Softbank selling ARM to Nvidia, finally after year of speculation...

This, as was Mellanox, and AI play, not us PC "master race" people. Big picture folks. Nvidia is doing lots of right things in the enterprise space right now.

It can be, and will be more than one thing.

It can be that eventual dominance of Mobile/Datacenter/PC business. That justifies the 40B price tag.
 
for sure, but people need to stop equating Nvidia with gaming for consumers only.

Yes, but it can be very much for the PC Master race as well, and this is less obvious to many.

Most people are focusing on mobile. Because that is currently the main thrust of ARM. But ARM already saturates mobile, and this does not justify 40 Billion.

Data Center is the next most obvious target. ARM can make good inroads into servers, and combined with parts NVidia already owns, it can build the whole system. But here they could just get cheap ARM licenses to do that.

But what is less obvious, and benefits more from taking over ARM holdings, is really driving ARM into the PC business. NVidia can have ARM specifically target PC oriented designs, and work with Microsoft to get Windows on ARM tuned up better.

If NVidia can make Windows on ARM successful, they win a big chunch PC market they can't touch: CPUs. They can also compete for the APU market that is eating low end GPU sales.

Windows on ARM success is the holy grail for a NVidia that owns ARM.
 
Looking at most of the stuff people use on Windows outside of games, well, those codebases are already running on ARM (and Linux / Unix kernels) elsewhere. Yeah, there's lots of legacy stuff that'll get left behind, but just like Apple isn't ditching x86 in one fell swoop, Nvidia with ARM won't be taking over the world in one day.
Most of that legacy stuff at this stage could be run on a clock radio. So even if emulation is horribly inefficient it will still work.
 
Yes, but it can be very much for the PC Master race as well, and this is less obvious to many.

Most people are focusing on mobile. Because that is currently the main thrust of ARM. But ARM already saturates mobile, and this does not justify 40 Billion.

Data Center is the next most obvious target. ARM can make good inroads into servers, and combined with parts NVidia already owns, it can build the whole system. But here they could just get cheap ARM licenses to do that.

But what is less obvious, and benefits more from taking over ARM holdings, is really driving ARM into the PC business. NVidia can have ARM specifically target PC oriented designs, and work with Microsoft to get Windows on ARM tuned up better.

If NVidia can make Windows on ARM successful, they win a big chunch PC market they can't touch: CPUs. They can also compete for the APU market that is eating low end GPU sales.

Windows on ARM success is the holy grail for a NVidia that owns ARM.
With Apple driving their hardware to ARM software developers will follow, that means a lot of the big software players are already working on that architecture. Microsoft’s .Net libraries and dev environments should be able to compile the same code for ARM just as well as it does for x86 so all that’s needed really is for an ARM based CPU that is powerful enough for a mainstream desktop and to convince Microsoft to release a version of Windows for it that isn’t crippled.
 
With Apple driving their hardware to ARM software developers will follow, that means a lot of the big software players are already working on that architecture. Microsoft’s .Net libraries and dev environments should be able to compile the same code for ARM just as well as it does for x86 so all that’s needed really is for an ARM based CPU that is powerful enough for a mainstream desktop and to convince Microsoft to release a version of Windows for it that isn’t crippled.
Do I smell a whole new flavour of Hackintosh's in the future....?
 
Yes, but it can be very much for the PC Master race as well, and this is less obvious to many.

Most people are focusing on mobile. Because that is currently the main thrust of ARM. But ARM already saturates mobile, and this does not justify 40 Billion.

Data Center is the next most obvious target. ARM can make good inroads into servers, and combined with parts NVidia already owns, it can build the whole system. But here they could just get cheap ARM licenses to do that.

But what is less obvious, and benefits more from taking over ARM holdings, is really driving ARM into the PC business. NVidia can have ARM specifically target PC oriented designs, and work with Microsoft to get Windows on ARM tuned up better.

If NVidia can make Windows on ARM successful, they win a big chunch PC market they can't touch: CPUs. They can also compete for the APU market that is eating low end GPU sales.

