Intel is in a very bad place, and they need to admit it

[Prosercunus]

Perhaps, but my 13900k is looking pretty sexy right now. Not going to lie though, those new 3D AMD processors coming out this month are looking pretty sexy as well.

 

[kiwigraeme]

This article is interesting - but without a breakdown of revenue etc then we are kind of blind
Their Foundry side has definitely not been great
The have lost their premium markups - no ones paying $2000 now for a CPU that is only 10% faster - ie those $1000 CPU for the whales of yesteryear - when the $500 one was practically just as good
The x86 is being hit by ARM etc and just capable low power alternatives
Plus the speed of change this decade is amazing.
Add in why update your laptop or PC ?

I imagine the server market pre-AMD was quite lucrative

Even if AMD and Intel want to price fix by mutual understanding ( ie not illegally ) - they have ARM etc to contend with
Nvidia etc want a lot of the new markets in AI , cloud etc

So Intel needs to get good
USA car manufactures post WWII probably had it real good - now it's tough
As a consumer - I hope this happens to all these big guys - Apple , Google, Facebook etc

 

[Mr Majestyk]

Well the worrying thing is we need a strong Intel as a competitor, but this is not guaranteed at all post 2024. AMD would become just like Intel in many ways if they had no competition. There is no incentive to innovate as much when the competition isn't there. Just look at the garbage Intel rolled out with year-on-year IPC uplifts of 1-2% due to how bad Bulldozer was. Now Intel is getting hammered by Apple and maybe, just maybe Qualcomm can start producing compelling ARM SoC. I d believe if they can execute their roadmap Arrow Lake could be a turning point, but boy that's a big if. Bit there there's still the problem of the foundries being able to compete and the money needed to get even close to TSMC and Samsung. Meteor Lake is Intel basically giving us 7nm. Sure there 7nm is probably competitive with current 5nm, but we will see 3nm on refined nodes by then.

Also the article did not mention the cluster fcuk that is the gpu group. Again a strong third player is required urgently in the market, but Intel don't have the resources going forward. Battelmage if it ever appears has been greatly scaled back in scope. And will be up against Blackwell and RDNA4 yet IIRC are targeting 3080 levels of performance. Hopefully it's 4080.

 

[Vanderlinde]

Intel has bin lazy for many many years. And now its time to play catch-up!

 

[quadibloc]

I think there's one basic omission from the article.
Yes, Microsoft does have a version of Windows that will run on ARM. But much existing Windows software won't run on it, even if some will because it's the managed code stuff for the new interface.
So the big reason for Intel's power and dominance remains fully intact: Intel owns the x86 architecture, and the x86 architecture is what you need to run most software.
How can Intel leverage that without running afoul of antitrust regulators? it's not as if the world needs to switch (or, rather, that Microsoft needs to switch Windows) from x86-64 to the Itanium because the Itanium is twice as efficient and powerful.
The licensing arrangement between Intel and AMD apparently means that Intel can't just come up with, say, a new set of vector extensions for the architecture that becomes the standard and AMD will have trouble licensing.
Like Nvidia, Intel has spent much more than AMD on software support for its products. If Intel was dying, AMD would need to keep Intel alive just about like Microsoft needed to keep Apple alive.
I think the general principle here is that Intel can't keep milking the cash cow; it has to come up with something new to keep the company alive. The problem is, of course, that the desktop computer is taken, it uses x86, and smartphones are taken, they use ARM. And any new niches for processors are looking hard at RISC-V. Companies do have a life cycle; look at the fate of IBM.
But IBM still hasn't given up, and Intel is at a much earlier point in its decline.

 

[yRaz]

Wouldn‘t it be a sweet irony if Intel had to go the same way they forced AMD to go- splitting off their fabs - by their dirty tactics back in the day ?

I think an Intel shrunk down to size to where they will be forced to compete only on merit will be best for anyone. They should never again be in a position where they can dictate what AMD based systems they offer and how they market them.

Why? So AMD can rail us like they did with socket 939?

These companies are not our friends, never forget that

 

[MunkenDrunkey]

The CISC vs RISC argument was put to bed nearly 30 years ago with the release of the P5 pentium. There has not been a full CISC chip since the death of cyrix. Pure RISC hardware on desktop dies with the G series from apple, as they were unable to scale to match the flexibility of the x86 instruction set. ...

The RISC instruction set is also licenseable, they are likely to prevail over CISC because at this point in the technology, there's so much specialization taking place, people are able to design their own SOC. With this new availability, the need for general purpose CPUs is fading. I guess I am not drawing a clear enough distinction between CISC/proprietary vs RISC/open.

 

[Irata]

Why? So AMD can rail us like they did with socket 939?

These companies are not our friends, never forget that

Where does it say that ‚AMD can / should rail us‘.

