Wednesday, April 15th 2020
![Google](https://tpucdn.com/images/news/google-v1723408558483.png)
Google to Design its Own SoCs for Pixel Smartphones and Chromebooks
Google is planning to take the Apple route in designing its own SoCs for its hardware. The company kicked off "Project Whitechapel," an initiative with technical assistance from Samsung (and possibly silicon-fabrication), to design Arm-based SoCs with specialized hardware to power Google's machine-learning tech. The first such chip will be 8-core Arm SoCs powering next-generation Pixel premium smartphones, but in the future, Google could use Arm-based SoCs to power Chromebooks.
Google has, in the past, collaborated with Intel and Qualcomm to put specialized hardware on smartphone SoCs, although the resulting chips would still be supplied by the two. "Project Whitechapel" would see Google play a dominant role in the SoC's design, with Samsung only providing technical inputs. It wouldn't be far-fetched to predict Google using lightweight variants of the SoC on its own IoT hardware, such as Chromecast and Home smart-speakers.
Source:
ArsTechnica
Google has, in the past, collaborated with Intel and Qualcomm to put specialized hardware on smartphone SoCs, although the resulting chips would still be supplied by the two. "Project Whitechapel" would see Google play a dominant role in the SoC's design, with Samsung only providing technical inputs. It wouldn't be far-fetched to predict Google using lightweight variants of the SoC on its own IoT hardware, such as Chromecast and Home smart-speakers.
16 Comments on Google to Design its Own SoCs for Pixel Smartphones and Chromebooks
What the project actually is about (and what ArsTechnica phrased way more clearly), are mobile SoCs that will include some ML optimization for the AI capabilities of mobile devices.
These will probably be tensor cores (like in Google's TPU).
Also, 162g or 193g of gold for $799/899 would be quite the bargain - at current prices that would otherwise cost $8900/10600 . An equivalent volume of gold to the internal volume of one of these phones even more so, given the density of the stuff! So I'd be all for buying a $799 Pixel full of gold!
(For the record, I completely agree that phone prices these days have gone entirely off the rails - my current phone is a Motorola One Zoom, which I'm very happy with even if I do wish it had a couple more GB of RAM.)
Smartphones are expensive because they've replaced many other devices for most people. It has nothing to do with manufacturing cost or anything.
Audio players are gone.
Cameras are quickly going to maybe 10% value from 2 decades ago.
PCs are doing relatively well, but many people choose not to buy them anymore.
All this cash goes into smartphones. In the end, prices are adjusted to the disposable income of consumers. :)
The upside is: if you buy an expensive flagship phone, but you don't own a camera or a PC, you probably end up spending less than you used to be 2 decades ago - "thanks to" consumers who buy more devices. :) And those who buy more stuff have to make some hard choices. Of course I'm talking about global average, not USA. :p
Edit: As a sidenote I hope Google doesn't do this for a few iterations and then abandon the idea like everything else they have done.
I just don't use the phone for media consumption/production. But I do appreciate the ability to backup and restore at the press of a button or being able to pull up a map in a pinch. Or Duolingo.
To be honest, most cheap smartphones are really bad as handhelds. Bad build quality, bad battery life, hard to use in direct sunlight. It's a similar issue I have with cheap gaming laptops.
The only good approach I can think of is buying flagships that are 2+ generations old, i.e. cheap and slow-ish, but still built like the real thing. I did that with iPhone 5S - easily the most fun and pleasant phone I had ever. I replaced it with the LG G6 - supposedly a "flagship on budget" and it's a mixed bag. But I bought it for the camera and that part delivers.
I f you know what to look for, there a re great phones to be had for under $300. That used to be under $200 during the Nexus phone days.
Will someone at least get a move on and start pushing out motherboards for ARM. Been waiting for over 4 years now, best candidate so far is a gigabyte server board. Unacceptable.
This is gonna be entertaining.
Me, I just paid for a DLSR like 10 years ago and only upgraded it last year. Believe me, when it comes to photos that matter, I'm covered. For less money than you overall :p
Also, the Nokia 8.1 can run Pixel's camera software which will output better images than the default one. I just didn't like the interface.
