ARM was a British company traded on the London stock exchange before it was purchased by Softbank. It is still based in Cambridge England. I made a good profit on my shares when they were purchased by Softbank
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
SoftBank Nears Deal to Sell Arm to Nvidia [Updated]
- Thread starter MacRumors
- Start date
- Sort by reaction score
Not unless there is some clause in their license that prevents it. I don’t think anyone’s seen the exact license they have with Apple.
1. It was a good price then. ( Some people will disagree ) ARM was sold to Softbank for ~100x+ its net earning / Profits. 100x P/E. Even ARM themselves dont foresee they could double their profits. ( Even if they could somehow double their revenue ). You will likely read some people argue AMD also has a P/E of 100, and blah blah... But AMD currently has less than 10% of Server Market ( They say they do but the figures dont quite align in my book ), and 10% in the "total" PC market. i.e They have lots of headrooms in terms of market size and revenue. Compared to ARM. They already have 100% of the Smartphone Market. And their business model dont scale well with those units. Perpetual license meant it doesn't matter if Apple sell 100M or 1M Chips. Although the market is now trending back towards licensing per unit using ARM design. Because not everyone can get an implementation that is fast and cheap. ARM currently has the upper hand here and will likely continue for the next few years.
2. Softbank ( specifically Softbank the VC, not the mobile phone network or the company ) invested in lots of startups. With money form different places. Most widely known are from Saudi. ( But not only from Saudi, so to say they are a Saudi Fund would be incorrect ). They made a investment into WeWork and somehow, didn't work out the details correctly. So WeWork scam Softbank for billions ( or may be tens of billions ), and now Softbank runs at a huge loss, with the current economy they likely need cash to fill up some holes in different places, and ARM is their most valuable asset.
Before my coffee... I spent whole 5 seconds thinking the corporate wars known from sci-fi are finally here, since I just read “Bank sells arms to Nvidia” 🙃
I'm not sure if you have noticed but it seems the staff of this site get some form of a weekend, and so, fewer articles are posted on the weekend. Sometimes this leads to a "24 hour delay".
You are not very well informed, nVidia already does this, nVidia does in fact make CPU's, or rather they make SoC's that include ARM based CPU's and nVidia graphics on a single chip. The nVidia Tegra SoC's have been around since 2008 and they are used by the Nintendo Switch, the nVidia Shield, Tesla Cars and in a bunch of machine learning applications by various customers.
What will be interesting is that nVidia will gain a lot of smart chip design engineers as part of the deal, experts in embedded chip design, as well as in mobile devices, I expect to see nVidia start producing mobile chipsets to rival the Qualcom Snapdragon line, with better graphics and inferencing, maybe they will even launch their own android phones (tablets already exist) focused around gaming and streaming as part of their Shield Brand.
that’s not going to happen.
we’re only a few years away from only having apple gpus
[automerge]1600086356[/automerge]
actually OS X has nvidia drivers. macOS not so much
I think it has more to do with the faulty chips NVidia provided to all laptop manufacturers and then told them to eat the cost when it came time for repairs. This was not the last time NVidia did this but it was enough for Apple to drop them.
Again, because going forward it does not matter. Do you think Apple GPUs need Nvidia drivers?
I’m not sure why people keep trying to turn this into a dramatic after-school special, throwing words like “hate” and “angry” around... Apple has what they need from ARM already, and there isn’t any risk to that. Thinking Apple cares about this deal (or Nvidia in general) is like thinking Tesla would care what ExxonMobil and BP do. It just isn’t relevant.
Any one can explain why would Nvidia would buy ARM for 20 years of revenue's worth?($2B/year)?
Nvidia and AMD and fighting to death on GPU meanwhile why not compete on other processors like RISC, PC CPU, smartphones...on the other side Intel just makes CPUs and after their long history no competition on GPU or smartphone CPUs.
Nvidia and AMD and fighting to death on GPU meanwhile why not compete on other processors like RISC, PC CPU, smartphones...on the other side Intel just makes CPUs and after their long history no competition on GPU or smartphone CPUs.
