Intel frees its Foundry biz – and that's just one of many major shake-ups today Intel will spin out its Foundry division as an independent subsidiary with its own board, in the hopes of bringing in new sources of capital for the ailing business unit. The decision, announced in a Monday letter penned by CEO Pat Gelsinger, comes just months after Chipzilla made the Foundry division a separate line item on …
For weeks? n his book Chip War (written 2021, published 2022) Chris Miller predicts this move and explains why. He has some less than complementary things to say about Gelsingers two immediate predecessors. I think its a fair summary that he blames these two for Intel's current problems which are really no different to the problems back then. The big difference today is that capital is expensive which is a problem for a capital intensive industry.
Charlie Clark,
The chip subsidies aren't about jobs. They're about market access.
Should Xi Xinping decide to go the full Putin in 5 or so years' time in Taiwan - then global markets are going to be in a total mess. Hence the policy is to get some fabs built. Of course, whether any of the politicians have considered the whole supply-chain - and whether those fabs can work without all the other inter-connected bits of the industry is another question.
Similarly there's been EU subsidies on offer to get rare earths mined and processed in Europe. Not that it needs a mine, you just need to find the most appropriate current mines, with the best levels of rare earths present and then change how they're processed.
It's pretty clear that making our economies so dependent on Russia and China was a bad move. Actually extricating from Russia was a lot less painful than it might have been - but China is a hell of a lot mroe important in the globalised world economy than Russia was. And though it's going to make everyone in the world poorer, the alternative is to leave our economies subject to the whims of the Chinese government - and more specifically Xi Xinping. He doesn't even have the people of China's best interests at heart, let alone ours.
"...Gelsinger, who has taken considerable flak for Foundry's mounting losses in recent quarters, claims the decision to establish the division as a subsidiary will offer multiple benefits.
"It provides our external Foundry customers and suppliers with clearer separation and independence from the rest of Intel," he wrote. "It also gives us future flexibility to evaluate independent sources of funding and optimize the capital structure of each business to maximize growth."
Typical management, P-R, and legalese horse-shit.
Intel is now irrelevant.
It's PR horseshit alright, with insane investors running the financial engineering asylum -- I don't like it at all. Intel will be alright though imo, as long as it's not been irreversibly straitjacketed into this dolorous posture.
Intel's cashflow is just getting worse. The company has some extremely talented people but several years of very poor decision making could cost it its future, especially given current valuations.
Is the fab side of the business massively unprofitable simply because of how Intel reports profits? If it's making it's own chips and counting the profits in the chips column not the fabs, then the fab's going to appear costly, once gone solo the chips team's going to look worse as they're now paying for them...
I'll opine that it's not lack of sales on AI chips that are Intel's problem, it's concentrating on "AI" in the first place. nVidia's "AI" is really just a leverage of their GPU design business, a business that has very little to almost no major competition and has for, what now, a good decade? And what does Intel do? Almost completely ignore the GPU space for *years* but now Intel is trying to play a power move on "AI", which nVidia's GPU chips have been playing in since the beginning.
So:
- go after an established, almost single-monopoly market (GPU) and be able to leverage the technology into the "AI" sector, or
- fight for relevance in a single-focus emerging market with an unknown future ("AI" exclusively).
"Let's go after AI! After all, it's this year's hype!"
smh
So, as usual, you can't fix the stupid.
The troubles extended back to itanium: a premium product developed to cash in on wealthy industrial customers (who were unexpectedly presented with a much cheaper yet equally as powerful 64-bit concept). Oh, the humiliation of Intel forced to adopt AMD64!
Intel was like the Evil Witch of West shrinking to a $90 billion emaciated shadow while AMD has swelled to a shaky $244 billion valuation. Intel isn’t exactly conquered, but battered and belittled.
Meanwhile nVidia is a $2,800 billion dollar target…
I think we're broadly in agreement.
Intel rountinely ignored other avenues in the hope that x86 on a die could and would do everything. So, while AMD did buy GPU expertise, and developed a client instruction set like CUDA for it, it was still competing for business with Intel in data centres, but is now in a much better position, largely because it realised there was life beyond x86.
