Nvidia is quietly preparing to drop the company's long-standing acquisition of Arm,

[mr moose]

1 minute ago, hishnash said:

 

 

I agree Nvidia would have not risked ignoring the EU regulators due to the consequence of doing so. 

 

NVIDIA don't care about EU regulation for the consequences you are proposing as they just would not happen.   The economic fallout would be way too big.  Imagine telling every phone and tablet maker they can't sell their products in the EU.   Even if you limit it to only products made after the acquisition, they are not going to just stop selling phones and tablets. 

[hishnash]

38 minutes ago, mr moose said:

Even if you limit it to only products made after the acquisition, they are not going to just stop selling phones and tablets. 

No its products made using IP licensed after the acquisition. this would mean non of the big arm players would be effected at all as they all have existing perpetual licenses for ARM IP. 

Apple could continue to sell ARM based systems for ever (they might not be able to license new ARM v10 ISA but ARM v9 licenses have already been issued) and apple existing license terms might already include "all future versions clause anyway" 

The same for Quaclcome and Samsung they already have exiting licenses in place that will let them continue to produce (an adapt) ARM IP. 
 

Such a ruling from the EU would not impact existing products being sold but it would hinder Nvidia from being able to issue new licenses to vendors effectively forcing the merger to keep ARM holding seperate from Nvidia (new IP contracts that merge NV IP and ARM IP would not be of much intrest to any vendor).

Existing ARM hardware vendors (apple, Qualcomm etc) would even lobby the EU to issue this as a ruling as it would consolidate there position in the market, it is these companies that would be consulted on the impact and they would say 0 impact to them and thus thier consumers. 

the impact would mostly be felt buy people buying hardware manufactured by Nvidia (GPUs and networking gear). Any hardware made by them/including chips contracted and sold by them to an AIB would be subject to that ruling (Nvidia would not be able to sell GPUs to Europe).  I think this is exactly the move the EU would do if Nvidia just ignored them. (it is also just the same as what the EU regulators would do if an EU company ignored them). 

[mr moose]

20 minutes ago, hishnash said:

No its products made using IP licensed after the acquisition. this would mean non of the big arm players would be effected at all as they all have existing perpetual licenses for ARM IP. 

Apple could continue to sell ARM based systems for ever (they might not be able to license new ARM v10 ISA but ARM v9 licenses have already been issued) and apple existing license terms might already include "all future versions clause anyway" 

The same for Quaclcome and Samsung they already have exiting licenses in place that will let them continue to produce (an adapt) ARM IP. 
 

Such a ruling from the EU would not impact existing products being sold but it would hinder Nvidia from being able to issue new licenses to vendors effectively forcing the merger to keep ARM holding seperate from Nvidia (new IP contracts that merge NV IP and ARM IP would not be of much intrest to any vendor).

Existing ARM hardware vendors (apple, Qualcomm etc) would even lobby the EU to issue this as a ruling as it would consolidate there position in the market, it is these companies that would be consulted on the impact and they would say 0 impact to them and thus thier consumers. 

the impact would mostly be felt buy people buying hardware manufactured by Nvidia (GPUs and networking gear). Any hardware made by them/including chips contracted and sold by them to an AIB would be subject to that ruling (Nvidia would not be able to sell GPUs to Europe).  I think this is exactly the move the EU would do if Nvidia just ignored them. (it is also just the same as what the EU regulators would do if an EU company ignored them). 

 

So you think the EU will limit everyone to only having access current ARM tech moving forward?  Not going to happen.

[thechinchinsong]

9 hours ago, Master Disaster said:

If I'm remembering correctly (and I might not be) they are being forced to sell after a high ranking Chinese employee got fired then went rogue, refused to quit and tried stealing the entire company.

 

I'm 99% sure that's what happened.

Is that process still ongoing? I thought it was like a Chinese ARM branch? Whole tech news sphere has been so focused on shortages and crypto the past few years I completely forgot about that.

[Kisai]

18 hours ago, kladzen said:

Summary

Nvidia is quietly preparing to drop the company's long-standing acquisition of Arm, which has met with strong opposition.

 

Pretty much everyone on this forum with any insight to what nVidia and ARM make, whatsoever, knew this was going to be the outcome.

