Qualcomm Opposes Nvidia's Acquisition of Arm as Deal Comes Under Increasing Scrutiny

[luvbug]

Why would tech companies spend money to contribute to the purchase of Arm for a consortium, only to license it to their competitors at a discount?

The consortium idea was to negate any single entity from controlling it, but allowing SoftBank to sell their asset. Perhaps fair-use licensing is a step too far, at least for now, but I believe it serves everyone's interest, that rely on ARM, to ensure it stays available on neutral terms. Nvidia won't provide that, and neither will Qualcomm. I actually shudder to think what Qualcomm would do if they were allow to purchase it!! JMO.

 

[jz0309]

I mean, I don’t know. This is why I’m asking. But as far as I understand, neither the European union or China (or any other state or region) has the jurisdiction to prevent this deal from happening, right? Only the US and Japan can legally prevent this purchase from happening?

no, regulators in EU, UK and China have a say ... that's how it works

 

[jz0309]

I mean, I don’t know. This is why I’m asking. But as far as I understand, neither the European union or China (or any other state or region) has the jurisdiction to prevent this deal from happening, right? Only the US and Japan can legally prevent this purchase from happening?

here's an example: US company (Google) buying US company (Fitbit) - EU had to approve ...


Google's Fitbit acquisition wins EU approval - The Vergewww.theverge.com › 2020/12/17 › googles-fitbit-deal-...

 

[Stella]

They have said that they will not change the licensing model. Of course you have to ask what benefit they are expecting from spending a huge amount of money to acquire ARM. At the very least Nvidia's needs would be prioritized, potentially to the detriment of other licensees. The beauty of ARM's business model was always that they were completely driven their licensees' needs and feedback. That would no longer be the case if they were part of Nvidia.

Yea, "nothing will change". Famous last words.

How many companies have uttered such statements when taking over another company and do exactly the opposite? Lots.


Anyone who believes "nothing will change" statement is beyond gullible.

 

[LDN]

Licensing an instruction set in this day and age is starting to look stupid. While it is just my own uneducated opinion, an instruction set shouldn't be a thing that can be licensed, core designs however make sense because every generation takes months and years of work. That's not the same for an instruction set.

 

[nwcs]

Well, given that the DOJ has routinely allowed mega mergers and acquisitions that reduced competition or created awkward situations why should this one be any different? Not that I think the deal should go through but it likely will.

 

[CubeHacker]

Forgive me for not understanding the complexity of the situation. Currently softbank, a mobile provider in japan owns arm, purchased back in 2016.

It seems to me that just about any company with enough money to purchase arm @40 billion dollars would use it to benefit themselves and potentially limit licensing.

So why isnt arm just spun off as its own independent company like it originally was? Arm apparently generates about 1.5 billion a year in revenue. Should be enough to stay afloat, and since their primary business model would be to continue to evolve arm and license the technology, no one could complain.

 

[farewelwilliams]

Oh, so Qualcomm has a problem with another company denying access to something essential.

 

[cmaier]

Forgive me for not understanding the complexity of the situation. Currently softbank, a mobile provider in japan owns arm, purchased back in 2016.

It seems to me that just about any company with enough money to purchase arm @40 billion dollars would use it to benefit themselves and potentially limit licensing.

So why isnt arm just spun off as its own independent company like it originally was? Arm apparently generates about 1.5 billion a year in revenue. Should be enough to stay afloat, and since their primary business model would be to continue to evolve arm and license the technology, no one could complain.

Because the owners of Arm will take as much money as they can get, and they get richer by selling to NVIDIA than by spinning out Arm.

 

[CubeHacker]

Because the owners of Arm will take as much money as they can get, and they get richer by selling to NVIDIA than by spinning out Arm.

So basically, greed, and turning a quick buck

 

[DrV]

I understand the worries about Nvidia becoming too dominant. The mechanism behind this may not be so obvious, after all.

Quite obviously, the possible merger would make life more difficult for Qualcomm, as Nvarm could provide very good CPU+GPU SoCs. This may not be that bad news for smart device or computer manufacturers, rather on the contrary.

Another possibility is that Nvidia aims at hiking ARM license prices. There are, after all, a dozen billion Cortex M controllers shipped every year in addition to the 10 billion or so ARM CPU cores. If ARM had a real monopoly, this would sound like a great plan.

Once, a very experienced friend of mine taught me that a monopoly is far too valuable to be wasted by high prices. ARM has been in the licensing business for a long time, and they have tried to find the sweet spot. If you tighten the screw too much, competing technologies will emerge.

Processor architectures come and go. A good architecture is energy efficient, requires little real estate, has low cost, and there are good tools available. ARM ticks the boxes at the moment, but it is not impossible to design as good or even better architectures if the price becomes unbearable.

My guess is that Nvidia is quite flexible when it comes to licensing the ARM cores or ISAs. They are aiming at highly integrated SoCs with a lot of edge computing capacity.