Windows on ARM success is the holy grail for a NVidia that owns ARM.
I don't think NVidia is buying ARM to focus on Windows and the PC business. Quite frankly, that would be a step backwards rather than forwards - there's a reason why Apple is dumping x86 for Macbooks, the same onus should fall on Microsoft to get with the times. This is all about the next generation of compute whether it's up at the datacenter or a localized network of many devices such as cars communicating with each other to predict traffic patterns, self driving, etc.
 
I don't think NVidia is buying ARM to focus on Windows and the PC business. Quite frankly, that would be a step backwards rather than forwards - there's a reason why Apple is dumping x86 for Macbooks, the same onus should fall on Microsoft to get with the times. This is all about the next generation of compute whether it's up at the datacenter or a localized network of many devices such as cars communicating with each other to predict traffic patterns, self driving, etc.

It's not about any one thing, it's about all of them. If you acknowledge Microsoft is eventually going ARM, then NVidia owning ARM puts them in the CPU driving seat for:

Mobile
Data Center
PC

And all the synergies in them. That is what makes this worth 40 Billion.
 
It's not about any one thing, it's about all of them. If you acknowledge Microsoft is eventually going ARM, then NVidia owning ARM puts them in the CPU driving seat for:

Mobile
Data Center
PC

And all the synergies in them. That is what makes this worth 40 Billion.
I don't acknowledge Microsoft will make a full arm compatible os meant to fill the market already occupied by x86 as frankly arm is worse for those applications. The fields arm can pickup traction is largely in the datacenter where they are able to impliment software and hardware soulutions without relying on a bunch of software devs to change there ways.

Why hasnt microsoft made a os to run on ibm power cpus? They are pretty powerful chips
 
And they are close to the chrome book range of things. I can run 64 bit windows on some phones but they wouldnt be a great solution for any computational tasks. Actual performance is still desired for most desktops and laptops and arm has not made great strides in providing such

Which is why NVidia owning ARM holding, and pushing for better desktop class ARM chips, makes the Windows on ARM more viable, and in the long run a greater reward for NViida.
 
Which is why NVidia owning ARM holding, and pushing for better desktop class ARM chips, makes the Windows on ARM more viable, and in the long run a greater reward for NViida.

The problem with arms viability for desktops is the architecture of arm cores. Coupled with the massive infrastructure currently developed for x86.

Yes its possible, no they wont beat the chips currently out there. Nvidia is not taking aim at the desktop CPU market. They are taking aim at the commercial computing market and the mobile computing market. If you see them at some point expand to laptops for a significant period of time to actually allow development to shift to supporting a arm cpu then they can evaluate if its worthwhile to explore the desktop market. That is not a decision they are making now and that is not why they droped this chunk of change on arm
 
Apple doesn't control a significant amount of the desktop market to change anything on the PC side. ARMs will not be a thing any time soon on PC. Apple is in their own little bubble and me and most people could care any less what they do their.
 
NVIDIA to Acquire Arm for $40 Billion, Creating World’s Premier Computing Company for the Age of AI
https://nvidianews.nvidia.com/news/...s-premier-computing-company-for-the-age-of-ai

“AI is the most powerful technology force of our time and has launched a new wave of computing,” said Jensen Huang, founder and CEO of NVIDIA. “In the years ahead, trillions of computers running AI will create a new internet-of-things that is thousands of times larger than today’s internet-of-people. Our combination will create a company fabulously positioned for the age of AI.

“Simon Segars and his team at Arm have built an extraordinary company that is contributing to nearly every technology market in the world. Uniting NVIDIA’s AI computing capabilities with the vast ecosystem of Arm’s CPU, we can advance computing from the cloud, smartphones, PCs, self-driving cars and robotics, to edge IoT, and expand AI computing to every corner of the globe.

“This combination has tremendous benefits for both companies, our customers, and the industry. For Arm’s ecosystem, the combination will turbocharge Arm’s R&D capacity and expand its IP portfolio with NVIDIA’s world-leading GPU and AI technology.

