Nvidia in Advanced Talks To Buy Chip Giant Arm (bloomberg.com) 66
Nvidia is in advanced talks to acquire Arm, the chip designer that SoftBank Group bought for $32 billion four years ago, Bloomberg reported Friday, citing sources familiar with the matter. From the report: The two parties aim to reach a deal in the next few weeks, the people said, asking not to be identified because the information is private. Nvidia is the only suitor in concrete discussions with SoftBank, according to the people. A deal for Arm could be the largest ever in the semiconductor industry, which has been consolidating in recent years as companies seek to diversify and add scale. Cambridge, England-based Arm's technology underpins chips in products including Apple Inc. devices and connected appliances. Financial Times, which has also independently reported about the deal talks today, adds: Buying Arm would further consolidate Nvidia's position at the centre of the semiconductor industry, at just the moment when the British chip designer's technology is finding broader applications beyond mobile devices, in data centres and personal computers including Apple's Macs. Arm would transform Nvidia's product line-up, which until now has largely focused on the high end of the chips market. Its powerful graphics processors -- which are designed to handle focused, data-intensive tasks -- are typically sold to PC gamers, scientific researchers and developers of artificial intelligence and self-driving cars, as well as cryptocurrency miners. The deal could alarm Arm's other big licensees, including Apple, Broadcom and Qualcomm, which may fear a unique asset being taken over by a potential competitor such as Nvidia.
A sale would mark a stunning reversal for SoftBank founder Masayoshi Son, who declared that Arm would be the linchpin for the future of the technology investment conglomerate. The company has failed to thrive under SoftBank, according to people with direct knowledge of the matter. Neil Campling, an analyst at Mirabaud, noted that Arm's annual revenues had risen from $1.2bn to $1.9bn since SoftBank bought it in 2016, while Nvidia's have roughly tripled in the same timeframe.
A sale would mark a stunning reversal for SoftBank founder Masayoshi Son, who declared that Arm would be the linchpin for the future of the technology investment conglomerate. The company has failed to thrive under SoftBank, according to people with direct knowledge of the matter. Neil Campling, an analyst at Mirabaud, noted that Arm's annual revenues had risen from $1.2bn to $1.9bn since SoftBank bought it in 2016, while Nvidia's have roughly tripled in the same timeframe.
Woah, there, cowboy (Score:3)
the semiconductor industry, which has been consolidating in recent years as companies seek to diversify
Uhh... nope. Less is not more.
Re:Woah, there, cowboy (Score:5, Informative)
Fewer companies, each making more types of products.
Re: (Score:3)
Leads to monopolistic behaviour in the absence of lynch mobs (forget the monopolies and mergers commission - they have evidenltly sold out - see phone network operators).
Re: (Score:2)
But still, that is what companies seeking to diversifying means. You as a company diversify into more types of products and or services.
Re: (Score:2)
Bye bye Intel (Score:2)
First Apple ditched you, then your once-leading process nodes fell behind the competition significantly, and now this. Intel you better wake up quickly, the next thing to happen that youâ(TM)re not gonna like is Microsoft releasing its 64bit x86 emulator for Windows on ARM.
What are you going to do then?
Re: (Score:2, Informative)
AMD bought up ATI.
nVidia buying up ARM is just as dangerous.
Intel are at risk because they can't do the graphics part - despite years of supposedly putting this stuff in chips, it's still just business-level graphics.
They could have put nVidia-level silicon in there and owned the market. Instead we have Optimus-shite where you have to pair the CPU with a real GPU if you want anything decent to happen.
That Intel don't have a graphics partner to put in their chips is really going to hurt them in the next few
Re: (Score:2)
Xe.
Re: (Score:2)
see also: Larrabee
Re: (Score:2)
AMD bought up ATI. nVidia buying up ARM is just as dangerous.
Quite a bit more, I'd think. The former probably saved one of the companies in the long run. nVidia controlling ARM could have serious impact on any computing field trying to get rid of x86/AMD64.
AMD, Nvidia and opensource (Score:3)
AMD bought up ATI.
nVidia buying up ARM is just as dangerous.
It's going to be worse.
At least AMD has put their promise into practice and Radeon cards nowadays enjoy a unified driver stack (AMDGPU) that is opensource on platforms where it is possible (Linux) and plays nicely with local infrastructure (AMDGPU uses standard Linux features for, e.g.: hybrid GPUs).
Nvidia has been absolutely catastrophic for the opensource world. No opensource driver, closed hard to obtain signed firmware, a driver which is basically a recompile of the Windows driver no matter on which pla
Re: (Score:3)
Intel doesn't have to change.
The Arm licensing model asks for a large sum upfront with small royalty fees. This was decided in order to put money quickly into the Arm business and to avoid relying on royalties alone, which is a gamble for a business, because the profit from royalties depends on the success of a licensee's product.
