A Google-Designed Processor for Computers May Be on the Cards As Soon as 2023 – Spiceworks News and Insights

Tháng Sáu 26, 2024 The Art Of Semi 0 Comments

Hardware, like software in the last decade, is re-emerging as a key differentiator.

Taking cue from Apple, Google reportedly already has plans in the pipeline for a 2023 release of Chrome OS laptops, or Chromebooks, running on its own processors. According to three sources familiar with the development, the new processors are based on the ARM architecture.
Looks like in-house development of semiconductors is the path Google has chosen for its line of notebook and tablet computers, according to a report from Nikkei AsiaOpens a new window . The search giant has drawn inspiration — a tad late perhaps — from Apple, and is reportedly developing its own central processing units for laptops including Chromebooks.
This is a big, albeit expected shift considering the company only recently announced in-house system on chips (SoCs) for the Pixel 6 seriesOpens a new window . So far, Google has registered a strong market sentiment for the Pixel 6 series thatuses the Tensor SoC. The move to self-developed chips is clearly Google’s attempt at putting itself in a better market position vis-à-vis its biggest competitor in the smartphone segment: Apple.
Apple was rewarded handsomely for using chips designed in-house in iPhones since 2010. Even as Google’s Android operating system for smartphones dominates the spaceOpens a new window , Apple is the one that has immensely benefited from using custom-made CPUs for its line of consumer electronics.
However, the iPhone maker until recently relied on IntelOpens a new window to power its macOS-based computers and laptops. Apple currently packages the macBooks and iMacs with the M1 chip. In that regard, Google is being cautious not to be left behind.
Google’s plans to bring its own chips for computers have been rumoured since April 2020. And according to the three sources familiar with the developments who spoke with Nikkei Asia, these chips will materialize commercially sometime in 2023.
See Also: IBM Telum: A New Chapter in Real-Time Fraud Detection
Well, almost everyone, right from computer and smartphone makers to cloud vendors and electric car manufacturers want their own chips. Problem is, not everyone has tasted success in this highly competitive, not to mention technically difficult area.
“Very few players have the skills or financial resources to design their own chips, so the typical players considering this path tend to be extremely large players, like the cloud service providers, or have very valuable applications for these specially designed chips,” Peter Hanbury, a partner at consulting firm Bain & Co. told Nikkei Asia.
Additionally, the cost of chip designing is in an upward spiral. It now costs over half a billion USD ($540 million) to design a 5 nm chip, $300 million to design a 7 nm chip, and $175 million for a 10 nm chip, according to global consultants McKinseyOpens a new window . For comparison, Hanbury told Nikkei Asia that a 28 nm chp design costs $50 million.
The following table illustrates which companies are designing chips in-house for respective use cases:
Company
2010
Computers
2011
Huawei
2016
Smartphones
Expected in 2023
Amazon
AI Cloud
2018
2020
Autonomous Driving
NA
Computers
2019
Facebook
2019
In one word, control. Having control on the software, as well as the hardware allows vendors to be more competitive and develop better quality products compared to their competitors that rely on external designs. It also allows device makers to design a smartphone or deliver a cloud service that punches well above its weight for a particular price segment.
Think of it this way: a suit tailored to a person’s measurements will fit better than what’s out there on the rack of a clothing store. It also means companies can control the supply of chips provided they are the ones who are in charge of manufacturing.
Hardware, like software in the last decade, is re-emerging as a key differentiator. To have a say in the development of both for better integration evidently results in better products.
Google is even banking on AI to design its next iteration of TPUs for its AI-processing data centersOpens a new window , and possibly other commercial use cases. In essence, AI will help design better chips, which in turn allows faster AI processing, which can help in designing even better chips.
The company explained this cyclic clockwork, research for which has been going on for well over a year, in the science publication Nature.com in JuneOpens a new window .
It is unlikely that Google, like Apple, will also manufacture self-designed chips. Google’s Tensor chips for Pixel are being manufactured by SamsungOpens a new window .
These new chips will primarily target ChromebooksOpens a new window , those light, snappy laptops running on the Chrome operating system. So it is reasonable to speculate that the first of Google’s in-house processors will not vie for a piece of the high-end computer market. At least not from the outset.
Google’s new chips will be based on the Advanced RISC Machine or the ARM architectureOpens a new window . ARM, a U.K.-based company also licenses its technology IP to nearly all chip manufacturers for use in bulk of the smartphones worldwide. The ARM architecture is notable for efficiency by reducing power consumption on devices, which is why smartphones are the primary target for ARM-based processors.
Since Google’s chip is for lightweight computers, it makes sense that the tech giant leverages ARM over Intel’s x86Opens a new window . ARM is also leveraged in Pixel 6’s Tensor chips. Besides smartphones, ARM is also used in multiple other consumer electronic products such as smart TVs, smart watches, vending machines, set top boxes, cameras, etc. The list is endless.
x86, on the other hand, is used mainly for medium to high performance use cases. This includes any computers (laptops, desktops, notebooks), servers, and supercomputers.
Currently, HP, Dell, Acer, AsusTek, Lenovo and Samsung build chromebooks using different processors (mainly x86). It is unclear if Google will mandate these companies to use its ARM-based chip for future Chromebooks.
See Also: Google Uses AI to Design Next-Generation Chips in Just Six Hours
In the short term, pretty much nothing. The reason being that manufacturing chips is an even bigger challenge than designing them, requiring billions in investment. Obviously money won’t be an issue. Rather, the technical, and surprisingly logistical limitations behind it.
Chip foundries, including those of Taiwan Semiconductor Manufacturing Company (TSMC), Samsung, Intel are moving to extreme ultraviolet lithography (EUV), pouring in billions. EUV is a next-generation lithography technology over which one Dutch company, ASMLOpens a new window , has a monopoly.
Each EUV machine costs $120 millionOpens a new window . Additionally, it costs $1 billion to set it upOpens a new window for high volume production. Moreover, ASML can only make 50 of these 180 ton machines in one whole year, which requires three Boeing 747s just for transportation.
So it’s safe to say that the impact on TSMC, Intel, AMD, Samsung, Qualcomm, UMC, etc., can remain minimal for now. However, considering there’s a chip shortage going on globally, Google may have to wrestle Apple, Microsoft, and others to get the wheels moving on production of its new in-house chips (both Tensor and those for Chromebooks).
Lucky for Google, Intel CEO Pat Gelsinger recently boosted the company’s efforts for Intel Foundry Services.
Google designing its own chips is a good thing for the industry as a whole. It eliminates overreliance on existing third-party vendors whose hardware may not deliver the desired result. More importantly, it keeps Apple’s burgeoning dominance in check. Microsoft is too far behind to be considered in the race as of now.
Let us know if you enjoyed reading this news on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We would love to hear from you!