 

ARM can be sold, but it has to be sold to someone who remains an IP holding company focusing on ARM and not a fabless chip manufacturer, nor a chip foundry. So the best option is in fact to sell it back to the UK and block any sales to outside-UK companies to keep another "foreign subsidiary goes rogue (and steals the IP)"  situation again.

 

[Hassan170]

14 hours ago, JLO64 said:

Have there been any updates on the ARM China division going AWOL? I understand that the Chinese Government has basically no incentive to do anything in this situation, but that fact that it may have seriously affected a $40 Billion (failed) acquisition can't be ignored. Also, given that they've hired private security to defend "their" company, I kinda think that this is going to end in a Rainbow Six style raid.

subsequently leadiint to a hollywood movie which will inspire the creation of a video game that ubisoft will develop half assedly and die

[soldier_ph]

Nvidia has been holding onto this acquisition for so long that their Arm is now starting to fall of. I'll see myself out.

[LAwLz]

On 1/25/2022 at 5:49 PM, Master Disaster said:

If I'm remembering correctly (and I might not be) they are being forced to sell after a high ranking Chinese employee got fired then went rogue, refused to quit and tried stealing the entire company.

 

I'm 99% sure that's what happened.

That's not what happened. 

What happened was that the Arm China CEO started a new company (while the CEO of Arm China) and tried to recruit Arm customers to his new company. As a result, the Arm China broad of members fired him, which he refused to accept.

This happened AFTER the sale of Arm was initiated. The two things were not really related, and even if they were the sale of Arm to Nvidia happened before the "coup".

 

 

On 1/25/2022 at 5:19 PM, Mark Kaine said:

why does softbank even want to sell, dont they know this is a goldmine and risc arm is the future?

Because Arm isn't a goldmine. It's a public toilet that is currently clogged.

Softbank overpaid A LOT for Arm to begin with, and Arm has been losing money for a while now.

 

Softbank bought Arm for 31 billion dollars. I believe their net income was something like 300 million dollars at the time.

It would take SoftBank roughly 100 years to recoup their investment. Not only that, but recently I believe Arm has been losing money.

 

So basically, SoftBank overpaid like crazy when they first bought Arm, then Arm started losing money and now SoftBank just wants to get rid of it. In comes Nvidia, willing to pay an even crazier amount than SoftBank paid, at around 40 billion dollars.

 

Imagine making an investment that in a good scenario will only pay off in 100 years. Now imagine that the company suddenly started losing money. The longer you hold on to the investment, the more money you lose. In comes some crazy guy willing to not only take this company that is bleeding money, but also pay you 40 billion dollars for it.

SoftBank would honestly have to be run by complete morons to not take the deal Nvidia proposed.

 

Designing processors costs a ton of money, so the operational costs of Arm are really high.

Not only are the costs really high, the competition (RISC-V and POWER) are literally free. If you try to raise prices, your customers will just start looking at your competitors. You have to keep prices low despite it costing a fortune to stay competitive against not just Intel and AMD, but also the free competitors.

SoftBank tried to steer Arm in a different direction (IoT) in order to become more profitable, but as it turned out the IoT division did even worse than the other division. 

 

The only logical reason as to why Nvidia would be willing to pay so much for Arm is because they see it as a strategic investment, probably to become more competitive in the data center space.

[Mihle]

I don't think SoftBank ever planned to recoup their investment by the income ARM made. I think the goal always have been to buy it, raise its value and then sell it to someone else.

[Curufinwe_wins]

https://www.anandtech.com/show/17245/nvidiaarm-acquisition-officially-nixed-softbank-to-ipo-arm-instead

 

It is officially dead now.

----------------------

 

Hot take, but I think this is the worst possible outcome here. ARM isn't making enough money off of licensing to survive (there is a reason softbank wanted to sell), and an IPO is not going to generate anywhere near enough real internal resources to be able to compete with Apple, Intel or AMD in the long run. If ARMs stock holdings were actually close to class leading you wouldn't have seen Apple, Qualcomm, Samsung, Mediatek all spend huge development resources to diverge from their base offerings. I also don't have much faith that ARMs base designs are going to hold out much longer in server space, with other companies crushing it with either substantially modified variants or rise of newer architectures.

