Skip to main content

Google designing its own phone processors would completely change the industry for the better

Google Campus Logo
Google Campus Logo (Image credit: Android Central)

Rumors of Google designing and manufacturing its own chips for Pixel phones are nothing new. But the latest comes at a time when it makes a lot more sense for it to be done; Google needs its own chips if it wants to carry on making products for everyone, and making them the way it wants them to be built.

Source: Google (Image credit: Source: Google)

This isn't complete conjecture, either. Google already designs the processors used in many of its server products and machine learning data centers. It also has designed a complete ARM processor for the Pixel phones that started as the Pixel Visual Core for camera features and has since moved into machine learning as the Pixel Neural Core. The company has poached some of the best in the industry from Qualcomm, Intel, and Nvidia to work in its new Bengaluru, India microprocessor design lab. That's not an accident.

If you build the chip writing the code gets easier.

Google does this for a very specific reason: customizing a chip means it works exactly the way Google wants it to work. And that is like music to the ears of a software company. Having control over the hardware so that high-performance software can be specifically tailored to a specific piece of hardware opens new possibilities and means less overhead.

In addition to not being reliant of Qualcomm for microprocessor code and updated firmware, this also means Google can decide the update path for future Pixel phones and Chromebooks should it build a custom SoC (System on a Chip). When a manufacturer designs a new phone it must secure a contract with chipmakers to provide updates. When you are the chipmaker, as we see with the Nvidia Shield TV, you can update products for as long as you like.

The technical side of things is very important for Google, but equally important is any cost reduction that could arise from building its own processors.

Pretty Pixelbook screen

Source: Daniel Bader / Android Central (Image credit: Source: Daniel Bader / Android Central)

Google is very happy that Samsung makes Android phones, or that ASUS makes Chromebooks. Google is in the eyeball business; the more eyeballs on the internet and using Google products, the more money Google makes. But Google also sells hardware, and the success of the Pixel 3a shows that Google needs to pare down prices if it wants to compete with Samsung and Apple. The 2020 version of the iPhone SE costs $399 in part because Apple doesn't have to purchase its A13 processors from another vendor.

The Pixel 3a is the best Android phone you can buy under $400

The Pixel 4 and the Pixelbook Go are wonderful products. They certainly aren't perfect, but each has a dedicated fanbase that loves them because of how well they work and their simple yet elegant design. But sales of both combined likely aren't near the numbers we see from the Pixel 3a. The reason? The Pixel 3a is affordable and works 90% as well as any other Android phone you can buy.

Google Pixel 4 XL in hand

Source: Joe Maring / Android Central (Image credit: Source: Joe Maring / Android Central)

The Pixel 3a was successful because Google was able to make its in-house version of Android run extremely well on cheap off the shelf hardware. The Pixel 4a should be similar, and we can expect to see a device with mid-range specs perform well enough that some will hail it as one of the best phones you can buy. Were Google to have more control over the internal hardware by building out its own SoC, a future Pixel flagship phone could run as well as Samsung's finest with hardware that's not nearly as powerful or expensive.

With an in-house chip we could see "budget-priced" Pixel phones that rival the best from Samsung.

This benefit can happen because Google writes its own operating system. Unlike Apple, which seems to be designing an ARM chip powerful (and expensive) enough to run a MacBook Air — and we will see an Apple A-series CPU in the MacBook Air within three years — Google could do the opposite and pare things down to balance performance against costs.

A $600 Pixel 6 XL that runs as well as a $1,000 Pixel 4 XL and brings new smart features to Android is the goal. Building its own in-house chip is one of the best ways to get there.

Jerry Hildenbrand
Jerry Hildenbrand

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.