“Arm will remain headquartered in Cambridge. We will expand on this great site and build a world-class AI research facility, supporting developments in healthcare, life sciences, robotics, self-driving cars and other fields. And, to attract researchers and scientists from the U.K. and around the world to conduct groundbreaking work, NVIDIA will build a state-of-the-art AI supercomputer, powered by Arm CPUs. Arm Cambridge will be a world-class technology center.”


Excerpt from the article.
 
None of what you guys or the article stated couldn't be done with just an arm license. None of it explains why Nvidia would be willing to overpay for something that was overpaid for the first time. Great, they want to build top/bottom ai systems for big business... They can do that with their arm license. They want to make a push into consoles....they've done it with the shield with just a license. If they want to work with microsoft to make desktop arm a thing, cool, they can already build arm + you systems with a license. The only way any of these cases makes sense to go from licensee to lincenser is if they can help control their competition. If they do a lot of arm + windows work right now, everyone with a compatible arm chip benefits, immediately creating their own competition. The only way to make arm work for them is to limit competitors. You can't do that with a license. The other reason is to have more control over the architecture itself, but if they need substantial updates/changes to the architecture and instruction set (either for them to license or for them to take over and stop giving updates to licensees).
Nobody knows the true motives besides a very select few, but it's not just to build an arm anything, they can do that now. If this gets approved we probably won't see an immediate change as I'm sure there would be a lot of stipulations built in, but over time you and other businesses will slowly notice the changes and eventually the true motive will be known.
 
NVIDIA to Acquire Arm for $40 Billion, Creating World’s Premier Computing Company for the Age of AI
https://nvidianews.nvidia.com/news/...s-premier-computing-company-for-the-age-of-ai

“AI is the most powerful technology force of our time and has launched a new wave of computing,” said Jensen Huang, founder and CEO of NVIDIA. “In the years ahead, trillions of computers running AI will create a new internet-of-things that is thousands of times larger than today’s internet-of-people. Our combination will create a company fabulously positioned for the age of AI.

“Simon Segars and his team at Arm have built an extraordinary company that is contributing to nearly every technology market in the world. Uniting NVIDIA’s AI computing capabilities with the vast ecosystem of Arm’s CPU, we can advance computing from the cloud, smartphones, PCs, self-driving cars and robotics, to edge IoT, and expand AI computing to every corner of the globe.

“This combination has tremendous benefits for both companies, our customers, and the industry. For Arm’s ecosystem, the combination will turbocharge Arm’s R&D capacity and expand its IP portfolio with NVIDIA’s world-leading GPU and AI technology.

“Arm will remain headquartered in Cambridge. We will expand on this great site and build a world-class AI research facility, supporting developments in healthcare, life sciences, robotics, self-driving cars and other fields. And, to attract researchers and scientists from the U.K. and around the world to conduct groundbreaking work, NVIDIA will build a state-of-the-art AI supercomputer, powered by Arm CPUs. Arm Cambridge will be a world-class technology center.”


Excerpt from the article.
SoftBank seems to be selling to try and soften their massive losses over investing in WeWork
 
The problem with arms viability for desktops is the architecture of arm cores. Coupled with the massive infrastructure currently developed for x86.

Nothing wrong with ARM core architecture. Yes legacy x86 SW is an issue, but it's far from an insurmountable one.

Now that NVidia owns ARM (subject to regulatory approval) they will be motivated to push it everywhere and that includes on the PC.

ARM making seriously inroads into the Windows eco-system is only a quesiton of "when", not "if".

NVidia is an aggressively competitive company. "When" is going to be a lot sooner with them owning ARM.
 
Am I the only one that thinks AMD is going to be in a very difficult position now? Before, they only had to contend with the behemoth intel in the cpu market, now two companies, which are much larger than AMD.
 
Last edited:
You can run a desktop on a potato. Not even figuratively speaking, but you really should be able to perform all of the basic tasks that most consumers do with a desktop operating system using the voltage supplied by a potato, powering a modern, efficient architecture.