Intel already has got an Arm license and has been using Arm CPUs in some of their products, i.e. as micro controllers on some of their network devices afaik.
Arm CPUs follow a diffe
That's what Blackberry thought.... (Score:3)
The whole industry has been shifting since the advent of the smartphone in 2006. Intel has lacked the agility to adapt. Competing companies have not only filled the void in non-x86 architecture, but i
Re:Bye bye Intel (Score:4, Interesting)
Arm CPUs follow a different design approach when compared to x86. Arm CPUs are designed as low-power CPUs, because their makers didn't want their CPU to be packaged in ceramic, but in the cheaper plastic and without the need for a cooler. This explains the success of Arm as a low-power CPU, which was a decision made about 30 years ago.
ARM was designed for speed in the first place, it was running circles around 80x86, 68000 and the other processors in the late '80s (I used to run MS-DOS, WordPerfect, dBase II and the TopSpeed compiler in an 80x86 software emulator on ARM in 1987 [computinghistory.org.uk]).
The low power requirements were discovered when Acorn engineers Furber and Wilson failed to switch of their development board because the CPU kept running on a capacitor or a spinning fan.
Re: Bye bye Intel (Score:2)
This is the key, the start of the RISC journey.
Over the last few years arm has been driven by marketers and makeweights and bought Pelion which is a white elephant, and failed to invest in mbed, their IoT tech. A return to an engineer led culture with a focus on the high value markets that capitalises on armâ(TM)s scale is a no brainier for NVIDIA.
Re: (Score:2)
ARM was designed for low power, and that's what enabled it to succeed where its RISC competitors failed. It may have outrun the 68000 and the 286, but was overtaken by subsequent generations. However, power consumption was where they remained the tops, and that got them adapted in the cellphone market.
Responding to the GP, Windows on ARM will go nowhere, as Windows RT found out. Heck, these days, there are even fewer Wintel apps, since the bulk of software seems to have become enamored w/ iOS and Andro
Re: (Score:2)
Actually, no, you're wrong. All CPUs are designed for speed and none has been designed to be slow of course, but you're simply attributing the change from RISC to CISC as an Arm feature, which it isn't. The 8086 and 68000 are older than any Arm CPU and are of course slower as such. But 80686 and 68060 were faster than Arm for example. Arm then didn't outperform other RISC CPUs, it was way behind the DEC Alphas, and x86 began to use RISC internally, which is why Arm still isn't a high performance CPU nor has
Re: (Score:2)
All CPUs are designed for speed and none has been designed to be slow of course
Except the RCA 1802 perhaps, which can clock down way to 0 Hz on the fly.
but you're simply attributing the change from RISC to CISC as an Arm feature, which it isn't.
I guess you mean "CISC to RISC" in stead of "RISC to CISC". And no, I never wrote that ARM was unique or special in this respect.
The 8086 and 68000 are older than any Arm CPU and are of course slower as such. But 80686 and 68060 were faster than Arm for example.
Now you are comparing mid-1990s CPUs to a mid-1980s CPU. Obviously the other CPUs caught up and overtook the ARM on speed, while ARM's enduring success was due to its low power requirements.
Arm then didn't outperform other RISC CPUs
I never claimed that. (The other RISC systems at the time however were at least an order of magnitude more expensive, an
Re: (Score:2)
I wasn't comparing mid-1990s CPUs to a mid-1980s CPU, but to the following Arm CPUs. Arm never achieved a status of making high performance CPUs. The same goes for Intel (i960) and Motorola (88k) by the way.
Re:Bye bye Intel (Score:4, Insightful)
You should not celebrate the end of the IBM PC compatible platform.
Everything that is not a PC compatible nowadays is effectively a game console that agreed to run the same user level software, with the OS being an encrypted signed proprietary piece of hell that probably sell all your data to shady data brokers that probably should be illegal.
Re: (Score:2)
Re: (Score:2)
uhhggg... consolidation (Score:2)
The nice thing about ARM is that it's a platform people desing silicon too, not some vendor's final product chip. NVIDIA's bussiness model is like Intels, They define the silicon.
Re: (Score:1)
Nvidia see lots of IP to license, change the terms of existing licenses and laugh all the way to the bank.
Now that Apple have announced that they are ditching Intel, I wonder if Nvidia see a chance to screw Apple over their licensing. Apple won't be happy if that happens. They'll have their lawyers going over their license agreement in fine detail just to see if there are any loopholes for Nvidia to exploit.
Apple and Nvidia are not the best of friends. Nvidia won't release drivers for MacOS due to Apple's t
Re:uhhggg... consolidation (Score:4, Interesting)
ARMs success comes in no small part from its licensing. Nvidia would be insane to tamper with that.