...that's not what @Ries meant by "Switzerland" - the point was that ARM was "neutral" and didn't have a vested interest in any particular chip vendor : they simply licensed designs to anybody willing to pay (whether it was NVIDIA Tegra, Qualcomm Snapdragon, Apple Silicon etc...) and their business model was "the more the merrier".
The question is now, why would NVIDIA license future versions of ARM designs to other chipmakers who are going to compete with its own Tegra chips? The same issue would apply if Apple had bought ARM - could you see them licensing Apple Silicon designs to Samsung (or NVIDIA for that matter)...?
The answer to that is that ARM's business model has probably played a huge part in ARMs success, and changing it would be a "courageous" decision... but only if your CEO actually gives a wet slap about the company's long term future, and isn't just counting on wringing enough cash out of the corpse to make a profit on the purchase price while trousering the IP. It would be nice to think that that's why Apple didn't buy it (but it's more likely that, as the highest profile user of the ARM ISA, their chances of getting it past the competition authorities would be zip - whereas NVIDIA are just a.n.other licensee and will probably only raise token opposition.
It may be that ARM has done its job, especially if Apple and others with existing ISA licenses are free to develop their own versions. The ARM-based mobile market has broken the back of the Wintel monoculture and created an environment in which alternatives such as RISC-V actually have a snowball's chance in hell of succeeding, and created a large, diverse talent base and production capacity for that which is not Intel/NVIDIA/AMD. We're really coming to the point where ISAs should only be relevant to specific parts of the OS kernel, a few critical libraries and JIT compilers, and the advantages of fine-tuning the CPU to the type of device should outweigh the disadvantages of having to re-write those. In fact, that's not even a new idea (its been the Unix philosophy all along) - it is only the never-to-be-sufficiently-cursed legacy of the IBM PC (a mundane, backward-thinking me-too system that nobody would have looked twice at if it hadn't had those three magic letters on the front) that made x86 king. Android, "modern" Windows and the whole Java world are already virtual machine based. Apple has bitcode technology that can generate tailored binaries for the target processor at the app store stage - not a magic bullet for running x86 on ARM but quite capable of dealing with minor variations on a RISC ISA.
I think Apple will be more than happy to "take it from here" with instruction set design. If they get the x86-ARM transition right - in particular by making sure that special-purpose instructions for vector processing, codecs, multithreading etc. are handled by OS frameworks rather than application code - it should be the last time that changing ISA for MacOS is a big deal.
WHY? Nvidia hardware has historically been pitiful on Macs where Apple has had numerous recall due to overheating and failed gpus. We don't need nvidia drivers or hardware.
$40B for ARM is overpaying by miles. In 2019 ARM probably made around $2B in revenues. Profitability was much less than that.
Not to give nVidia the finger, but to make the best chips they can make. Apple can evolve Apple Silicon into whatever they want, if it no longer resembles "official" ARM anymore (or eventually breaks compatibility, etc), that makes absolutely zero difference to Apple.
Because modern hypercapitalist finance doesn't give a rat's backside about long-term prospects and growth - its about being the one holding the parcel (or the one not standing up) when the music stops. It's about being the casino that owns the roulette wheel and rakes in the cash while the suckers make money. It's about being the one that makes money when shares go up and makes money when they go down.
Maybe they can pump it up and sell it for $50bn in a few years' time (like Softbank did). Maybe they can run it into the ground by taking excessive profits and not re-investing it, then keep or sell off the valuable IP. Maybe their share price will temporarily shoot up and all the execs can cash in their shares... or maybe it will dip down and all the short sellers will make cash.
Always remember: the difference between high finance and a pyramid scheme is that - at some point in history - somebody (probably the tax people) successfully lobbied for a law against pyramid schemes.
I just don't like nVidia's business ethics, but they tend to make the better hardware. AMD is a solid and (excluding some recent driver issues) reliable 2nd place though.
If anyone wants the image without the watermark...Regardless, Apple declined to buy ARM. They don't seem to think it will affect them.
Apple is reportedly an Architectural license of Arm
A long look at how ARM licenses chips
People routinely ask SemiAccurate about ARM CPUs which is an odd question because ARM makes absolutely no silicon.semiaccurate.com
It does make SoftBank the largest shareholder of Nvidia. That’s still only 10%, but they very likely get a seat at the table.
if we talk about why Apple didn't buy ARM, reasons are mentoined here and second it will be not financialy good decision since Apple have ARM instruction license and TSMC as partnership.