Intel competing on merit means that they have to offer good products at competitive prices rather than writing out checks to keep selling mediocre / uncompetitive products.
It also means that we should no longer see competing products limited to certain screen / graphics card / … combinations.

I want actual competition and until Intel is cut down to size that won‘t happen as it‘s too easy for them to get back to / stay with their bad habits.

Side note: Am mainly talking about OEM products where the volume is.

 

[Avro Arrow]

I really think that this article is being overly dramatic. Intel isn't in any danger of dying, it's far bigger than both AMD and nVidia COMBINED.

nVidia's Total Assets: $44,180,000,000
AMD's Total Assets: $67,580,000,000
COMBINED ASSETS: $111,760,000,000

Intel's Total Assets: $182,100,000,000

By this article's standards, both AMD and nVidia are both at death's door! Can you say "clickbait"?

 

[moon982]

The RISC instruction set is also licenseable, they are likely to prevail over CISC because at this point in the technology, there's so much specialization taking place, people are able to design their own SOC. With this new availability, the need for general purpose CPUs is fading. I guess I am not drawing a clear enough distinction between CISC/proprietary vs RISC/open.

Well smartphones and tablets may use ARM I don’t think servers, factories, warehouses, schools, hospitals, government buildings and military so on ever use ARM.

I think ARM will mostly be for public for simple stuffs like smartphones, smart watch and tablets and netbooks.

May be some stores and small offices may start to use Apple computers in the future.

 

[passwordistaco]

Well smartphones and tablets may use ARM I don’t think servers, factories, warehouses, schools, hospitals, government buildings and military so on ever use ARM.

I think ARM will mostly be for public for simple stuffs like smartphones, smart watch and tablets and netbooks.

May be some stores and small offices may start to use Apple computers in the future.

Amazon is using custom ARM processors on their servers, and they're not the only one.

Microsoft has a military Hololens contract. I bet those units have ARM chips.

 

[someOtherGuy]

This is totally comparing apples and oranges. Intel's business is primarily CPUs and makes around 70% of computer processors; that is not a mere a niche in the market.

I think the the article is talking about silicon output. Intel could spinoff the foundry business and become fabless, like nVidia or AMD are, but until that happens, if it happens at all, Intel Co. performance is not only tied to CPU and GPU design but also to their production, which is a side of the business that had lagged behind the competition for years now, and the reason to worry for the writer and other stockholders. Then, when you compare the foundry business to the competition you're not talking Intel vs AMD but Intel vs the wildly known TSMC and Samsung but also another LOT of foundries.

 

[KaitouX]

Intel needs to come to the realization that they need to swallow their pride and go with a chiplet design much like AMD. Only then will they be able to more easily hit target dates, reduce product manufacturing complexity, and reduce costs.

Intel announced they are going to be using tiles(what they call their version of chiplets) years ago. Meteor Lake is going to be the first major architecture to use it.
I don't see how tiles are supposed to help hitting target dates, if any thing you are making it harder by adding a extra possible failure point.

 

[trparky]

Intel announced they are going to be using tiles(what they call their version of chiplets) years ago. Meteor Lake is going to be the first major architecture to use it.
I don't see how tiles are supposed to help hitting target dates, if any thing you are making it harder by adding a extra possible failure point.

It's just like how AMD does just about everything with their chiplets.

Need an Epyc chip? Use the chiplets.
Need a Threadripper? Use the chiplets.
Need a desktop processor? Use the chiplets.
Need a mobile processor? Use the chiplets.

Lately all Intel chips have been monolithic dies. They have to have several die types going on at the same time, each for a different job. A whole unique die for a server chip, a whole unique die for desktop chips, and finally a whole unique die for mobile chips. That's a lot of wasted R&D, manufacturing space, and money.

 

[NoLifeDGenerate]

Intel completely deserves to be exactly where they are and worse after all the ridiculous and underhanded **** they did to competitors like AMD over the years.

 

[Irata]

I really think that this article is being overly dramatic. Intel isn't in any danger of dying, it's far bigger than both AMD and nVidia COMBINED.

nVidia's Total Assets: $44,180,000,000
AMD's Total Assets: $67,580,000,000
COMBINED ASSETS: $111,760,000,000

Intel's Total Assets: $182,100,000,000

By this article's standards, both AMD and nVidia are both at death's door! Can you say "clickbait"?

The question is what are those assets ? Looking at the balance sheet, goodwill is listed at $ 27.5 Billion, ‘property, plant and equipment‘ at $ 80.8 Billion.

Not all assets are equally good - some can easily be liquidated, others actually cost money. Inventories are difficult as they can be liquidated but may also lose in value.

 

[Avro Arrow]

The question is what are those assets ? Looking at the balance sheet, goodwill is listed at $ 27.5 Billion, ‘property, plant and equipment‘ at $ 80.8 Billion.