Re: (Score:3)
Re: (Score:2)
Nvidia will most certainly tamper with ARM's licensing model if they buy them. There is literally no reason to buy ARM unless you intend to alter the licensing model because any licensee can already do anything they want including design their own chips using the instruction set.
I'll say it again, nVidia has no incentive to buy ARM unless their intent IS to alter the licensing model.
Re: uhhggg... consolidation (Score:3)
Apple and Nvidia are not the best of friends. Nvidia won't release drivers for MacOS due to Apple's terms.
If I was Tim Cook, I'd be starting a research project to develop or use a CPU that does not have this sort of threat hanging over it. Risc V might be a good starting point.
According to nVidia, it was Apple that refused to sign nVidiaâ(TM)s Drivers.
However, as other people have pointed-out in other Forums, Appleâ(TM)s spectacular performance in their Arm-based SoCs is based in large measure on their custom, homegrown subsystems (âoeperipheralsâ in Arm-parlance), and thus Apple could relatively easily pivot to another, likely-homegrown CPU core, and not feel a thing.
Afterall, with their Arm Architecture License, they are already designing their own CPU silic
BAD (Score:5, Insightful)
This may help drive up support for RISC V; otherwise, it's not good for the industry.
Re: (Score:2)
That is what I wrote about a week ago [slashdot.org]
RISC V (Score:4, Insightful)
Another reason for everyone contribute to, and then move to, RISC V.
There is no reason our CPUs have to be locked into a proprietary tech when more open alternatives are available.
Re: (Score:2, Informative)
Risc-V is interesting from a technical perspective, but won't be relevant in markets anytime soon...
Re:RISC V (Score:5, Interesting)
Risc-V is interesting from a technical perspective, but won't be relevant in markets anytime soon...
That's where you're wrong kiddo. Western Digital has invested in moving to Risc-V. [anandtech.com] FYI, they have released their cores to the public [github.com] and seem to be in the process of testing.
Now, WD could just be hedging their bets and should this buyout go through then they are going to be ahead of the game on all their competitors. Nvidia is incapable of playing nice and as a result, they are going to drive customers to looking for alternatives to ARM. ARM is no longer a unicorn but there has been no incentive to switch. That incentive is Nvidia's management because they have a completely different culture which will no doubt obliterate all the good will that ARM has built up.
regulatory scrutiny? (Score:1)
Any deal with Nvidia, which is a customer of Arm, would likely trigger regulatory scrutiny as well as a wave of opposition from other users of the company’s technology.
What with the UK out of the EU, any regulatory resistance will have to come from the US or UK regulators.
Not entirely convinced that that's not a rather mixed blessing in this particular instance.
The new Intel (Score:5, Interesting)
Judging by Nvidia's tactics trying to make new GPU extensions as proprietary as possible (RTX), vendor-locked game graphics options (PhysX early on, TressFX), and their prominent marketing slogans (the way it's meant to be played)...
I'd say we'd just have another Intel on our hands
Re: (Score:2)
Re: (Score:2)
Tegra never died. It has been flourishing this entire time. It is called the Nintendo Switch.
Re: (Score:2)
Re: (Score:2)
I didn't realize that 59 million units sold counted as "life support"
Re: (Score:2)
When Qualcomm sells 4 times more snapdrogons every quarter, yeah that kind of blows.
Re: (Score:2)
And the Shield TV. Probably not the volume that Nintendo moves, but it's still a great product that sells.
Wish I could buy arm holding stock (Score:5, Interesting)
Before ARM is sold to NVidia, I predict that Apple will get into a bidding war. I could see other players getting into the war as well, such as Amazon and AMD. On the the major values of ARM was that it was a neutral vendor that didn't play sides. Going in the hands of NVidia will change that, and potentially realign the market.
Re: (Score:3)
I do not see Apple in any way bidding for ARM. AFAIK they have a perpetual licence to use the instruction set, so not much of a gain, then they would have to support Qualcomm and all of the other ARM customers which would be a technical and regulatory nightmare. Not happening.
Re: (Score:3)
This is the main reason the biggest people with money like Apple, Intel, and Qualcomm are staying out.
It's why nVidia and AMD (if they wanted) are likely the few companies that can buy it, but neither of them is particularly flush with cash.
Quite likely SoftBank would just retain that investment because they know that Apple and Qualcomm and Intel, if they could get over regulatory hurdles, will be far more interested and thus more valuable. Just the regulatory headaches aren't worth the
Re: Wish I could buy arm holding stock (Score:2)
Before ARM is sold to NVidia, I predict that Apple will get into a bidding war.
They were actually privately offered to buy Arm a few months ago, and turned it down.
Re: (Score:2)
This will be good for RISC-V (Score:3)
No one likes to put money in a competitor's pocket for a multitude of reasons.