 

[cmaier]

So basically, greed, and turning a quick buck

When you sell stuff so you try to sell it for less than the market will bear?

 

[Piggie]

Would love to see multicore GTX graphics from Nvidia build directly into some new custom ARM designs for phones, tablets, consoles? anything.....

It's a shame we have companies great at ARM and others great at GPU's.
We al want one that can pull both sides together.

 

[BeefCake 15]

I hope nvidia are refused by regulators to buy ARM. It’s not in the best interest of the consumer. And yes, I know this doesn’t affect Apple.

It does somewhat because Apple still has licenses for ARM. It's an ARMs race at this point betweent Nvidia, Apple, M$, Google and AMD.

 

[BeefCake 15]

When you sell stuff so you try to sell it for less than the market will bear?

Everyone turns into a None Profit saint when it comes to reading about corporate profits...

 

[Td1970]

So basically, greed, and turning a quick buck

What makes the world go round.

 

[Krevnik]

Would love to see multicore GTX graphics from Nvidia build directly into some new custom ARM designs for phones, tablets, consoles? anything.....

It's a shame we have companies great at ARM and others great at GPU's.
We al want one that can pull both sides together.

I'm a little skeptical that Nvidia is well positioned here. Mostly because like Intel of years past, they've been designing for a high power envelope, mixing some efficiency gains with brute force expansion of the power envelope. But also because Nvidia has tried to get into ARM designs already, and pretty much failed to make a dent in the market.

Tegra for phones is dead. Tegra X1's successor will be focused on cars. I'm honestly wondering where Nintendo will go for their next SoC, assuming their next bit of hardware is an evolution of the Switch. This makes me wonder what Nvidia's game here is, considering it looks like they've already picked up their chips and are attempting to walk away from the table.

 

[Piggie]

Yes you are correct.
Nvidia's GPU's are pretty much untouchable, and pretty unobtainable (thanks scalpers)
But as you say, they are going flat out with MASSIVE GPU's and thousands of cores to create this super high end performance.
And this means, as you can using a lot of power to achieve this.
Doing a total turn-around and going all in on the low power market for portables would be a big shift for them.

I was sad what happened with the Tegra line (which I think is still used in some nice devices)
I recall the early graphics demo's of what Tegra was going to be able to do, and it looked stunning.
It was a shame something went wrong (I don't know what) and it was never continued to build into a full line of products.
They seemed to get off to an great start with much promise and then didn't follow through.....

 

[EvilEvil]

I can’t wait for the privilege to buy one (the card that is)

I had no trouble getting a 3080.

 

[DrV]

Yes you are correct.
Nvidia's GPU's are pretty much untouchable, and pretty unobtainable (thanks scalpers)
But as you say, they are going flat out with MASSIVE GPU's and thousands of cores to create this super high end performance.
And this means, as you can using a lot of power to achieve this.
Doing a total turn-around and going all in on the low power market for portables would be a big shift for them.

I was sad what happened with the Tegra line (which I think is still used in some nice devices)
I recall the early graphics demo's of what Tegra was going to be able to do, and it looked stunning.
It was a shame something went wrong (I don't know what) and it was never continued to build into a full line of products.
They seemed to get off to an great start with much promise and then didn't follow through.....

Interestingly, the very high end and the very low end meet each other when it comes to power efficiency.

In the low end you just do not have the electric power, as you want to conserve the battery. In the high end you would have as much power as you wanted, but as all power you feed into the chip turns into heat, cooling becomes a major headache.

So, at both ends of the spectrum you need to worry about power efficiency. Not so much in the mid range (not so fast desktop GPUs). There money comes first.

Of course, the inter-core and memory communication differentiates frugal mobile and fast high end GPUs, but you still want to have as much oomph per watt as possible.

 

[MauiPa]

In other words, Qualcomm is concerned that nvidia will charge a licensing fee for a lot of patents that they would not be using if the simply bought rights to license ARM. Strangely that sounds familiar

 

[wilhoitm]

Yeah, but afaik these two companies don’t need the approval/permission of the EU to make a deal right? As I understand it, an American company would be buying a Japanese company. China is a market as big if not bigger than Europe. Will these two companies need permission from China as well?

Huawei is opposing the deal also!

 

[wilhoitm]

NVIDIA is threatened by Apple's M1 processor which is faster than a lot of NVIDA's GPUs!

NVIDIA buying ARM must be stopped at all costs! Apple's M1 processor almost makes NVIDA irrelevant!

 

[Peepo]

Can't wait to see some RTX 30 series GPUs in Macs...

Can't wait to seem some RTX 30 series GPUs anywhere!!!

 

[Peepo]

I bet Apple is working on (or already has) their own instruction set so they can 100% move away from ARM. It would most likely just have have one instruction - analogous to one mouse button.