Intel kept wanting to force x86 + on everyone and, you have to admit, this was remarkably successful even given the mobile fuckup until about 2020, when it was generally considered that GPUs were mainly for games and video decoding and encoding. It had a good track record in video, but was happy to be an also ran in games. So that, if like Google and Amazon, you wanted cheap and cheerful (integer only, low power, small, etc.), er, chips for ML or embedded, Intel was not an option.
That said, Intel is now probably undervalued and both AMD and nVidia overvalued. It would be a pity to see all that process engineering knowledge lost. Maybe the US government will coax AMD into doing the decent thing because the one thing the world should be looking to avoid is reducing choice when it comes to foundries.
Itanium and AMD64 would disagree with your assessment of Intel as betting everything on x86 and AMD realising there is life beyond the 32-bit chip they extended to 64 bits.
I have no idea what your last sentence is supposed to mean given AMD got rid of their foundries years ago and are not about to start building new ones given the success they have had without them.
Seeing that you're still here, debating this, on Thursday, here's a related question that might also be interesting:
What do you all make of this HPCWire article that suggests (from an Intel "Monday letter to employees") that Falcon Shores (intel GPU) is on the chopping block? Reasonable? Grate plan? Otter nonsense? Udder BS?
No, the argument that Gelsinger has made is precisely the opposite: he says that Foundry has been unprofitable for nearly two decades. And that it’s CPUs and the profit from them which has been hiding this.
He also says that the vertical integration has been massively distorting, and caused them to make stupid decisions and operate the foundry inefficiently. Specifically, that when the CPU guys want to make a new SKU or try a new proof-of-concept, they just tell the foundry “go do it”, and it gets done. But what they don’t see, or cost, is the massive disruption of taking things off and on the line. If they really appreciated that cost, he reckons they wouldn’t ask for / wouldn’t get approval for, many of the projects that Intel CPU run.
I expect he is correct in this analysis, that is probably the truth of the matter. The question is; does he have the cure? Or is the upheaval making Intel un-tenable? Is there a profitable *fabless* Intel on the other side of this, because that’s the only way I see this going?
” Gelsinger noted that Intel would move forward with its advanced packaging facility in Malaysia but would "align" the startup of the facility with market conditions.”
Srill using cheap Far East Asia. Labour and long supply chains means most of the CHIPS funding intentions are largely bollocks…. but that’s been the case forever as most output will just be shipped to the Faf East where end product assembly is - regardless of Laptops, servers, consoles, consumer electronics, auto components etc.
Having spun off the foundries, then selling the foundries company to a south korean company, that the newly renamed KC foundries ends up making a shed load of profit for its owner
Meanwhile, back in the US , Intel shareholders sue the company for loosing them a shedload of money....
Based on what's happened in Ireland, a lot of middle management (which is mostly not a huge loss, from the feedback of some who are remaining) but indeed there are a fair wedge of good and experienced engineers and tool owners who are taking the opportunity.
From what I've heard from on-site colleagues the place is currently like a ghost town (empty and the fab not running much).
Those who remain are varying between nervous, depressed or just utterly fed up.
The other problem of course is the knock-on effect for suppliers and vendors. On the one hand they need us more due to that loss of experience, but on the other they haven't got the cash available to fund the contacts etc, so our lads and lasses are on dodgy ground too (or on flights to Germany or further afield).
If you're, for example, AMD, would you trust the separation that's supposedly now here between Intel foundry and Intel design? I wouldn't.
This is useless unless it's a full, Global Foundries-esque spinoff, but Gelsinger wants to MAKE chips... I don't think he has it in him to really do what's required.
I feel to massively underestimate the capabilities of a corporation with revenues in the tens of billions. Even allowing for inefficiencies inherent in any org, I read comments assuming they did not do their job by skipping strategic analysis of costs and market trends. They did not.
In my layman’s mind Intel troubles can be traced back to the 14nm process. Continuous delays were never recovered in follow up nodes. What caused those delays is anyone’s guess. Astonishingly complicated process steep of trade secrets.
AMD piggyback on Apple’s economies of scale achieved with TSMC fab of leading edge mobile SoCs, and the intro of a competitive design, came along with Apple’s decision to abandon ship, and at the same time find a competitive position independent of nVidia accelerators, and AWS long term plan of promoting it’s own design, coming back now as a fab only customer, were in the nature of things.
Something I would like to read more about is the environmental cost of newer nodes to power smartphones. And then the rest.
My two cents.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