Well, that's probably a reach, but what's not a reach is that modern 'desktop stuff' just doesn't need a lot. The Raspberry Pi line is plenty evidence of that; the Pi 4 can do pretty much any desktop task, including some gaming, quite well while being extremely cheap and using an architecture (both the various ISAs and the manufacturing process) that is nearly obsolete itself.

Apple has shown what a modern ARM, really modern 'not x86', system can do. And by exercising strict control over their product stack, they've been able to ensure that software and hardware are optimized together alongside a focus on efficiency and security. Imagine how fast desktop computers could be if they received the same focus!
 
Am I the only one that things AMD is going to be in a very difficult position now? Before, they only had to contend with the behemoth intel in the cpu market, now two companies, which are much larger than AMD.
With Ryzen they're in a pretty good position, not just because it's a good technology, but also because they have a manufacturing partner that seems pretty reliable at the moment. What's critical is that they keep innovating. Intel's faltering is due to a number of pressures, many self imposed, but none that they can't recover from.

And right now demand for CPUs that are great at chewing through difficult to parallelize workloads really isn't being met. Aside from production volume, that's the one thing that Intel and AMD bring with x86, and the one thing that ARM or Power or RISC-V will need to accomplish before either Intel or AMD are 'in trouble'. Apple has shown that ARM is 'fast enough' for user-facing tasks so long as the cores are supplemented with generous coprocessors that can handle the heavy lifting for the most common workloads like graphics, but they haven't shown the single-threaded performance of x86 yet.
 
Nothing wrong with ARM core architecture. Yes legacy x86 SW is an issue, but it's far from an insurmountable one.

Now that NVidia owns ARM (subject to regulatory approval) they will be motivated to push it everywhere and that includes on the PC.

ARM making seriously inroads into the Windows eco-system is only a quesiton of "when", not "if".

NVidia is an aggressively competitive company. "When" is going to be a lot sooner with them owning ARM.

Ibms risc implication is a better desktop and server core. And modern x86 cores are undoughtably more powerful even with all the junk they carry.

They will be trying to push it everywhere it can generate money that is not the desktop scene. Thats the rivers of money flowing through the commercial space. If at some point in the future the desktop makrt becomes a easy river to fish in they will do that as well but it currently is not

I also expect to see a competitive supercomputing soluion now that nvidia has aquired all of the infinity stones lol
 
SoftBank seems to be selling to try and soften their massive losses over investing in WeWork
They are also heavily invested in Lyft, and Uber, and a whole lot of other Tech companies that are looking to crash out hard.
 
Ibms risc implication is a better desktop and server core. And modern x86 cores are undoughtably more powerful even with all the junk they carry.

I think there is plenty of doubt about that. Apple ARM Macs will soon be showing what ARM can do with a bit bigger power budget. IBMs RISC (Power) is irrelevant for PC, for the same reason Apple abandoned it. Most PCs are laptops, and perf/watt rules there.

If at some point in the future the desktop makrt becomes a easy river to fish in they will do that as well but it currently is not

So you also agree that it is only a matter of When, not if. So what are you arguing about. I never said it was overnight. It's obviously a long term play.
 
They will be trying to push it everywhere it can generate money that is not the desktop scene. Thats the rivers of money flowing through the commercial space. If at some point in the future the desktop makrt becomes a easy river to fish in they will do that as well but it currently is not
Desktops less so, but laptops?

You're lucky to get a day of work out of a laptop. Far less if it's built for workstation use. If Nvidia can make ARM fast enough for Windows with native apps, like Apple is looking to do, that's a pretty big win in a huge market itself.
 
Am I the only one that thinks AMD is going to be in a very difficult position now? Before, they only had to contend with the behemoth intel in the cpu market, now two companies, which are much larger than AMD.


No not the only one, that is what I said from the start, though I think Intel is in trouble too. Though that's the long view, not in the next couple of years.

x86 clings to it's stronghold largely because of Windows. Everything elsel left, is running fairly portable code that can be easily ported. So x86 bulwarks are maintained by Windows.

When ARM makes significant inroads into Windows, that will hurt both Intel and AMD as the stronghold crumbles.
 
