NVIDIA Interested in Acquiring Arm from SoftBank

[medi01]

AMD has CPU and GPU products in portfolio, can combine the two.
Intel has CPU and GPU products (the latter was lacking, but they have realized it's a problem)
NV, with just aGPU products is in a risky position (e.g. what will happen to NV's MX chips when Intel rolls out its own?)

#metoo moment for them on one hand.

On the other, of all the imaginable things that could have happened to Arm, getting into The Leather Man's hand is the worst: otrageous business practices, track record of killing tech like PhysX and pissing off: Microsoft, Apple, TSMC. The guy seems to play only zero-sum games.


And, last, but not least, does company's stock price going up automatically mean it has more cash?

 

[Frick]

So much talk, so little knowledge in this thread.
It seems like most of you have no idea what the implications would be if ARM ended up with a company that would stop licensing the architecture.
Everyone thinks about phones, but that's really just a tiny, tiny part of ARM based hardware.

Think about all the MCUs out there based on the Cortex-M series and suddenly you realise that there will be a lot less chips for everything from bluetooth to wireless headphones to thermostats to rice cookers and what not.
ARM chips are in just about everything we use that have some kind of processing power today, even more so in lower power devices.
Sure, there's some competition here by old 8051 8-bit MCUs and some other 8, 16 and 32-bit MCUs, but I would say 75%+ is based on ARM today.

The Cortex-R series is used in SSD controllers, car computers and anything that requires an RTOS.

The Cortex-A series of SoC processor cores are also the most widely used processor core in routers today, even though MIPS technically has a better architecture for routers, but they simply didn't keep up with ARM. They're obviously in phones, but also in pretty much all modern "smart" TVs, set-top-boxes, media players, smart speakers and what not.

It would be an utter disaster if ARM ended up being sold to a company that didn't continue to license the technology, as it we'd end up in a situation where there would be no new products for at least a couple of years until there's something available to replace all these different parts that are based on ARM architectures. RISC-V seems to have the best place in the market right now to step up, but so far, it's not a simple task as there are no simple drop-in replacements, as the architectures are quite different, despite both being RISC based. MIPS seems to have committed harakiri, so I'm not sure they'll ever come back from that. ARC/Synopsys might be able to replace some MCU designs as well. PowerPC is mostly dead in this space as well. Intel has proven they can't do low power. AMD, who knows?

Maybe the world has become to reliant on ARM and its processor designs, but they haven't really had too much competition either. So yeah, this is very disconcerting to say the least.

Which is why it's more likely they will not be bought by any large company with obvious interests in the industry. Spin off, IPO. Makes more sense from a regulatory view.

 

[kings]

If Nvidia ends up buying ARM, it will certainly have many clauses from regulators, one of which is to continue to license the technology.

 

[LemmingOverlord]

A way to re-enter the phone market, maybe with an nVidia phone.

This is not Nvidia's business model. The only product they have which is not a component, per se, is the Nvidia shield, and that is because it is more of a technology demonstrator of the platform's capacities.

If anyone can continue ARM's work, it is Nvidia or Intel. I'd rather have Nvidia doing this, though. They are more familiar with ARM design and IP than Intel, and it would open up a world of options for the datacentre space where Nvidia has been pushing its ML/AI/HPC solutions.

EDIT: I'm seeing a lot of people commenting on Nvidia buying ARM and pulling licenses. That's just daft. The biggest reason Nvidia has to buy ARM is to continue ARM's licensing model. What would ARM be without its licensing revenue? Does Nvidia have a competing alternative to ARM? No. That would just be shooting yourself in the foot.

Which is why it's more likely they will not be bought by any large company with obvious interests in the industry. Spin off, IPO. Makes more sense from a regulatory view.

Spin off & IPO would end up handing the company to a Chinese investment fund (because, let's face it, that's what would happen) and it's the surest way to kill off ARM.

but realistically when it comes to SoC there are not many choice for that either. Qualcomm is pretty much forcing all smartphone maker to use their snapdragon package. other SoC maker most often have only two choice when it comes to GPU: ARM own Mali or Imagination Technologies PowerVR. AMD also now present but since the effort is a joint venture between AMD and samsung there is possibilities samsung want to keep AMD GPU to their line of SoC as an advantage over others. for majority of SoC maker they most often go with ARM Mali. imagination real customer for the last decade really is only apple. hence when apple decided to make their own custom GPU Imagination basically screwed back then....until some chinese company bought them.