If NVIDIA buys ARM, Fujitsu, makers of, among other things processors for an ARM based Top 500 Supercomputer, will not be to happy to put money in the pockets of a supercomputer provider competitor. Same for Amazon (gravitron), Ampere, Cavium and Qualcomm...
I guess Qualcomm, Mediatek, Asrock and Samsung will not be to happy to put money in a company that competes with them with processors sold in the open market for set top boxes, consoles, tablets and such (not phones, tough, at least for the time being).
Then, there is the question of Mali Vs. Tegra.
Does a company, as an ARM licensee, gets access to NVIDIA graphics tech? or just plain old Mali, and nVidia reserves the good graphics to themselves?
So, If this purchase goes through, I foresee a lot of companies moving from ARM to RISC-V. It will take a few lustres, but it will happen.
Re: (Score:2)
#12: AI Bridging Cloud Infrastructure (ABCI)
NVIDIA Tesla V100 SXM2
#24: Gadi
NVIDIA Tesla V100 SXM2
Mellanox HDR Infiniband (also NVIDIA)
#53: ITO - Subsystem A
Mellanox InfiniBand EDR
Fujitsu is already putting money into NVIDIA.
Re: (Score:3)
Actually it is a problem for samsung. You see, Samsung has to pay royalties for the priviledge to use the ARM instruction set on exynos, and sells exynos in some markets where nVidia sells tegras.
Also, apple has to pay royalties for the priviledge to use the ARM instruction set, while also selling things like AppleTV and Apple Speakers, that compete with things (like nVidia Shield) that nVidia Makes.
Huawei has to pay for the priviledge of using the ARM instruction set in Kirin. And these go on tablets that
This is really bad.... (Score:2)
They will find a way to screw everyone, this would be a catastrophe to ARM...but perhaps a warning to the market and invest in a more open cpu, like Risc-V or POWER.
Nvidia is not a team player, this wont end well.
Could the chickens come home to roost for Apple? (Score:3)
Apple's poor treatment of Nvidia over the past few years--dickish behavior of not allowing them to sign their drivers, basically not allowing Nvidia GPUs on the Mac, due to a long-ago beef over Nvidia sorta pre-announcing new Apple hardware--could just come back to bite Apple in the butt if Nvidia gets ahold of ARM. I'm sure Apple has a pretty solid licensing agreement in place with ARM currently, but their move to Apple Silicon will still necessarily rely on ARM IP. Might just make a future re-licensing discussion interesting; can't imagine Nvidia doesn't have that in mind right now.
Re: (Score:3)
Nope. Apple has an irrevocable permanent ARM license from being a partial owner / investor way back in the Newton days. This puts them on the same footing as Samsung (Exynos), Qualcomm (Snapdragon), Intel (Xscale), etc.
They probably still get frozen out of GPU tech, but the actual ARM stuff they're good on.
Re: (Score:2)
Re: Could the chickens come home to roost for Appl (Score:3)
Nope. Apple has an irrevocable permanent ARM license from being a partial owner / investor way back in the Newton days. This puts them on the same footing as Samsung (Exynos), Qualcomm (Snapdragon), Intel (Xscale), etc.
They probably still get frozen out of GPU tech, but the actual ARM stuff they're good on.
Apple has a perpetual Architecture license from Arm. They already design their own âoeArm-compatibleâ CPUs.
And as far as GPUs go, they have been rolling their own proprietary, native Metal2-based, GPUs for a few years now, and they are already quite good (including hardware ray-tracing, BTW).
Re: Could the chickens come home to roost for Appl (Score:2)
I'm sure Apple has a pretty solid licensing agreement in place with ARM currently, but their move to Apple Silicon will still necessarily rely on ARM IP. Might just make a future re-licensing discussion interesting; can't imagine Nvidia doesn't have that in mind right now.
The fact that Apple turned down the purchase of Arm when Softbank offered it to them privately a few months ago, tells me that Apple's own ARMy of engineers and lawyers aren't the least bit worried about the prospect of another owner of Arm.
Re: (Score:2)
It wasn't just the product leak, it was the failed GPU's that required a full motherboard swap, as well as Nvidia asking like complete blow-hards towards Apple engineers. It's all well-document. Apple got tired of their behaviour, and coming from Apple, you know how bad it must have been.
"Failed to thrive" (Score:2)
What a crazy world where a better than 50% revenue increase in under 4 years is a failure to thrive.
I have an idea (Score:2)
Apple is biding its time (Score:2)
Not the Best Title (Score:2)
I read it and thought “Why does Intel want to buy a giant arm?”
Re: (Score:2)
Errr, nvivida
Typical predatory US companies (Score:2)
The UK should have forbid the sale on the grounds of national security.