Of course, as you say people's practices, needs and wants change as technology develops, just as technology changes as people's wants and needs develop - so you are indeed right that fewer people today regularly buy PCs, cameras and other electronics that they would have bought in the past decade or decade and a half. On the other hand the 2000s and first half of the 2010s represented an explosion in spending on consumer electronics, with replacement rates for TVs changing from a decade or more to 1-2 years for a time, with cellphones going from a luxury item to a commodity, and with PCs changing from enthusiast toys/work tools to an integral part of everyone's life - and then nearly back again, as most of what most people did with them was suddenly possible with a smartphone. The smartphone rush of the early 2010s, with its push for yearly replacements and drastic performance increases and radical feature additions each generation is also a thing of the past, with average replacement rates even three years ago (when I last worked in retail and followed this stuff closely) starting to stretch out past the two-year mark. This drop in replacement rates from the relatively recent highs is largely the reason why prices are going up - overall sales numbers are stagnant or dropping, manufacturers are running out of "emerging markets" to enter to artificially inflate or maintain sales numbers, and we have reached a level of performance and features where >99% of users would be happy with a lower midrange device. Artificial market segmentation to keep a significant advantage with buying a flagship has also largely become untenable as various budget-focused and "flagship killer" brands have emerged. Flagship pricing is therefore skyrocketing mainly to cover the exorbitant R&D costs for marginal differentiation of products in what is destined to become (and arguably already well on the way to becoming) a high-end niche. Five to eight years ago "everyone" bought flagship phones, but five to eight years ago flagship phones were $400-600 and were radically better than any cheaper options, with every aspect of the phone also being a dramatic improement over the previous model. Today midrange devices occupy those price points, and are catching up to the higher tiers in sales despite flagships receiving all the subsidies, marketing and hype, which alongside ever-slower replacement rates (as devices are largely now "good enough") tells us that the average spending on smartphones isn't actually increasing.
In short: while smartphones represent a convergence in terms of them replacing multiple devices, this is not the reason for their prices going up. Remember, smartphones five years ago could do roughly everything smartphones today can (albeit slower and not quite as well, but current phones don't do much more), so feature additions don't explain the near 100% price increase for flagships since then. Increasing commodification, spiking R&D costs for premium features to stand out, and the appearance of good enough mid-range and lower mid-range options is the main reason.
The same thing happened with PCs a decade or so ago (more for desktops, less for laptops) - when good enough budget PCs started appearing, we suddenly saw a rise of high-priced "premium" devices that try to stand out by being incrementally better in various ways, from build quality to features to design to branding.
Of course you are thus right that flagship smartphone prices are likely to keep rising for a while more, but as they do they will become ever less relevant for the average person. The midrange will on the other hand grow, improve, and become the new battleground for sales volumes and profits, with the flagships being relegated to an aspirational "I'd get one if I could afford it, but I'll buy this cheaper sibling" marketing-first position (which they arguably already have). I would call myself a minor photography enthusiast, yet I wouldn't buy a flagship phone for its camera abilities these days. Phones - even high end ones - aren't suitable for serious photography simply due to their in-use limitations. Sensors are becoming good (and with the advent of large ones like in the S20 series, big enough to not have inherent issues due to sensor size) but they still can't compete with using a proper camera. And even with good manual modes, these cameras still don't have the flexibility to give the kind of creative freedom you get from any semi-serious camera. Non-adjustable apertures, for example, are ... frankly a deal-breaker. Computational photography is quite exciting and can do some really good stuff, but it's not quite there yet, and likely won't be for another five years or more. Digital zoom is largely unusable (and if you're shooting raw, not actually available), and even with current multi-sensor layouts phones lack the flexibility of a DSLR or mirrorless camera. Optical quality is also a serious issue - I have yet to see an ultrawide smartphone camera image that doesn't look like garbage up close, and they universally have tons of chromatic aberrations and distortion. And this won't get better, simply because a lens that small can never be that good. Telephoto is another issue, as I've yet to see a phone exceeding even 100mm - due to obvious physical limitations - but that nonetheless seriously limits its useability.
Of course phones are fantastic for everyday photography - but at that point you're back to the point where even a current mid-range phone is plenty good enough for most people (my One Zoom is surprisingly great if you're not looking to edit or enlarge a lot). There are definitely perceptible differences in optical quality, sharpness, dynamic range and noise between cheaper and more expensive phones, and the quality of the processing algorithms matters a lot, but ultimately, most everyday photography is largely documentary in nature, with aesthetics mattering less than capturing the content, and if any processing is done it's left to heavy-handed filters that obliterate detail and most image quality advantages of better cameras anyhow. For social media use like IG or Snapchat, again, the pros use DSLRs, and for everyone else the difference between a decent mid-ranger and a current flagship are negligible. Hopefully not, those core designs are ... ugh. Back when Samsung announced they were working on a "large area" core design to rival Apple I was quite excited for it, but we've seen how that panned out. I'm hoping Google uses ARM designs as a base, but widens them, enlarges the caches, and goes after the areas where Apple is currently miles ahead. Though that would likely take a few years.