The main problem is that you can achieve better economies of scale and performance, as well as lower costs to develop (i.e. the buy in to develop a mobile graphics architecture would cost billions and take years). So they buy the whole package from the same vendor (e.g. Qualcomm). It isn't pretty, but it's what there is. Samsung and AMD will definitely keep their product locked to Exynos SoCs. As for the other vendors who develop their own SoCs, like you said, they rely on PowerVR or Mali (that latter of which, I hear, is quite expensive).

Does anyone know whether there is - somewhere - in the Intel Xe product stack a mobile part?

The only "independent" GPU IP that is around that isn't half-bad, and could possibly be the object of investment is Broadcom's VideoCore (VI, right now) which powers the Raspberry PI 4... If Broadcom were to part with it.

 

[Anymal]

Its obvious, SoftBank would get most on the brink of Apple transition to ARM. Its all about money and now its the best time to do that. I am pretty sure Apple is the only one that has enough money to buy it and SoftBank know it, there no other buyer.

 

[TheLostSwede]

If Nvidia ends up buying ARM, it will certainly have many clauses from regulators, one of which is to continue to license the technology.

Current technology yes, but can they be forced to license future technology?

EDIT: I'm seeing a lot of people commenting on Nvidia buying ARM and pulling licenses. That's just daft. The biggest reason Nvidia has to buy ARM is to continue ARM's licensing model. What would ARM be without its licensing revenue? Does Nvidia have a competing alternative to ARM? No. That would just be shooting yourself in the foot.

Well, there are many ways it could go, but today, Nvidia is not a technology licensing company for sure. With their current business model, it doesn't make sense, although I guess they could change that.

I highly doubt it's the biggest reason for Nvidia to buy them though, the biggest reason would be to own their own ecosystem and become competitive with Intel and AMD.
Looking how successful, or rather not successful Nvidia has been with a lot of their acquisitions in the past, to me, this is scary.

Some examples of companies that Nvidia bought and killed: Exluna, MediaQ, iREady, ULi (ok, so they sold chipsets for a while), Hybrid Graphics, PortalPlayer (parts might be used in Tegra) and Icera (3G/4G modem company, killed in less than four years). That leaves Ageia (although PhysX doesn't seem to be a thing any more), PGI (compilers, debuggers etc.) and hopefully Mellanox as successful buys. Not exactly a splendid track record. Obviously some of the technology might've been used in other products, but they seem to be no better than the competition in making something out of their acquisitions.

So say hey buy ARM, I can see them killing off a lot of the business units within ARM, GPUs for one, as they'll want everyone to use Nvidia and not ATI based GPU technology. Maybe they'll sell off or kill the Cortex-M series. They'll most likely want to hang on to the "new" server architecture from ARM, as they seem to be wanting to build their own servers. This might also mean that they'll cut off the likes of AppliedMicro, Cavium and Fujitsu. Not sure how interested they would be to continue to support the phone chip makers, as that requires a lot of staff, although, if they can get their graphics technology into that part of the market, they might do it.

Too many unknowns, but as I said, this is disconcerting, as ARM technology is in vastly more products than most people think about.

 

[Caring1]

I can't wait for the Nvidia FE Fridge with HDR1000 screen complete with game controller and wireless connectivity to the latest HMD for Virtually browsing the contents.
PC Partner editions can add RGB and special edition cooling such as Windforce chiller for the vegetable crisper.

 

[Xaled]

Scary news, not only that they may kill it, make it much more expensive, or let it have slow to no progress, but they may finally force their GPP program or something similar to it

 

[efikkan]

Funny, it seems to me like Nvidia's ARM effors have slowed down in recent years.

Remember that AMD wasted billions and a lot of their manpower on ARM efforts for years and getting "nowhere", and is largely responsible for AMD to failing to compete in the CPU market for years. It's better to stick to the stuff you know, than trying to conquer new markets with several established players.

