Hacker News new | past | comments | ask | show | jobs | submit login
Apple signs new agreement with Arm that goes past 2040 (theverge.com)
74 points by mikece on Sept 6, 2023 | hide | past | favorite | 36 comments



That's not quite accurate based on the filings. Rather it should read that ARM has existing agreements with a number of companies, Amazon, Apple, Samsung, even Nvidia.

The ARM architecture is so ubiquitous that everyone wants a piece. I kind of think of ARM as a DMZ: The business model of licensing IP to a broad range of clients means that a lot of companies have a vested interest in ARM remaining a neutral party. We've seen with Nvidia and others that ARM isn't a clean acquisition target, so now everyone gets a place at the table, and thus license security for the long term. It's a bit similar to OpenAI, and how I think that in today's market, no regulator would allow a major tech company to acquire them.


I'd never heard of the term DMZ used in a computing context. For those like me who didn't know, it does mean De-Militarised Zone


It has been used in the context of firewalls forever.

https://en.m.wikipedia.org/wiki/DMZ_(computing)


And by "forever" it is meant ~1994; see Cheswick and Bellovin (§3.1):

> Up to this point, we have used the words “firewall” and “gateway” rather casually. We will now be more precise. A firewall, in general, consists of several different components (Figure 3.1). The “filters” (sometimes called “screens”) block transmission of certain classes of traffic. A gateway is a machine or a set of machines that provides relay services to compensate for the effects of the filter. The network inhabited by the gateway is often called the demilitarized zone (DMZ). A gateway in the DMZ is sometimes assisted by an internal gateway. Typically, the two gateways will have more open communication through the inside filter than the outside gateway has to other internal hosts. Either filter, or for that matter the gateway itself, may be omitted; the details will vary from firewall to firewall. In general, the outside filter can be used to protect the gateway from attack, while the inside filter is used to guard against the consequences of a compromised gateway. Either or both filters can protect the internal network from assaults. An exposed gateway machine is often called a bastion host.

* https://en.wikipedia.org/wiki/Firewalls_and_Internet_Securit...

* https://archive.org/details/firewallsinterne00ches

Ingham and Forrest wrote a good history of firewalls:

* https://www.cs.unm.edu/~treport/tr/02-12/firewall.pdf


Are there any historical examples of such a company - ubiquitous yet still a small fish, slightly troubled corporate history, but too important to be acquired by someone else?


No doubt a publicity move agreed between Apple and ARM, ahead of ARM IPO.


The lack of detail is troubling, will new prospective shareholders be happy about pricing given the sheer time length.

Was Apple trying to get this in specifically before IPO


I believe Apple was a founding investor in ARM and has a perpetual license to much (all?) of ARM’s IP. I’m not sure ARM’s prospective shareholders should ever have expected much revenue from Apple.


Apple could still potentially pay to get some say in ARM'S design decisions or be able to direct part of their resources to their needs.


Fair enough, but I don't expect ARM investors are expecting that to be substantial revenue.


Anything that sounds like Apple making a 15+ year commitment to the ARM platform is going to benefit ARM's IPO massively, so the real question is what Apple got from this.


Stability, legal leverage. It’s likely been hashed out in response to nVidia’s attempt to buy ARM.

https://www.gq.com/story/tim-cook-global-creativity-awards-c...

Tim Cook: we are very much focused on the long term.

If ARM ends up being acquired, Apple can unleash the lawyers to demand adherence to their 40 year deal.


> If ARM ends up being acquired, Apple can unleash the lawyers to demand adherence to their 40 year deal.

Hate to break it to you but it’s not 2003 anymore. It’s barely a 20 year deal.


Whoops; mental math fail at bedtime.

Not sure that undermines the premise it’s about locking down IP access on Apple’s term’s. An ARM buyer or even just new management will have a harder time coming in and jacking up prices on Apple before 2040.


> Was Apple trying to get this in specifically before IPO

Only, I'm sure, to the degree it gave them some advantage in negotiating with ARM.

ARM's IPO is troubled, for reasons I haven't been following, but I am sure this helps them.


> But given how much Apple relies on Arm’s designs for the custom chips powering its most popular and profitable products, like the iPhone, Mac, iPad, and Apple Watch, in addition to the upcoming Vision Pro, it signals that Apple plans to continue utilizing Arm’s technology platform for a very long time to come.

But Apple doesn't rely on Arm designs for anything? I think it's just sloppy writing, but I'm not sure I'd trust the analysis that follows weak reporting.


>But Apple doesn't rely on Arm designs for anything?

Of course they do. ARM started is still offering a lot of embedded design.


Apple relies on the ARM ISA so this new agreement may cover ARMv9 or something. If you want to argue that an ISA is not a "design" journalists don't understand the difference.


As I understand it Apple already has a perpetual license as a result of their being a founding member or something?


We simply don't know the specific terms of the license because it's never been made public.


When did Apple sell their shares in Arm?

Arm was cofounded by Apple, Acorn and VLSI.


After the Newton was shut down, 1998, according to this:

https://www.cpushack.com/2010/10/26/how-the-newton-and-arm-s...


Great find! I really dig these circular stories that shape the electronics history. The story behind ARM has always fascinated me. The search for low power with efficiency kind of seems like the holy grail of the time. The Acorn story reminds me a lot of the Po Bronson novel "The First $20 Million Is Always the Hardest: [1] 1. https://www.goodreads.com/book/show/306422.The_First_20_Mill...


They were really out of cash. Jobs convinced Gates to give them $150MM as well (Apple's existence helped MS with antitrust; Apple desperately needed the cash).


As I recall, the dash was less important than the commitment to keep Office on the Mac. (Maybe IE too?)


They really were desperate for cash.


But as Jobs pointed out at a conference where this came up, the perception that the Mac had legs (because Microsoft still was committed to them) was far more important than the money. If people don't have confidence in your platform, all the money in the world doesn't mean squat.


They had around $1.2 billion in cash reserves at the time.


Pity ... RISC-V is so much better perspective ... and free.


Patience. RISC-V is an unstoppable freight train but it isn’t yet up to speed.

RISC-V is not yet ready for the high end needs that Apple has. RISC-V doesn't yet have the single core performance nor the ability to scale to many cores like ARM does -- or at least I haven't seen it.

I think that when RISC-V can do this and is competitive with ARM, Apple could look at it. Apple isn't married to ARM forever, remember it used to be Motorola, then IBM PowerPC, then Intel x86 and now ARM.

Given that Apple just did the transition to ARM, I do not expect they will look around for the ARM successor for a while. But a decade from now it wouldn't be surprising for them to consider RISC-V, at that point its capabilities should rival ARM if not exceed it while also being much more easily customized.


Apple has used RISC-V in some embedded controller implementations recently. Perhaps due to licensing cost or perhaps just to get experience with the chipset.


Not in hardware you can buy, yet. But there's e.g. Ascalon[0], design lead by Wei-han Lien, who previously led M1 at Apple.

0. https://tenstorrent.com/risc-v/


That is amazing stuff. Thanks for mentioning it! Next up I guess is some type of RISC-V chiplet design and then they have all bases covered.


>Next up I guess is some type of RISC-V chiplet design

Aegis CPU: Ascalon-based Chiplet System Architecture

From that same page.


This doesn't make sense. Apple doesn't use publicly available ARM core designs.

What about the ISA would make Apple's core designs impractical or less efficient?


I don't think this deal excludes Apple from using RISC-V.

What it does is secure their ability to still use ARM for a while.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: