


Arm Is Launching Its Own Chip This Year With Meta As a Customer (techcrunch.com) 14
Arm will reportedly start making its own chips this year after signing Meta as a customer, according to the Financial Times (paywalled). TechCrunch reports: The chip is expected to be a CPU for servers in large data centers and can be customized for various customers. Arm will outsource its production. The first in-house Arm chip will be unveiled as early as this summer, the Financial Times also reported.
This is a notable change in strategy for the semiconductor company, which usually licenses its chip blueprints to companies like Apple and Nvidia. Making its own chips will turn some of its existing customers into competitors.
This is a notable change in strategy for the semiconductor company, which usually licenses its chip blueprints to companies like Apple and Nvidia. Making its own chips will turn some of its existing customers into competitors.
Dangerous strategy (Score:5, Insightful)
This is a notable change in strategy for the semiconductor company, which usually licenses its chip blueprints to companies like Apple and Nvidia. Making its own chips will turn some of its existing customers into competitors.
Directly competing with your customers instead of selling them things that they can sell to make money is a fast way to lose your customers.
Re: (Score:3)
Eh, maybe. They're producing Neoverse chips for the server room. Not many players are doing that for sale to the public at large. Amazon won't sell their Graviton hardware on the general market, and NV's Grace chips are mostly tied to their GPU platforms. There's Ampere, but Softbank is buying them out...
Re: (Score:3)
They F'd up by selling permanent/perpetual architectural license to Qualcomm and Apple (pre-2007, before the iPhone .. they thought getting $10 million was huge). https://www.theregister.com/20... [theregister.com] They've been trying every legal loophole to rescind it and failed, therefore to really grow revenue they need something like this.
Re: Dangerous strategy (Score:4, Informative)
You might have information I'm not aware of, granted I'm not following ARM closely, but the article you linked to talks about things that are quite different to what you summarized.
The license deal with Qualcomm is that Qualcomm acquired a small company that had a very generous license with ARM. That small company was founded by former Apple employees but it was not owned by Apple. Qualcomm took advantage of the license and ARM is not pleased about that obviously but failed to litigate to cancel it.
Apple has worked with ARM long before the iPhone as they used them for the Newton. They are actually a founder of ARM as a joint venture between Acorn, Apple and VSLi in 1990. Apple provided the seed investment of $3M. They more recently renewed their partnership with ARM until 2040.
https://en.wikipedia.org/wiki/... [wikipedia.org]
https://www.tomshardware.com/n... [tomshardware.com]
Re: (Score:2)
Apple pays 30 cents per chip and that's because ARM came up with newer instruction sets/versions .. but it is still far below the amount ARM gets from its newer licenses. Apple only brings in 5% of ARM's revenue. https://mobilesyrup.com/2023/1... [mobilesyrup.com]
Re: Dangerous strategy (Score:2)
Those customers, none of whom built their business around building server chips, will sure run to licence chip designs from Intel or AMD instead.
Re: (Score:2)
ARM surely makes money from low volume customers like Apple, NVidia, Qualcom... But a billion cores may not be much.
Consider high volume customers who are very low maintenance and buys tens of billions of cores.
I have a few thousand ARM server cores in production. Cores that are running highly specialized and optimized for ARM code are amazing. And Apple ARM is amazing because Apple specializes and optimizes their toolchains and code for Apple processors. But every test I've do
Re: (Score:2)
Where are the customers going to go? RISC?
ARM is leveraging their market leading position to open up a new market - not surprising
Not a big change (Score:2)
"... The chip is expected to be a CPU for servers in large data centers and can be customized for various customers. Arm will outsource its production."
Customisable means it's still a soft IP block, albeit maybe fully ready to synthesise. The change is that its customers, like Meta, are also the end users.
A match made in hell. (Score:1)
Arm and Facebook - Two of the scummiest companies imaginable.
Thank goodness for x86 and riscv,
Re: (Score:2)
ARM worse than Intel? I don't think so.
But seriously, RISC-V is no competition at the high end because the instruction set just doesn't scale up well. It's great for low-power applications. I can see it displacing MIPS derivatives (e.g. Microchip's PIC32 and various Chinese MIPS-based microcontrollers), maybe squeezing out stuff like AVR, and becoming the default royalty-free soft core for FPGA-based designs, but it just doesn't scale up.
The fact that LibreSoC switched from RISC-V to OpenPOWER should tel
Re: (Score:2)
Today I Learned (Score:2)
Today I learned that ARM didn't actually make any chips.
Re: (Score:2)