Not sure where this should live but an article popped up on notebookcheck about rumours of Intel AMD merger https://www.notebookcheck.net/Unite...al-merger-between-Intel-and-AMD.912749.0.html To me the possibility seems crazy with the size of Intel plus antitrust and competition. Any thoughts or just rumour mill on a slow news day?
Zero chance of that happening - it'd leave you with a single company dominating 90-some percent of the x86 market. Better odds on Qualcomm ($184bn market cap) buying Intel ($100bn), or at least parts thereof.
From what I've read elsewhere, both Apple and Samsung are also sniffing around Intel. Current protectionism suggests that Apple would be the US government's preferred bidder, should the rumours turn out to be true. https://www.techpowerup.com/328355/apple-and-samsung-in-the-fray-to-acquire-intel-rumor
I just can’t ever see the possibility of Intel being bought out tbh, even if it is by a Left-Pondian Beacon Of Brilliance like Apple and not a Filthy Foreign Devil like Samsung. Even if I were to try and speculate realistically, the value has surely got to be in its foundries. You can’t underestimate the value or importance of the x86 ISA, but surely someone like Apple would really rather prefer it if the success of their fancy M-series CPUs didn't come with the downside of a big chute funnelling money to TSMC... I don't know, maybe the industry needs a big upset like Intel collapsing at the side of the road and being picked apart by vultures... (Sorry, that was a rather grim way to put it...) Maybe it might force some real competition or innovation instead of the slow drip-feed we've had with x86/x86-64. Maybe we need ARM to make a big push into the desktop (from someone other than Qualcomm, please), or a truly performant & competitive* RISC-V design to hit the market... Oh sure there have been massive improvements in x86/x86-64 IPC, power efficiency, PCIe & USB bandwidth etc... But there's been very little that has made me sit up and think "holy balls, that's impressive" - which is more or less the reaction I had to Apple's M-series CPUs, especially after I started using one for work... And I know, I know, the M-series chips are, in essence, just "faster, less power, better bandwidth, etc"... But it's the degree to which those improvements were made that blew me away. I really wouldn't want to put a bet on my 117W i5-12400F beating my work laptop's ~40W M2 Pro in pure CPU workloads... (Of course my 4070S GPU would utterly humiliate anything that even the beefiest Apple Silicon has to offer...) But to get Apple Silicon you have to pay the Apple Tax, and with prices for the latest Mac Mini (complete with innovative "we put the power button on the bottom!!! lol!!!" design features) starting at £600 for something with a pitiful 16GB RAM & 256GB storage... Yeah... you can get right tae **** with that, pal... *Definition of "performant" and/or "competitive" is entirely at reader's discretion.
Funny you should say that: guess where Intel's leading-edge stuff is made? Yup, TSMC: Every single Arrow Lake and Lunar Lake part is fabbed at TSMC then packaged by Intel. They're targeting Panther Lake as 70% Intel 30% TSMC, and as much of Nova Lake in-house as it can manage. ...but "targeting" and "will achieve" are, of course, very different things, and Intel's fabs have been *struggling* of late.
One has to wonder how much AMDs absence in the market led to intel's current state. Sit on your dominance, only do as little as they could get away with thinking they could turn it up when needed. Then they hit a brick wall with 10nm, can't ramp up as they thought they would and trounced by AMD on desktop and hedt, AMD makes inroads in the server market, Nvidia show up with ridiculous compute power and all of a sudden intel are where they are now, building fab all over the place that could well end up like global foundries, picking up the crumbs when the latest greatest node isn't required or financially viable.
Yes, I did see something earlier about their most recent parts being made by TSMC. I haven't really kept up with the latest in CPU news to be quite honest, it's all a bit boring... It would have no doubt been a sting in the tail, but I suspect that Intel's Really Big Numbers come from the same place that everyone else's does these days: the data centre. There had long been speculation and rumour that Apple were planning their own desktop/laptop oriented ARM chip. They've been in the ARM chip-making business since the days of the A4 iPad/iPhone chip (fabbed by Samsung at the time), and, Darwin, the core underlying component of their operating systems (which, as I understand it, is not just the kernel), has been running on ARM hardware for a long time in the form of iPadOS, iOS, watchOS, etc. They also have prior form in doing a complete switchover, with the switch from PowerPC to Intel way back in t' day.