30 Comments
  • This is why Apple's iPhone hardware and software is so seemleas so successful as they design the processor as well as writhe the iOS software themselves and Google doing this would get them close to par with Apple because as it stands they like every other Android OEM have to rely on Qualcomm.
  • > The 2020 version of the iPhone SE costs $399 in part because Apple doesn't have to purchase its A13 processors from another vendor. The other part would be that they are already making quite a few of those and have done so for about a year. I am sure they were not in a position to put that chip into $399 phone last fall, no matter who the chip was designed by. And that's the part Google will be missing on, so I would not hold my breath for the $600 Pixel 6 XL with the custom CPU.
  • Even if Google does start making their own SoCs they still won't be in Apple's league, even Samsung's own Exynos series is struggling against the might of Apple's A series which is due to Samsung having to rely on Google for the OS and then bloat it out with their bloated custom akin on top of Android.
  • " even Samsung's own Exynos series is struggling against the might of Apple's A series" In what way specifically? What specifically runs worse on a modern Galaxy phone, compared to an iPhone?
  • The GPU in the Exynos series is always inferior to the ones in Apple's A series and so is Qualcomm's but at least Qualcomm is closing the gap on Apple unlike Samsung.
  • Hmmm...my S10e is a smokin good phone!
  • The note 7 was truly a smoking good phone. Lol
    Sorry, I know that crap is old.....lol
  • Based on the statement ($1,000 vs $600 price) are you saying Qualcomm is charging $400 per chip?
    Hopefully with that saving, we can get front facing speaker (Pixel 2 XL) again.
  • Lol Google dropping the price of their flagship from 1000---- 600, yeah right. If anything they will use these mid range chips they design to keep their pixel a series devices cheaper. 
  • Yeah, it's certainly done wonders for Samsung, Exynos models are craved and lusted after in regions that don't have them lol.
  • This. And aside from the performance, it's not like Exynos ever did anything for prices.
  • The great illustration of your point was Galaxy Tab S2 9.7 (probably more, but this is the one, I have the firsthand experience with): T813 (Qualcom) and T810 (Exynos) were sitting on the same shelf at the same price in the identical packaging -- the only difference was the actual model number in the small print. I was hunting for T810, because Exynos was supporting MHL video out and Qualcom did not...
  • Samsung isn't a good comparison because they don't control the OS. They can't have the same vertical integration that Apple has or Google could have unless they have their own OS. Sure they have Tizen but that's not anywhere near ready to compete with the established and mature OS's that are ios and Android...even if they could get app developers on board (they can't).
  • While Samsung can make some very good individual parts, the company is shit when it comes to assembling them into a cohesive unit then supporting them with quality software. An Exynos SoC is excellent on something like a small single-board computer mostly because other companies write an operating system that supports it well. Using them in an Android phone requires more support than Samsung was willing to provide if performance was the goal. IOW, Google isn't going to do all the work that Samsung should be doing. This wasn't always the case and some of the early Exynos SoCs were more open and performed much better than other mobile processors. The Nexus S is a great example.
  • > While Samsung can make some very good individual parts, the company is **** when it comes to assembling them into a cohesive unit then supporting them with quality software. While the assessment of the quality of the final product is, certainly, in the eye of beholder, I am not sure that you can pin that on the CPU. I own two otherwise identical Samsung devices, one with Exynos and one with Qualcomm and, apart from one of them supporting Miracast and the other being capable of outputting video via MHL connector, I find little difference in the consumer experience. Maybe both of them are "****" in your opinion (I do like them), but not because of the CPU. OTOH, as the owner of the few Nexii and someone, who is using OG Pixel XL as the daily driver, I would have to say that the quality of the final product that came out of Google... varies. You seem to be of opinion that the custom CPU will fix that... I guess, the time will tell.
  • I really don't think this will help. Google can't compete with Apple on the high end. Google won't compete with Apple on the low end. Plus, it's presumptuous to think they'd pass any savings along to consumers.
  • Plus its Google. They'll probably give up on it like they do everything else lol.
  • I wonder if them designing their own chips will also make it possible for them to better control battery management. Get more battery life out of a smaller battery
  • It absolutely could.
  • When I saw the battery size for the Pixel 4 and 4XL I thought. Google has to have something up their sleeves that isn't ready yet but hoping to be before the next launch that would optimize the OS to better manager the battery usage.
  • Do they put the Pixel Visual core in anything? :/
  • Time to break Qualcomm's monopoly. I'm sure this is one reason why phones are getting more and more expensive.
  • Really? Did you compare prices of Galaxy S20 with Exynos and Qualcomm?
  • Sure, so Google could just sell ppl processors now that are three generations behind, just like they do their phones. I see Google as always in their five year plan. Not giving what they could and what the people want, but saving features for the next release and so forth.
  • This chip isn't being designed for Android or ChromeOS. It is being designed for Fuchsia. Android and ChromeOS are Linux OSes with monolithic kernel. Fuchsia is going to be a microkernel OS. From what I have read, microkernel OSes are capable of starting off very minimal and then adding to it only what is needed - sometimes at runtime - and as a result are very scalable to the point where they can start with a barely visible IoT device and scale up to a data center server because everything on it is served as/consumed as a service that are layers (or modules?) that you can add or remove as needed. You can see why such an OS would have conceptual appeal to what is still 90% a cloud and Internet services company. The downside is performance and memory management issues if you don't tune the hardware for it. This site doesn't talk about Fucshia much - though it should as Android (or some container like application layer construct) - will definitely be one of its services. So this is totally different from what Microsoft and Apple have. All their operating systems - from the Windows CE and iPod stuff that was basically embedded firmware to the big data cloud stuff that Microsoft uses for Azure - are monolithic kernels and all CPU hardware is based on/designed for it. Google is going to have the challenge of designing a CPU to maximize performance and stability for the first widely and commercially available microkernel OS, or perhaps a bit more accurately a microkernel OS that is more than embedded firmware. It is curious how they are going to handle it. I have read, for example, that for ChromeOS the Android container launches each Android app instance its own virtual machine that lives for the duration of the app and is destroyed when the instance terminates, with only the data being preserved. Will Fuschia do something similar by creating a server for each Android app? Or will the server be launched for the Android container, who will then launch servers (or servlets since most Android apps are Java-based?) for each Android app instance? And how will Linux be handled and the applications - bigger, more complex and resource intensive than nearly all Android apps - also? Back in the day there were a bunch of neat-but-no-useful-real-world-application "turn your old Nexus into a server" type articles like this one: https://www.digitaltrends.com/mobile/how-to-make-an-android-server/ for example. Well if Google can get the OS and hardware to its potential, a (relative to the cost to technology) cheap Fuschia device <b>could actually be a server</b> with other computing devices being clients with capabilities far more than using your Google Home app on your smartphone while in the living room to play a show on the TV connected to the Chromecast in the basement . If Google realizes this, then there could potentially be a whole lot more at stake here than merely taking Qualcomm down a peg (as happy as that would all make us).
  • You can say that again!
  • How will it 'change the industry for the better '...? It might change something for Google and potential buyers of Pixel phones, but the industry as such would hardly feel any change.
  • Would be interesting to see how those A series chips performed with other OS's if apple decided to start selling their chips to others (doubt they would, but could be interesting).
  • Boys and girls, this is the roadmap, accept it and brace for impact. 🤭
  • Google is following the Microsoft path with Windows and Surface. I think Google will work with Qualcomm more and invest in semi-custom chips, hardware driver support and optimization. This will lead to 3 or more generations of Android working on precious generation hardware. The Snapdragon chips have matured to the point all its capability is no explored by manufacture because the life span is to short. With models now being sold for more than one year (Galaxy s10 series and Note 10 series) we have reached the tipping point. There is no escape, ther are to many manufacturers and to short of a phone life span artificially influenced by hardware support.