Last edited:
Am I the only one that thinks AMD is going to be in a very difficult position now? Before, they only had to contend with the behemoth intel in the cpu market, now two companies, which are much larger than AMD.

What's funny and ironic about this statement is that way back in 2007 (I think) when AMD acquired ATI, Kyle wrote an editorial saying that it was going to be Nvidia in a difficult position. IIRC, he suggested that perhaps Intel might buy Nvidia?

Talk about a 180 in change of fortunes.
 
I think there is plenty of doubt about that. Apple ARM Macs will soon be showing what ARM can do with a bit bigger power budget. IBMs RISC (Power) is irrelevant for PC, for the same reason Apple abandoned it. Most PCs are laptops, and perf/watt rules there.



So you also agree that it is only a matter of When, not if. So what are you arguing about. I never said it was overnight. It's obviously a long term play.
Ibms cpus are irrelevant for the same reason arm is. No software support. (And expencive as heck)

And we will have to see what the desktop market shapes up to be in the future if it even still exists in the same capacity. We are not talking about the near term. Abd no that wasnt nvidia play. Nvidias play involves big money in the commercial segment (and possibly mobile) that would be some icing on the cake way down the road (small money compared to what they are thinking

Desktops less so, but laptops?

You're lucky to get a day of work out of a laptop. Far less if it's built for workstation use. If Nvidia can make ARM fast enough for Windows with native apps, like Apple is looking to do, that's a pretty big win in a huge market itself.

I can see arm laptops coming around but in the current software/gaming scene they would be of zero use to me. They would be fighting for market share with chrome books and tablets.
 
Well, I think Tim Cook will play ball, and hopefully some serious Apple/Mac workstations powered by ARM chips and Nvidia GPU's will come out of this.

Gonna be fun to watch, my last Apple workstation is about 10 years old now haha..
 
I can see arm laptops coming around but in the current software/gaming scene they would be of zero use to me.
Games I'll give you. That's a really big hurdle to clear. But it's also not a market that can't be cleared: the biggest games are not the most graphically intensive games, even on the desktop, and with Nvidia graphics, I don't see it as impossible. Just not... soon.
They would be fighting for market share with chrome books and tablets.
What's the practical difference between an ARM Chromebook and a Windows x86 laptop?

To me, it's performance and application availability. The performance is what Nvidia is now in a position to address, and application availability from a strictly consumer perspective is a non-issue. Porting to ARM isn't going to be fast, but it's not going to be that difficult either.
 
With Ryzen they're in a pretty good position, not just because it's a good technology, but also because they have a manufacturing partner that seems pretty reliable at the moment. What's critical is that they keep innovating. Intel's faltering is due to a number of pressures, many self imposed, but none that they can't recover from.

And right now demand for CPUs that are great at chewing through difficult to parallelize workloads really isn't being met. Aside from production volume, that's the one thing that Intel and AMD bring with x86, and the one thing that ARM or Power or RISC-V will need to accomplish before either Intel or AMD are 'in trouble'. Apple has shown that ARM is 'fast enough' for user-facing tasks so long as the cores are supplemented with generous coprocessors that can handle the heavy lifting for the most common workloads like graphics, but they haven't shown the single-threaded performance of x86 yet.
It depends on the task, and Japan's new supercomputer is equiped with ARM CPUs that have far more FLOPS (FPU/Vector) than any x86-64 CPU has, and Power ISA CPUs far outreach x86-64 in threading and raw computation.
x86-64 is hardly the gold standard for embarrassingly parallel workloads - this is why GPUs are being used for such tasks.

x86-64 is far from having the most powerful single-threaded performance as well - that crown goes to POWER CPUs by a wide margin.
It also depends on the task at hand - the only CPU ISA that is generally good at everything, again, is the Power ISA (assuming GPUs aren't used) - for a price.
 
IBMs RISC (Power) is irrelevant for PC, for the same reason Apple abandoned it.
It was too costly for IBM to move forward with such small sales from Apple, and at that time, x86-64 was on the rise with Core and Core 2 CPUs from Intel.
It had nothing to do with PowerPC being "irrelevant for PC" - again, x86/x86-64 is hardly the gold standard for everything, and ARM is now proving that, just as x86 proved that m68k was hardly the gold standard for computers back in the late 1980s and into the 1990s.
 
Japan's new supercomputer is equiped with ARM CPUs that have far more FLOPS (FPU/Vector) than any x86-64 CPU has
This is more a product of different design goals though, from what I've seen. You could make an x86 CPU do more, you'd just have to sacrifice somewhere else. We can't really say the inverse with ARM yet.
x86-64 is far from having the most powerful single-threaded performance as well - that crown goes to POWER CPUs by a wide margin.
It also depends on the task at hand - the only CPU ISA that is generally good at everything, again, is the Power ISA (assuming GPUs aren't used) - for a price.
I'll admit to knowing very little of the current state of Power, since availability is near zero.

But I should be more clear on 'single-threaded performance'. I'm not really talking about being able to run through one thread quickly, but also to be able to deal with the branching that makes single-threaded workloads truly single-threaded. That's where I see x86 as having a real advantage in current implementations.
 
If I was Nvidia, I'd be incredibly giddy over the idea of being able to make a my own console that uses my graphics cards paired with my own highly specialized CPU, to essentially build a console that's fully in-house and extremely competitive, with ultimate power. No more having to put Nvidia cards inside an AMD or Intel motherboard. Pure Nvidia.

As a consumer though, this would be just about the worst thing.

You do know outside of the CPU nvidia did this before right? Motherboards with nvidia chipsets hosting AMD or Intel CPU's depending on the build. With 'high end' audio solutions and custom on board graphics solutions as well... right? Because they did and they lost their shirts with that part of the business and backed out of it.
 
Because they did and they lost their shirts with that part of the business and backed out of it.
Intel did stop allowing third-party chipsets at that time too.

Nvidia also didn't really have a value-add at the time. Their last products were solid, certainly more so than VIA, but they had no place in volume sales of say HP and Dell corporate desktops and laptops.

Also, Nvidia still builds 'chipsets', in terms of the core logic to hook everything up. Their Jetson boards run circles around Raspberry Pi SBCs and have all of the connectivity potential a consumer desktop or laptop would need.
 
Why is it abuse if they pay $40bn for it? Is it a crime if they try to make returns on it? Since when did technology become a community property? I see a lot of pearl clutching going on about evil nvidia taking over Arm but no solid reasons offered of why it’s such a bad thing. Apple controls it’s chain top down and so does Samsung yet I don’t see people crucifying them. In fact I've seen stupid suggestions that Apple or Amazon should take Arm over, as if that would be any better.

It is not a crime to make returns. However, it *is* a crime to use a significant market position in one market to link sales to push products in another market. In other words, nvidia is free to make profits in licensing arm technology and selling their own chips, but they are likely restricted from offering bundled arm-gpu/ai technology licensing. They are also not allowed to increase the prices of arm licensing so much that their own chips become the only viable arm solution.
 
Last edited:
Why is it abuse if they pay $40bn for it? Is it a crime if they try to make returns on it? Since when did technology become a community property? I see a lot of pearl clutching going on about evil nvidia taking over Arm but no solid reasons offered of why it’s such a bad thing. Apple controls it’s chain top down and so does Samsung yet I don’t see people crucifying them. In fact I've seen stupid suggestions that Apple or Amazon should take Arm over, as if that would be any better.

It is a crime if they use their leverage after buying ARM to put other ARM licensees who are their competitors at a disadvantage.

That is the very definition of anti-competitive behavior.

SoftBank didn't face this scrutiny because they didn't directly compete with ARM licensees. (At least not that I am aware of)

If Nvidia continues to run ARM like it has been run in the past there is no problem. If they use the fact that they have bought ARM to give themselves and advantage over other ARM licensees in other markets in which they compete, it definitely runs afoul of the Sherman act or the Clayton act (or possibly both)

I don't know if this will be held prior to going through, but if they start using their ownership of ARM Holdings to hurt other entities they compete with as a user of ARM chips I bet it will go to the courts in a hurry.
 
Back
Top