Why can't ARM just exist as a stand-alone company? I don't get the "ownership" issue. In a significant way, it's just a standards organization, like the RFCs defined the protocols for the Internet, or the standards for relational databases, or Java, or CSS.
A small matter of $40 billion (note that's with a B ) that Softbank paid for the company.
Nvidia is the best of a bunch of bad options for ARM Holdings. ARM Holdings has a lot of licensees - 500 - but their licensing fees are low. Add that to their significant R&D expenses and ARM Holdings doesn't make much profit. This is why ARM Holdings' shareholders chose to cash out by selling to Softbank for $32 billion in the first place. And it is also why Softbank has been trying to unload ARM Holdings almost ever since, secretly at first but finally openly in a desperate bid to dump them. So whether Softbank spins it off as an IPO or finds another buyer, they wouldn't get anywhere near the $40 billion that Nvidia offered wouldn't even get the $32 billion that they paid for it at this point. Everyone's best scenario - an entity who doesn't have a competing or related product buying ARM to make money off licensing fees without getting involved in the company's operations - doesn't exist. Anyone who buys ARM is going to do so in order to use their IP in their products. Of those, Nvidia is the best bet because Nvidia isn't a major CPU company: the Shield TV streaming box and the Nintendo Switch are the only consumer devices that use them. The combined lifetime sales of those devices are less than the number of Android smartphones and tablets sold by Samsung alone in a good quarter. Nvidia is a GPU company, not a CPU one. They want ARM's AI/ML-related IP for their GPUs, not their Cortex IP for their CPUs. Microsoft, Google and Qualcomm all have CPU-based AI/ML platforms and would rather not see Nvidia's GPU-based one get even better by adding ARM's IP and expertise.
Also, the ARM Holdings Cortex cores IP is going to depreciate in value fast. Their Cortex cores were never a match for x86 in performance. Instead their sole advantage was that until recently x86 couldn't be used in low-power devices (the power/heat thing). But at 7nm and lower, x86 can be while providing much better performance. This means - for example - that x86 Android is finally going to become a thing ... though on edge/IoT/appliance devices that are currently Cortex-based ARM, not on phones and tablets. But x86 Windows and ChromeOS devices that offer comparable performance/battery ratios to the iPad Pro will be on the market with 5nm AMD chips (Athlon architecture) and 10nm Intel chips (big.LITTLE architecture) by 1Q2022.
And that is just short term. Long term, the real threat to ARM Holdings is RISC-V. RISC-V is open source. Why is this significant? Even though it is very possible to design much better ARM cores "from scratch" like Apple did, because ARM Holdings owns the IP you can use ARM Holdings core designs as a base. But with RISC-V anyone can take the open-source base and use it to create their own truly custom CPUs. Google, Nvidia, Samsung, Western Digital and Marvell are all RISC-V platinum members. There are already RISC-V based IoT devices in the wild and Micro Magic recently created a prototype CPU that doubles the performance of a Samsung Exynos chip.
So not only is ARM Holdings not making anywhere near enough revenue on licensing to justify a big sale price now, in a few years competition from x86 and RISC-V is going to drive their licensing revenue even lower. Just like Google buying FitBit was better than allowing FitBit to go broke and having who knows buy the FitBit data at bankruptcy auction to do (again) who knows what with it, letting Nvidia buy ARM Holdings today is much better than having Softbank be forced to liquidate it 5 years from now. And the regulators would have far less control over Softbank's liquidating ARM Holdings' IP to cover demonstrable losses than they do approving Softbank's selling ARM Holdings above board.
ARM, as it stands, merely licenses its products to whoever wants to use them. Apple, like all others, already has full access to ARM. There is no need to buy it.
But, many fear what will happen if NVIDIA (or anybody likely to restrict access) takes over those patents and licenses.
So, what would happen if Apple took control of that essentially open library? Others would have to find an alternative to ARM and everybody would lose.
In the end, this is global / globalization problem: After 4 years of attacks, China for one would block any efforts of a U.S company to take control of the ARM licenses. They simply would not tolerate the political weaponization of that technology.
I would imagine that, technically, Apple has an Architectural licence, which is higher than a pertupetual licence.
Apple has a perpetual license (i.e. "lasts forever") for ARM's ISA (Instruction Set Architecture). Has nothing to do with a physical chip, or design of the ARM chip. It's just the instruction set.
Apple doesn't care what happens to ARM the company.
From the article: "SoftBank, the current Japanese owner of ARM, was said to have approached Apple to discuss a takeover of the chip design firm in 2020. Apple, reportedly, wasn't interested in the acquisition due to antitrust concerns -- despite the reliance of Apple Silicon on Arm technology."
I would imagine that, technically, Apple has an Architectural licence, which is higher than a pertupetual licence.
Apple has a perpetual license (i.e. "lasts forever") for ARM's ISA (Instruction Set Architecture). Has nothing to do with a physical chip, or design of the ARM chip. It's just the instruction set.
Apple doesn't care what happens to ARM the company.
Please be more careful, ARM offer something called a Perpetual License, which is not what Apple have. They have an Architecture License. It is speculated that the license terms are in perpetuity, so Apple will have access to the ISA effectively forever, but calling it a perpetual license is inviting confusion.
Apple will certainly want ARM to remain active and healthy, developing the spec.
Comments
Also, the ARM Holdings Cortex cores IP is going to depreciate in value fast. Their Cortex cores were never a match for x86 in performance. Instead their sole advantage was that until recently x86 couldn't be used in low-power devices (the power/heat thing). But at 7nm and lower, x86 can be while providing much better performance. This means - for example - that x86 Android is finally going to become a thing ... though on edge/IoT/appliance devices that are currently Cortex-based ARM, not on phones and tablets. But x86 Windows and ChromeOS devices that offer comparable performance/battery ratios to the iPad Pro will be on the market with 5nm AMD chips (Athlon architecture) and 10nm Intel chips (big.LITTLE architecture) by 1Q2022.
And that is just short term. Long term, the real threat to ARM Holdings is RISC-V. RISC-V is open source. Why is this significant? Even though it is very possible to design much better ARM cores "from scratch" like Apple did, because ARM Holdings owns the IP you can use ARM Holdings core designs as a base. But with RISC-V anyone can take the open-source base and use it to create their own truly custom CPUs. Google, Nvidia, Samsung, Western Digital and Marvell are all RISC-V platinum members. There are already RISC-V based IoT devices in the wild and Micro Magic recently created a prototype CPU that doubles the performance of a Samsung Exynos chip.
So not only is ARM Holdings not making anywhere near enough revenue on licensing to justify a big sale price now, in a few years competition from x86 and RISC-V is going to drive their licensing revenue even lower. Just like Google buying FitBit was better than allowing FitBit to go broke and having who knows buy the FitBit data at bankruptcy auction to do (again) who knows what with it, letting Nvidia buy ARM Holdings today is much better than having Softbank be forced to liquidate it 5 years from now. And the regulators would have far less control over Softbank's liquidating ARM Holdings' IP to cover demonstrable losses than they do approving Softbank's selling ARM Holdings above board.
"SoftBank, the current Japanese owner of ARM, was said to have approached Apple to discuss a takeover of the chip design firm in 2020. Apple, reportedly, wasn't interested in the acquisition due to antitrust concerns -- despite the reliance of Apple Silicon on Arm technology."
Apple will certainly want ARM to remain active and healthy, developing the spec.