https://www.bloomberg.com/news/articles/2020-07-31/nvidia-said-in-advanced-talks-to-buy-softbank-s-chip-company-arm
6 replies (most recent on top)
@pai+16erYiCF Bean counter Bob is managing the bottom line and wall street as he doesn’t know better or other. He is selling assets quietly to improve the bottoms line and hide terrible 10nm finances which are now showing up on the margin side of things.
No way way intel will be allowed to buy ARM, intel doesn’t license x86 and sues anyone who tries to do x86, good luck letting them try and get ARM. If it was an ARM license, they had it and sold it, another case of squandering a lead.
Impossible for Intel to buy ARM due to antitrust. It will be hard for Nvidia to pull off, and Intel will definitely lobby against it
32 billions. Why Intel not buying companies like ARM, instead of buying unknown and unproven startups?
$32M is the price of an average mansion in San Francisco.
You are not wrong, but the facts you have cherry picked fail to portrait the whole picture.
For example, you paint it as if Nvidia is the only company Apple has had problems with, yet Apple has parted ways with Intel, IBM (Power PCs), and many other companies in the past.
The claim that Nintendo is the only company nvidia successfully collaborates with is just wrong:
- nvidia manufactures GPU chips, collaborates with dozens of OEMs to ship graphics cards
- nvidia collaborates with IBM which ships Power8,9,10 processors all with nvidia technology
- nvidia collaborates with OS vendors like microsoft very successfully
- nvidia collaborated with mellanox successfully and acquired it
- nvidia collaborates with ARM today...
The claim that nvidia is bad at open source because it does not open source its Linux driver is also quite wrong, since NVIDIA contributes many many hours of paid developer time open source, has many open source products, donates money to many open source organizations, contributes with paid manpower to many open source organizations as well...
I mean, this is not nvidia specific.
You can take any big company, e.g., Apple, and paint a horrible case by cherry picking things (no Vulkan support on MacOSX forcing everyone to use Metal, they don't open source their C++ toolchain, etc.), yet Apple does many good things too (open sourced parts of their toolchain like LLVM, open source swift, etc.).
I mean, you even try to paint this as if Nvidia is the only company that Apple has parted ways with, yet Apple has long track record of parting ways with other companies (IBM PowerPC processors, Intel, ...). I'm pretty sure that the moment Apple is able to produce a competitive GFX card, they will part ways with AMD as well.
This is quite concerning honestly. I don't mind ARM being acquired, and I don't mind Nvidia acquiring things. But I'm concerned about this combination.
Nvidia is a pretty hostile company to others in the market. They have a track record of vigorously pushing their market dominance and their own way of doing things. They view making custom designs as beneath them. Their custom console GPU designs - in the original Xbox, in the Playstation 3 - were considered a failure because of terrible cooporation with Nvidia [0]. Apple is probably more demanding than other PC builders and have completely fallen out with them. Nvidia has famously failed to cooporate with the Linux community on the standardized graphics stack supported by Intel and AMD and keeps pushing propietary stuff. There are more examples.
It's hard to not make "hostile" too much of a value judgement. Nvidia has been an extremely successful company because of it too. It's alright if it's not in their corporate culture to work well with others. Clearly it's working, and Nvidia for all their faults is still innovating.
But this culture won't fly well if your core business is developing chip designs for others. It's also a problem if you are the gatekeeper of a CPU instruction set that a metric ton of other infrastructure increasingly depends on. I really, really hope ARM's current business will be allowed to run independently as ARM knows how to do this and Nvidia has time and time again shown not to understand this at all. But I'm pessimistic about that. I'm afraid Nvidia will gut ARM the company, the ARM architectures, and the ARM instruction set in the long run.
[0]: An interesting counterpoint would the Nintendo Switch running on an Nvidia Tegra hardware, but all the evidence points to that this chip is a 100% vanilla Nvidia Tegra X1 that Nvidia was already selling themselves (to the point its bootloader could be unlocked like a standard Tegra, leading to the Switch Fusee-Gelee exploit).