In other hand i see big risk for Nvidia to invest a such money buying ARM. In other hand this will be a big push for wide ARM migration since x86 are starting to juice itself out. Let's see what Apple will show us tomorrow at conference and what they are prepared for reviel.
Edit: If Apple decided to swith to arm, there must be a really big reason for speed, productivity and power consumtion. And of course nVidia looking the same way.
In other hand i see big risk for Nvidia to invest a such money buying ARM. In other hand this will be a big push for wide ARM migration since x86 are starting to juice itself out. Let's see what Apple will show us tomorrow at conference and what they are prepared for reviel.
Edit: If Apple decided to swith to arm, there must be a really big reason for speed, productivity and power consumtion. And of course nVidia looking the same way.
Apple could make the move , not necessarily that they have to make the move. If Nvidia manages to destabilize ARM as an open IP solution then Apple would need to adjust to the circumstances. Over the course of a decade in technology not all of the decisions are going to be 100% up to Apple. Times change , strategies can change.
If the open ARM IP space was healthy in 10 years then Apple may not have to do much.
A substantive large 'fork' by Apple of the instruction set over ten years isn't all that pragmatically different than moving to a new instruction set. Without constraints to be part of the group, Apple is likely to drift into their own 'world'. ( not tracking open 3D graphics OpenGL/Vulkan --> Metal not tracking open compute OpenCL --> Metal ) . Where Apple goes after abandon the future mainstream updates for ARM wouldn't be maximally optimized for backward, legacy compatibility over the long term. That is not what Apple does.
they have a legal perpetual licence. Apple were one of the original owners / Partners of Arm. The next major iterationOf Arm won’t arrive for years and Apple have the perpetual licence.
Perhaps the "Switzerland" above should have been in quote. It is not really referring the Switzerland as a location (i.e., location of the company). But as 'Switzerland' the canonical neutral party ( not taking sides). ARM was neurtral because there were not a company that sold silicon products. So giving money to ARM for chip development was not giving money to a competitor to construct something to your own product. Far more closer to ARM being 'outsourced' R&D expensive were several companies all contribute to the development and all can get access to that shared intellectual property developed.
Nvidia is an implementer of ARM. Some folks think they just only make GPU chips. That isn't true.
- Archives Page 1 | NVIDIA Blog
shield.nvidia.com
Buy the Latest Jetson Products
Get Jetson AGX Xavier, Jetson TX2, and more for your development work.
developer.nvidia.com
Nvidia Jetson - Wikipedia
DOCA Software Framework
Accelerate application development for the NVIDIA BlueField DPU.
www.mellanox.com
Several years ago Nvidia was trying to compete with Qualcomm , Samsung ,etc in getting placement in phone ( Android ) products and largely failed to get any substantive traction. If they own ARM and have the inside track to new tech ... are they still going to fail to get maximal traction?
A very significant way of making back the $40 billion they just laid out would be to take product market share away from other ARM licence holders is by replacing their products with higher profit margin Nvidia products.
Nvidia isn't a "neutral" party in the ARM 'open license' model . When ARM implementer tell the ARM division that they would like X, Y , and Z in future processors they will also be telling their competitors that would X, Y , and Z. If they need new feature "A" that is a design trade-off to new feature "B" and Nvidia's products highly need feature "B" is t at going to turn out to be a good outcome for that implementer ? Probably not.
Nvidia is probably going to talk up that they aren't going to change the ARM business model at all. If they said anything else they probably wouldn't be able to buy it. So the creditibilty of that is quite marginal.
ARM makes about $200-400M per quarter in revenue. ( $400M is high). Lets say that things are looking up for the company (more Mac CPUs 12-18M more units sold) and that high side normal for all 4 quarters in the future. So about $1,6B per year. If they devoted every last penny of that revenue to pay off the $40B purchase price, it would take 25 years to pay off that amount.
The ARM business model can't pragmatically pay off that amount. Nvidia silicon products with large margins might. That is the "definitely not Switzerland" issue.
Last edited: