What if instead of buying ARM, they put $40 billion into developing RISC V?
$40 billion is a huge amount of money. They could find the world's top ten chip designers and offer them $1 billion each in cash to come up with a new superior CPU design, and still have $30 billion left over to pay their minions.
If this deal goes through expect lots of ARM licensees like NXP, ST, Broadcom, etc. to start something along these lines.
I just don't want to redesign all my embedded boards because I can no longer easily get ARM cores or nVidia raises the price of random SoCs I use in my designs. I keep hoping that the silicon houses knew this is coming and will start putting out some compelling RISC V chips I can use in future products.
Why not $1 Billion into SPARC? It's still unclear to me why after successfully delivering the 5 Ghz, 8 core, 256 thread, SPARC M8, Oracle wasn't able to at least sell its Microelectronics group for a nominal sum. Toss in the Server group as an added incentive for free.
What if instead of buying ARM, they put $40 billion into developing RISC V?
Actually, NVidia is already using internally developed RISC-V cores in their GPU's. These cores aren't exposed to the outside world, but they're integral to the internal functioning of the GPU. People from NVidia have publicly presented this work at RISC-V events over the last few years. Here's a recent example: https://youtu.be/7Lx3692cbAg [youtu.be].
Well, on that note, why not just buy the same kind of licenses Apple has bought to the ARM IP? It's not as if Nvidia lacks expertise, which, together with the IP, is the only other thing Nvidia needs from ARM to build CPUs.
I don't think this has anything to do with nVidia making CPUs. And that makes the whole thing a bit of a puzzle because the only reason ARM has been successful as a CPU architecture has been that its been controlled by a neutral company as a semi-open standard. It offers almost nothing
You're right.. spending this much money doesn't seem like it's for the technology. They are likely doing it to stifle the competition and get ahead that way. Makes me wonder if they're unable to advance GPU architecture anymore beyond die shrink performance improvements.
To get back the $40 billion is going to take a while unless they dramatically increase prices. ARM makes only about $400 million in revenue per quarter (not including expenses, so profit is less). How are they going to get back the $40 bill
They are likely doing it to stifle the competition and get ahead that way.
Why would they do that, from a business perspective? It's not like Nvidia doesn't care for dollars. Trying to out-compete the competition is at best a risk-full strategy. Much safer and future-proof approach is to just let the royalties roll in.
They might have the notion that they could directly get the sales of the ARM licensees. Such that if Xiaomi wants to make the next Redmi Note, instead of buying an ARM CPU from Mediatek, they would buy it direct from nVidia instead. nVidia can price the CPU high. If Xiaomi still wants to get it from Mediatek it would cost them a lot more becaue nVidia would ask for a high licensing price. Xiaomi won't have much of a choice but to pay unless they want to be stuck on an older CPU design. I do see the side of
Why not $40 billion into RISC V ? (Score:5, Interesting)
What if instead of buying ARM, they put $40 billion into developing RISC V?
$40 billion is a huge amount of money. They could find the world's top ten chip designers and offer them $1 billion each in cash to come up with a new superior CPU design, and still have $30 billion left over to pay their minions.
Re: (Score:1)
I just don't want to redesign all my embedded boards because I can no longer easily get ARM cores or nVidia raises the price of random SoCs I use in my designs. I keep hoping that the silicon houses knew this is coming and will start putting out some compelling RISC V chips I can use in future products.
Re:Why not $40B into RISC V ? (Score:2)
Re:Why not $40 billion into RISC V ? (Score:4, Interesting)
What if instead of buying ARM, they put $40 billion into developing RISC V?
Actually, NVidia is already using internally developed RISC-V cores in their GPU's. These cores aren't exposed to the outside world, but they're integral to the internal functioning of the GPU. People from NVidia have publicly presented this work at RISC-V events over the last few years. Here's a recent example: https://youtu.be/7Lx3692cbAg [youtu.be].
Re: (Score:2)
Well, on that note, why not just buy the same kind of licenses Apple has bought to the ARM IP? It's not as if Nvidia lacks expertise, which, together with the IP, is the only other thing Nvidia needs from ARM to build CPUs.
I don't think this has anything to do with nVidia making CPUs. And that makes the whole thing a bit of a puzzle because the only reason ARM has been successful as a CPU architecture has been that its been controlled by a neutral company as a semi-open standard. It offers almost nothing
Re: (Score:2)
You're right .. spending this much money doesn't seem like it's for the technology. They are likely doing it to stifle the competition and get ahead that way. Makes me wonder if they're unable to advance GPU architecture anymore beyond die shrink performance improvements.
To get back the $40 billion is going to take a while unless they dramatically increase prices. ARM makes only about $400 million in revenue per quarter (not including expenses, so profit is less). How are they going to get back the $40 bill
Re: (Score:2)
They are likely doing it to stifle the competition and get ahead that way.
Why would they do that, from a business perspective? It's not like Nvidia doesn't care for dollars. Trying to out-compete the competition is at best a risk-full strategy. Much safer and future-proof approach is to just let the royalties roll in.
Re: (Score:2)
They might have the notion that they could directly get the sales of the ARM licensees. Such that if Xiaomi wants to make the next Redmi Note, instead of buying an ARM CPU from Mediatek, they would buy it direct from nVidia instead. nVidia can price the CPU high. If Xiaomi still wants to get it from Mediatek it would cost them a lot more becaue nVidia would ask for a high licensing price. Xiaomi won't have much of a choice but to pay unless they want to be stuck on an older CPU design. I do see the side of
Re: (Score:2)