to have a FireBook laptop) and build chips that have the power to do exactly what an operating system needs to do and do it the most efficient way possible is how and why Apple's M-series chips are so good when they are inside an Apple laptop.

An ARM chip designed with Microsoft and built for a Surface book, or a project with Google for an all-new class of Pixel is what Qualcomm sees in its headlights. That's the only thing that makes sense.

And Apple is right to be pretty upset about it. Expect some courtroom drama with this one folks.

Jerry Hildenbrand
Senior Editor — Google Ecosystem

Jerry is an amateur woodworker and struggling shade tree mechanic. There's nothing he can't take apart, but many things he can't reassemble. You'll find him writing and speaking his loud opinion on Android Central and occasionally on Twitter.

2 Comments
  • Qualcomm just wants a M series cpu
  • 0. Apple has already stated that their lawsuit is against Williams only, and they have no intentions of stopping Nuvia and its other employees from designing and selling chips. 1. I really do not think that Qualcomm merely wants this just to be able to make better laptop and desktop chips for Chromebooks and Windows PCs. The reason is that they could accomplish this on their own in a couple of years merely by coming up with a custom design that merges the best parts of the Cortex X1 and the Cortex A78C. The Cortex X1 provides single core performance that is far more competitive with the M1 Firestorm performance cores than the Apple fan club wants to admit. When the Cortex X1 gen 2 comes out and addresses bottlenecks with caches and instruction sets that they weren't able to resolve by the gen 1 deadline, it will be even more competitive. More important, the Cortex A78C will FINALLY allow designs of more than 8 cores, which this group has been stuck on since 2015. (MediaTek tried a 10 core design in 2017 and it actually in some respects performed worse than the 8 core designs did). So by 2022 we could potentially see a big.LITTLE design of 4 Cortex A57 efficiency cores and EIGHT Cortex X1 gen 2 performance cores that would give a multicore benchmark surpassing the Intel Core i7 desktop chips ... and get it on a 3nm process no less. In other words, buying Nuvia wasn't necessary: Qualcomm or Samsung could already issue 12 core Chromebook and Windows 10 chips with 8 Cortex A78C performance cores and 4 Cortex *** efficiency cores on a 5nm process by 3Q 2020 if they wanted to. The reason why they don't is that such a chip would use too much battery to be viable as a mobile phone processor and neither of them take tablets or laptops seriously enough to put that sort of effort into anything other than a phone. 2. So yes, this actually may be for servers. Note that Marvell left the ARM server business, leaving Ampere as the only major player and HP as a minor one. (AMD continues to be "evaluating" whether they will make ARM chips, but since they will catch Apple on a 5nm process for their x86-64 chips by 4Q2021, they may not need to.) Please realize that Qualcomm doesn't merely make mobile chips. They make chips for IoT and edge devices too. Who is the main competitor in that arena? Nvidia. WHO JUST BOUGHT ARM. But even if not wanting to rely on ARM designs anymore isn't a motive - though it could be - going from supplying chips for an edge server (which Intel also makes CPUs for) to chips for workstations, data center servers and even cloud servers ... is simply a matter of performance. Qualcomm can easily get customers who already buy their edge chips to buy their server chips if those chips are powerful enough. By contrast, while Nuvia has chips that are powerful enough in theory, their issue is finding buyers for them, a problem that Marvell had that Qualcomm doesn't.