I question Nvidia's motivation with ARM. Nvidia's primary market is desktop graphics, ARM CPUs would stand no chance there. and they would risk losing their entire market share.

Also, Arms will be everywhere in the future, not just in phone. This would give Nvidia the ability to compette against AMD and Intel who both do CPU + GPU

But i suspect they won't be alone there, and i wouldn't be surprised if they go public instead for even more money.

ARM is used pretty much everywhere in embedded devices, but have no chance against the performance of x86.

ARM is the future and everyone is tired of legacy x86. Make it happen Jensen!

ARM would have to become CISC to compete with x86 performance.

And ARM is "old" as well, 35 years already, we have been waiting for this new "future" since the 80s.
The whole "legacy" argument about x86 has been dead since the mid 90s as x86 microarchitectures uses micro-operations, combining the best of CISC and RISC.

 

[Recus]

Tears confirming that ARM is future. At least for datacenter/HPC.

 

[Fluffmeister]

Even if nothing happens, I just took delight from reading the comments and seeing people squirm.

 

[freeagent]

Their aim is to be the new supreme being.

 

[mak1skav]

All that I can read and understand from that movement is that Japan, who once was a giant and the biggest innovator in electronics and technology overall, will lose and sell another tech company to a richer player out there.

 

[TheLostSwede]

All that I can read and understand from that movement is that Japan, who once was a giant and the biggest innovator in electronics and technology overall, will lose and sell another tech company to a richer player out there.

You are aware ARM was British until not that long ago, right? It's got nothing to do with Japanese tech.

 

[lexluthermiester]

lol, the fanboyism in this thread is just amazing. Ya'll talking like all those billions belong to you.

The special snowflaking in your comment is as equally amazing. There is nothing wrong with people sharing how they feel about something that will very likely affect us all long term.

 

[jabbadap]

Hmh, what about Mali? Nvidia did try to license their own IP, but no-one took it(Kepler if I remember correctly).

 

[prtskg]

I think Softbank is going to make ARM public and these tactics are just to increase IPO.

 

[efikkan]

I think Softbank is going to make ARM public and these tactics are just to increase IPO.

Usually there are two main reasons to go public; raise a lot of money or inflate the stock price while slowly dumping shares. The latter is harder to do with a company which is not consumer-facing.

 

[ARF]

Nvidia has no enough cash to buy ARM.
No one in the right mind will allow this, the number 1 most important and strategic technology holding to be sold to one of the most corrupt, terrible and unfair US corp that could ever exist.

 

[Punkenjoy]

ARM is used pretty much everywhere in embedded devices, but have no chance against the performance of x86.

Amazon Graviton 2 ARM based server is already competitive with x86 after just 2 generation. The Apple ARM cpu is also quite powerful while still being a low power part.

The main advantages of x86 was never that this architecture was more efficient. It was always the installed base, the software catalogue of binary apps and the lead that intel had with their Fabs...

Now there are so many developpers working on arms devices, there are new technologies for Just in Time, pre-run binary translation, very good compiler and the best Fabs in the world is no longer Intel.

And the main reason ARM will take over is probably not because ARM is a better architecture, but to the fact that it's more open. There are more company and more money right now that have stakes in ARM than in x86.

That's it

 

[cyneater]

Am I the only one sort of excited about Nvidia buying arm. things going sideways

And then maybe open risc v will become a thing...
Or maybe IBM could pull there head out of there back side and make power PC great again.

 

[prtskg]

Usually there are two main reasons to go public; raise a lot of money or inflate the stock price while slowly dumping shares. The latter is harder to do with a company which is not consumer-facing.

Raising a lot of money seems like a good enough reason to me.

 

[lexluthermiester]

Or maybe IBM could pull there head out of there back side and make power PC great again.

That would take both a gruntload of money and a serious amount of innovations that IBM doesn't have up their sleeves..

 

[Anymal]

Nvidia is perfect for ARM future, but maybe not enough money to buy it.

 

[RoutedScripter]

Please no. Worst news of the year so far.

Anti-Trust, ... to much saturation of everything in a few companies is always bad for consumers.

Nvidia is perfect for ARM future, but maybe not enough money to buy it.

Who the hell are you, a cheerleader for corporate greed? A**hole.