The really interesting one is the tech comes from ASML, intel is a fairly major investor in them, but TSMC were able to utilise ASML's tech better than Intel repeatedly! Just as AMD have utilised the manufacturing better than Intel. But Intel going away would be terrible. It'd hand AMD the monopoly for at least the next few years, especially with the ARM vs Qualcomm mess that's brewing, and Apple will hoard their tech to themselves! Maybe Nvidia would move into the segment, they're familiar with ARM chips from their Jetson platform, which I'm really familiar with, but they're far from cutting edge AMD processors. Although, don't discredit Intel yet, they're still HUGE in server, networking and cellular! They're looking to sell off some under performing sections of their business. Looking into the numbers, the bulk of the loss were deferred tax losses for 3 years ($9.9 billion), being 'over staffed' ($2.2 billion), and Mobileye driverless cars (~$2.9 billion), so thats $15 billion bang on. Now assuming the staffing cuts stem most of the tax losses, and they just bin off Mobileye (literally, driverless cars are a dead end with current road infrastructure IMO, I've worked in it and worked with key researchers in it, there needs to be social overhaul for it to happen), then they're maybe at £1 billion in the negative, which is fairly ok. They're super strong in datacentre and networking. I think it's being blown out of proportion by the media in a huge way. Don't fall for clickbait. Alternative direction, Intel pulls out of the consumer CPU business... All numbers from here: https://www.theregister.com/2024/11/01/intel_q3_2024/
They always have done. Since the days when Jobs came back and kinda “re-launched” Apple, they’ve always kept a tight rein on their software & hardware. Even though Darwin/XNU form the core of all Apple OS’s and are technically open source, a bare kernel and OS core does not make for a functional modern operating system. All the stuff you actually see and use is all proprietary, even if it’s built on open source foundations. You ain’t getting macOS unless you’re on Apple hardware; “hackintoshing” will soon drop off a cliff when they drop support for x86 machines, which is widely expected next year. And you ain’t getting Apple Silicon unless you pay Apple prices for an Apple computer. See: most basic spec Mac Mini starting at £600 with insultingly low RAM & storage (and yet another brilliant design innovation by slapping the power button on the bottom of the machine). Definitely. The old adage “no one ever got fired for buying IBM” largely became “no one ever got fired for buying Intel”. AMD and ARM outfits are making gains in the data centre, but AFAIU it’s still Intel’s market to lose. Even all the fancy (and highly lucrative) “AI” boxes from Nvidia still need a CPU nestled somewhere in there to run the show. Yeah, I think if anything, this is probably the most likely “upset” we’d see. But I still doubt that it’s a realistic prospect TBH, any big silicon producer will still have a need to do something with the parts that don’t make the grade for the higher-end SKUs. As much as I might want to see a big upset and shake-up in consumer CPUs, I just don’t see Intel leaving that market.
I think that's a little unfair, albeit maybe only just, and I say that as someone who bought a MacBook once and never again. Apple acquired CUPS in 2007 and continued to make it available under the same open licence; it makes WebKit, its in-house browser engine, available under a mix of LGPLv2.1 and the highly-permissive BSD two-clause licences (though it doesn't have much of a choice in the former, as WebKit's a fork of KHTML and KJS); and, yes, there's Darwin. On the hardware side, Apple was literally the originator of the open standard IEEE 1394, or FireWire, though it was developed in partnership with others - and, funnily enough, Sony holds nearly double the number of applicable patents. Thunderbolt? Joint venture between Apple and Intel. Mini DisplayPort? Apple, made available under a royalty-free licence - albeit one it can cancel if the licensee tries to bring a suit for patent infringement against the company. There's more, but that's just off the top of my head.
Thinking back to the AMD/ATI merger and all that followed anyone buying Intel is potentially strapping a millstone around their neck. Not only do they have they to sort the problems Intel are having ATM, they need to catch up to AMD on the CPU side, do something on the GPU side (arguably a market that is exploding with AI), do something with FABs never mind the rest of the business AMD are struggling to compete on both CPU and GPU and already spun their Fabs out into GloFo. Nvidia dabbled with ARM processors but arguably being locked out of the CPU market they drove the GPU side with CUDA. I do wonder is Intel actually too big to fail? Apple use ARM, Snapdragon on windows has put up a good showing. AMD is still fighting the X86 flag and RISC V is waiting in the wings. The CPU market feels healthy. Maybe several companies picking the bones or breaking up Intel might be an outcome. On the surface numbers look bad but I don't think things are dire on the CPU side and even if they are circling the drain it isn't the end of the world. I suspect they can continue a bit longer and recover. Losing the Fabs might be a concern to certain parties but AMD Apple and Nvidia have shown being fabless isn't fatal. Maybe the government will want US based manufacturing so selling it for nationalisation might be an option. I really do want Arc and Battlemage to succeed. Nvidia are running rampant and that doesn't look like changing. I think there is an element of hardware but the biggest issue for Intel GPUs seems to be software.
As I said before, I don't understand your position of them failing at all. The underlying finances don't support it at all. Intel is doing fine, their consumer division is contracting, but is still huge, and most of the losses seem to be from Altera and Mobileye. As I said, $15 billion of the losses is non-recurring, so where are they failing exactly? My work just spent a huge amount of money on systems (we got the first H100's in the country), all of it is Intel CPU's and networking. They are doing fine and aren't going anywhere.
As an aside, the capitalist mentality that companies need to have forever increasing profits is not only stupid, but actually insane. Unless you're a growth phase start up, or you happen to have a good few years like Nvidia, it'll be ups and downs. There is no infinite money! It should be totally fine for companies to have a contraction year, it should be expected!
I agree that Intel ain't going to curl up and die in the immediate future, but I dunno about "doing fine": AMD just overtook it in data centre revenue, for the first time in either company's history: ...oh, that white line? The one that's higher than both Intel and AMD? Yeah, that's Nvidia's revenue... from networking products alone. Doesn't include any of Nvidia's actual data centre compute products like accelerators. Surprised me, too.
They bought mellinox or someone didn't they? Its what enables all of their 400Gbps comms they use on their rack scale systems. And thats why its so high, each one of those mega nvidia systems needs Nvidia networking.
Mellanox. And those insane speeds are why Nvidia bought them . Who wants to keep giving money to another company for parts you need?! For a trifling $7bn up-front investment you can simply buy the company that makes the parts you need - now you can stop giving the middle-man money
And more importantly, deny it to anyone else. I've written mergers and acquisition reports before, and denial to competitors is just as strong as utility to yourself...