 

What I mean by IPO not generating resources is that stock price increase after set IPO doesn't make the company profit, and you can be darn sure Softbank will profit every cent it can from this leaving ARM in a shitty catch-22 with even less leverage than before.

I honestly believe getting sold to nVidia was the only way you could maintain some semblance of open access while also giving the resources needed to take the fight to other teams. Hell any resources given ARM is a loss making company.

Clearly most people disagree there, but that's why it is a hot take.

 

 

[dalekphalm]

2 hours ago, Curufinwe_wins said:

https://www.anandtech.com/show/17245/nvidiaarm-acquisition-officially-nixed-softbank-to-ipo-arm-instead

 

It is officially dead now.

----------------------

 

Hot take, but I think this is the worst possible outcome here. ARM isn't making enough money off of licensing to survive (there is a reason softbank wanted to sell), and an IPO is not going to generate anywhere near enough real internal resources to be able to compete with Apple, Intel or AMD in the long run. If ARMs stock holdings were actually close to class leading you wouldn't have seen Apple, Qualcomm, Samsung, Mediatek all spend huge development resources to diverge from their base offerings. I also don't have much faith that ARMs base designs are going to hold out much longer in server space, with other companies crushing it with either substantially modified variants or rise of newer architectures.

 

What I mean by IPO not generating resources is that stock price increase after set IPO doesn't make the company profit, and you can be darn sure Softbank will profit every cent it can from this leaving ARM in a shitty catch-22 with even less leverage than before.

I honestly believe getting sold to nVidia was the only way you could maintain some semblance of open access while also giving the resources needed to take the fight to other teams. Hell any resources given ARM is a loss making company.

Clearly most people disagree there, but that's why it is a hot take.

Frankly, if what you're saying is true, ARM is doomed pretty much no matter what.

 

NVIDIA owning it, sooner or later will become a problem for everyone *not* NVIDIA.

 

The ideal solution to me is something like the USB Implementers Forum or the HDMI Forum, or other scenarios where there's an Industry conglomerate that all pool resources into development. But since that was seemingly never an option, if it's NVIDIA buys it, and slowly turns it into something anti-competition, or Softbank spins it into an IPO and it slowly starves to death, I'd rather the second happen, since it's doomed either way.

 

Now, sure, there's no guarantee that NVIDIA would have actually done anything that would make it less useful to NVIDIA's competition, but their track record indicates it's likely enough to be a concern.

[LAwLz]

1 hour ago, dalekphalm said:

Frankly, if what you're saying is true, ARM is doomed pretty much no matter what.

Sadly, it is true.   

On 1/26/2022 at 9:13 PM, LAwLz said:

Arm isn't a goldmine. It's a public toilet that is currently clogged.

Softbank overpaid A LOT for Arm to begin with, and Arm has been losing money for a while now.

 

Softbank bought Arm for 31 billion dollars. I believe their net income was something like 300 million dollars at the time.

It would take SoftBank roughly 100 years to recoup their investment. Not only that, but recently I believe Arm has been losing money.

 

So basically, SoftBank overpaid like crazy when they first bought Arm, then Arm started losing money and now SoftBank just wants to get rid of it. In comes Nvidia, willing to pay an even crazier amount than SoftBank paid, at around 40 billion dollars.

 

Imagine making an investment that in a good scenario will only pay off in 100 years. Now imagine that the company suddenly started losing money. The longer you hold on to the investment, the more money you lose. In comes some crazy guy willing to not only take this company that is bleeding money, but also pay you 40 billion dollars for it.

SoftBank would honestly have to be run by complete morons to not take the deal Nvidia proposed.

 

Designing processors costs a ton of money, so the operational costs of Arm are really high.

Not only are the costs really high, the competition (RISC-V and POWER) are literally free. If you try to raise prices, your customers will just start looking at your competitors. You have to keep prices low despite it costing a fortune to stay competitive against not just Intel and AMD, but also the free competitors.

SoftBank tried to steer Arm in a different direction (IoT) in order to become more profitable, but as it turned out the IoT division did even worse than the other division. 

 

The only logical reason as to why Nvidia would be willing to pay so much for Arm is because they see it as a strategic investment, probably to become more competitive in the data center space.

 

Small correction to my quote above. I looked through the latest numbers from Arm and it seems like they didn't lose money during 2021. They made 27.5 million dollars in profits, before tax.

So if Arm keeps being this profitable, and somehow avoids paying any tax at all, it would take SoftBank roughly 1100 years to break even on their investment.

 

 

1 hour ago, dalekphalm said:

The ideal solution to me is something like the USB Implementers Forum or the HDMI Forum, or other scenarios where there's an Industry conglomerate that all pool resources into development. But since that was seemingly never an option, if it's NVIDIA buys it, and slowly turns it into something anti-competition, or Softbank spins it into an IPO and it slowly starves to death, I'd rather the second happen, since it's doomed either way.

I totally agree that Arm should turn into some kind of consortium. Let companies like Apple, Samsung, MediaTek, Qualcomm and other chipmakers control it. 

I think that would be the best option for everyone. Those companies won't ruin ARM since they rely on it themselves, and to them the cost of developing it would just be taken from their R&D budgets. Hell, they could even make it open and free if they wanted.

[Kisai]

1 hour ago, dalekphalm said:

Frankly, if what you're saying is true, ARM is doomed pretty much no matter what.

 

NVIDIA owning it, sooner or later will become a problem for everyone *not* NVIDIA.

Pretty Much(tm).

 

The problem would also be the same if Intel, AMD, or even Apple or Microsoft had purchased it. 

 

None of these companies have the interests of "ARM" in mind, only their own vertical integration. Quite frankly we can do with a lot less vertical integration, because it creates products and services that depend on other services and products from the same company, and induces vendor lock-in.

 

The only way ARM survives is by standing alone.

 

1 hour ago, dalekphalm said:

The ideal solution to me is something like the USB Implementers Forum or the HDMI Forum, or other scenarios where there's an Industry conglomerate that all pool resources into development. But since that was seemingly never an option, if it's NVIDIA buys it, and slowly turns it into something anti-competition, or Softbank spins it into an IPO and it slowly starves to death, I'd rather the second happen, since it's doomed either way.

 

Now, sure, there's no guarantee that NVIDIA would have actually done anything that would make it less useful to NVIDIA's competition, but their track record indicates it's likely enough to be a concern.

Yep, and nVidia's track record is nearly as bad as Apple's is in that regard. Apple acquires something, and suddenly nobody but Apple is allowed to use it (see Final Cut, the technology behind FaceID/ARKit, etc.) nVidia acquires something, CUDA-fies it, and now it doesn't work on anything but nVidia's own chips. Microsoft buys things as defensive acts towards competitors (it bought Nuance, the technology behind Siri, long after Apple bought the technology behind Microsoft's Kinect. Microsoft also bought n-trig, the technology behind stylus's used on touch screens, and then we have Activision as a defense against Sony)

 

It doesn't take much of a crystal ball to see the fallout of nVidia having ARM, in that Apple, Intel and AMD go and invest heavily in RISC-V or an entirely new ISA that nVidia isn't invited to.

 

[Curufinwe_wins]

3 hours ago, dalekphalm said:

Frankly, if what you're saying is true, ARM is doomed pretty much no matter what.

 

NVIDIA owning it, sooner or later will become a problem for everyone *not* NVIDIA.

 

The ideal solution to me is something like the USB Implementers Forum or the HDMI Forum, or other scenarios where there's an Industry conglomerate that all pool resources into development. But since that was seemingly never an option, if it's NVIDIA buys it, and slowly turns it into something anti-competition, or Softbank spins it into an IPO and it slowly starves to death, I'd rather the second happen, since it's doomed either way.

 

Now, sure, there's no guarantee that NVIDIA would have actually done anything that would make it less useful to NVIDIA's competition, but their track record indicates it's likely enough to be a concern.

I mean sure, but USB forum is a flipping disaster on the marketing (not as much technical end, but still not good there either)... and at least once ARM dies in the opensource sense, there would still be nVidia pushing some semblance of competition out there...

 

Here it will wither and collapse until someone buys out the IP and they have fallen too far behind to be relevant.

[Curufinwe_wins]

3 hours ago, Kisai said:

Pretty Much(tm).

 

The problem would also be the same if Intel, AMD, or even Apple or Microsoft had purchased it. 

 

None of these companies have the interests of "ARM" in mind, only their own vertical integration. Quite frankly we can do with a lot less vertical integration, because it creates products and services that depend on other services and products from the same company, and induces vendor lock-in.

 

The only way ARM survives is by standing alone.

 

Yep, and nVidia's track record is nearly as bad as Apple's is in that regard. Apple acquires something, and suddenly nobody but Apple is allowed to use it (see Final Cut, the technology behind FaceID/ARKit, etc.) nVidia acquires something, CUDA-fies it, and now it doesn't work on anything but nVidia's own chips. Microsoft buys things as defensive acts towards competitors (it bought Nuance, the technology behind Siri, long after Apple bought the technology behind Microsoft's Kinect. Microsoft also bought n-trig, the technology behind stylus's used on touch screens, and then we have Activision as a defense against Sony)

 

It doesn't take much of a crystal ball to see the fallout of nVidia having ARM, in that Apple, Intel and AMD go and invest heavily in RISC-V or an entirely new ISA that nVidia isn't invited to.

 

See that is if ARM even can survive alone, which doesn't seem likely to me...

 

And not as though AMD buying ATI has been an abject irrevocable disaster compared to say a world in which ATI dies on its own (yes AMDs cpu end being hot trash for the better part of the decade probably stripped any and all resources from the GPU buisness...)

 

I guess what I'm saying is not that I like vertical integration, but you can't compete with that without your own, or a metric assload of condition free money, which clearly ARM isn't getting on their own.

 

I also don't have a lot of faith in industry consortiums because we think they wouldn't ruin it, but they would be claiming that everyone else was ruining it and their way was the obvious and only solution, and so you get mired in inconsistent direction and do-nothing-ness until death by attrition.

 

Speaking as someone who sits in on ASME code discussions lolz.

 

 

----------------------

EDIT, in a not terribly shocking follow along to the deal falling through and the Softbank trying to do damage control/hype for the IPO, they released a presentation (using numbers with some pretty intense disclaimers and not audited at all, but that doesn't make it automatically wrong) including EBITA for ARM. Supposedly after drop of more than 50% from 2016 to 2019, 2020 and 2021 were near record recovery, which would be much more promising than had appeared.

 

https://www.forbes.com/sites/patrickmoorhead/2022/02/08/surprise-nvidia-deal-off-arm-is-very-profitable-has-a-new-ceo-and-rene-haas-is-looking-forward-to-its-ipo/?sh=3ca2969b6649

 

It is worth noting these numbers do *NOT* align with the most recent EBITDA presentation I had found which was from Q1 2020. Obviously some things have changed (cough cough), but there may be some number adjustments that have been included for marketing purposes. Likewise, EBITDA is not the same thing as profit, but things might not be as dire as had been projected based on the more recent softbank disclosures.

 

[Kisai]

9 hours ago, Curufinwe_wins said:

See that is if ARM even can survive alone, which doesn't seem likely to me...

 

And not as though AMD buying ATI has been an abject irrevocable disaster compared to say a world in which ATI dies on its own (yes AMDs cpu end being hot trash for the better part of the decade probably stripped any and all resources from the GPU buisness...)

 

Look, my view here is that AMD acquiring ATI was not in the industries best interests, it was only in AMD's interest because Intel was doing APU designs and AMD had no APU design, which had AMD not bought ATI, none of the game consoles since the GameCube would be what they are. Ask yourself why Nintendo didn't make another TV console. PowerPC wasn't really a good fit for game consoles, but both Sony and Microsoft ALSO used Power for the 360/PS3 for some bizarre reason. When they switched to an AMD APU,  now the PS4 and PS5 can be backwards compatible for once. Same with Microsoft's XB1 and "Xbox Series" consoles. Nintendo? The only thing the switch could have been backwards compatible with is the GBA. The DS/DSi, and 3DS, despite having the same ARM-compatible parts, have input gimmicks that don't support backwards compatibility without re-porting the game. The GC, Wii, and Wii games can be ported (read: recompiled) but Nintendo historically didn't have "one platform" SDK, so that's also a bunch of effort. So Nintendo pretty much shot themselves in the foot for backwards compatibility while it's competition didn't have to, for once.

 

But, what would have happened had ATI not been acquired by AMD, and wasn't acquired by anyone else? Well for one, I think it would be in a similar position to nVidia is now, where it produces GPU chips that AIB's turn into full GPU add in cards. But I also believe that Nintendo would have still used their GPU, but with ARM parts early on, and Intel would have just licensed ATI's GPU's instead of developing Xe, and Apple would have just kept using ATI's GPU parts in their M1. But AMD would be at a losing disadvantage against Intel, and not have the resources for APU parts. 

 

So in the short run, we will see. My opinion is that ARM might struggle a bit to become profitable, but they will likely raise licensing costs for future designs, or maybe be forced to build standard ATX/mATX/ITX motherboard design around an ARM CPU for AIB's to build, now that Windows can boot on ARM. I'm sure there are plenty of people who want to keep building silly ITX/NUC style systems as media players/STB/HTPC as silent as possible. This is still an under-served market that basically only has the nVidia Shield STB (Nintendo Switch class hardware) and Raspberry Pi (woefully underpowered for gaming) in it, with neither device being very useful as a desktop PC.

 

[Curufinwe_wins]

7 hours ago, Kisai said:

Look, my view here is that AMD acquiring ATI was not in the industries best interests, it was only in AMD's interest because Intel was doing APU designs and AMD had no APU design, which had AMD not bought ATI, none of the game consoles since the GameCube would be what they are. Ask yourself why Nintendo didn't make another TV console. PowerPC wasn't really a good fit for game consoles, but both Sony and Microsoft ALSO used Power for the 360/PS3 for some bizarre reason. When they switched to an AMD APU,  now the PS4 and PS5 can be backwards compatible for once. Same with Microsoft's XB1 and "Xbox Series" consoles. Nintendo? The only thing the switch could have been backwards compatible with is the GBA. The DS/DSi, and 3DS, despite having the same ARM-compatible parts, have input gimmicks that don't support backwards compatibility without re-porting the game. The GC, Wii, and Wii games can be ported (read: recompiled) but Nintendo historically didn't have "one platform" SDK, so that's also a bunch of effort. So Nintendo pretty much shot themselves in the foot for backwards compatibility while it's competition didn't have to, for once.

 

But, what would have happened had ATI not been acquired by AMD, and wasn't acquired by anyone else? Well for one, I think it would be in a similar position to nVidia is now, where it produces GPU chips that AIB's turn into full GPU add in cards. But I also believe that Nintendo would have still used their GPU, but with ARM parts early on, and Intel would have just licensed ATI's GPU's instead of developing Xe, and Apple would have just kept using ATI's GPU parts in their M1. But AMD would be at a losing disadvantage against Intel, and not have the resources for APU parts. 

 

So in the short run, we will see. My opinion is that ARM might struggle a bit to become profitable, but they will likely raise licensing costs for future designs, or maybe be forced to build standard ATX/mATX/ITX motherboard design around an ARM CPU for AIB's to build, now that Windows can boot on ARM. I'm sure there are plenty of people who want to keep building silly ITX/NUC style systems as media players/STB/HTPC as silent as possible. This is still an under-served market that basically only has the nVidia Shield STB (Nintendo Switch class hardware) and Raspberry Pi (woefully underpowered for gaming) in it, with neither device being very useful as a desktop PC.

 

So the AMD/ATI example was kind of contrived because the situations were flipped, ATI had a pretty swell time right at the start of that merger while AMD was struggling like crazy... I don't actually think it was good for the merger in that moment (though it 100% saved AMD as we know it today).

 

I guess my point was more that if you start from the assumption that the acquired company cant compete without significantly more resources (something I do hold to be true for ARM, but perhaps less obvious based on the supposed release from Softbank), the choice is wither on the stalk and have nothing or get bought out. In that sense, getting bought out is almost always more preferable than straight up dying as long as the org buying isn't already one of the main competitors in the same market. Sure would it be better that ARM was given huge funding and stayed independent of nVidia? Absolutely, I just dont think that is a realistic or plausible option. The biggest issue with ARM is that the majority of its products by volume (and profit margin for their source companies) come from perpetual license holders, which I don't expect they have any leverage over to raise royalty fees (see Apple). If they can make better new products to capture additional streams on future designs, that is a significant growth opportunity, but I worry that needing to do better to get money and needing to get money to do better is exactly the way in which a company like AMD gets mired in almost a decade of decay barely limping along (again thank goodness ATI was covering during most of those lean years).