Having assets of that value mean that the company is far from existential danger. That's the point that I was making. Sure, they're getting their butts handed to them in the CPU space but Intel makes so many different things that even if they left the CPU market completely (which would never happen), Intel would still exist.

Not all assets are equally good - some can easily be liquidated, others actually cost money. Inventories are difficult as they can be liquidated but may also lose in value.

Yeah but when you're a cutting-edge tech giant like Intel, most of your corporate assets are extremely valuable and easy to liquidate. Comparing Intel to Goodwill is kinda... strange.

 

[wiyosaya]

I agree with the article. As I see it, the main problem with Intel is their hubris. I once worked for a major company that has all but evaporated. As I saw it, their main problem was their hubris. It was company policy to never go back on a decision even if it turned out to be a bad decision.

All Intel needs to do is continue resting on their hubris - as I see it, it will not matter how much they have in terms of assets. Assets are limited and will, without better business practices, eventually deplete.

 

[hwertz]

They're in a tough position for sure:
Servers -- the traditional cash cow, but with the concern for best performance per watt, and large economies of scale, both Google and Facebook (for two) have gone to using their own custom CPUs. Some other cloud sales have gone to ARM systems too. AMD has cut into the sales of many-core x86-64 CPUs so Intel has had to cut the quite profitable Xeon prices to be competitive. Overall sales are down too.

Desktop/laptop -- sales are down. Apple's switch to ARM could not have helped a bit. AMD has cut into these sales too. Besides the weak economy, the other factor that seems to be overlooked is the rather dreary specs many new systems have in an effort to keep the costs down. The new system will have a much faster CPU, yes. But it's hard to get excited about buying a new laptop/desktop when your existing system has like 8-16GB RAM, and a 1TB HDD (for instance) and you see a new system with 4-8GB RAM and a 256GB or 512GB SDD -- less RAM and less storage? Yes please (not.) Also with the zeal for some laptops to solder everything on, without checking detailed specs (that Acer for one makes easy to find but HP or Dell for two other examples don't....) you can't buy a system with like 4GB RAM and know you can actually upgrade it to a more reasonable specs.

Edit: Just to point out, I did run Ubuntu on a Chromebook with an ARM, and it was drama-free, just like a normal laptop; I could even run x86 and x86-64 apps under qemu (for a printer driver that was Intel-only, and one or two other small applications.) So Windows on ARM may not be that great but macOS on ARM and Linux on ARM are perfectly fine. (Years back, I worked at U of Iowa surplus and was into buying up cheap used UNIX systems; I had a successful Linux desktop running on DEC MIPS, DEC Alpha, HP PA-RISC, SGI MIPS, and Mac PowerPC. So I'm sure if RISC-V dekstop/laptops shipped a Linux desktop can be brought up on those too. Really the only reason I have an Intel system now is ubiquitousness, I'd be perfectly fine running a different CPU if a nice laptop or desktop shipped with one in it.)

GPUs -- ARC has excellent Linux support, I have a Tiger Lake notebook and it's great, full OpenGL and Vulkan support that runs every game I throw at it (through Wine or Steam/Proton.). But what gamer (using Windows) is going to buy an ARC GPU for like $200-300 when it has DX12 support but no support for DX9, 10, or 11, or OpenGL?

 

[HardReset]

It's just like how AMD does just about everything with their chiplets.

Need an Epyc chip? Use the chiplets.
Need a Threadripper? Use the chiplets.
Need a desktop processor? Use the chiplets.
Need a mobile processor? Use the chiplets.

AMD mobile processors are mostly monolithic. That is because chiplet design pretty much sucks power wise when it comes to low power, low usage (mostly idle) scenarios. Chiplet design is great for servers. For everything else it's pretty much good only for saving money and/or top notch manufacturing capacity.

 

[Noah Genda]

The best analogy we can think of here is automobiles. Mercedes sells around 10% of cars in the US, just as Intel has about 10% of industry capacity

Intel Revenue in 2022: $63.1 Billion. Intel Net Income in 2022: $8.0 Billion
AMD Revenue in 2022: $23.6 Billion AMD Net income in 2022: $1.3 Billion.

To compare capacity to sales or profit is absurd. AMD has no capacity at all, as they
don't own a foundry.

Intel may be struggling, but by no means are they in danger of going out of business.

 

[WaltC3]

The current management at Intel is, as the article correctly states, not to blame for the present situation. Indeed, the current management was a knee-jerk response to AMD beginning to seriously eat into Intel's lunch, just a couple of years ago, by competing in Intel's product space with demonstrably superior products. Company bean counters longed for the halcyon days when Intel was the high-end x86 monopolist back in the (original) Pentium heyday. Thus, Gelsinger was dragged from the retirement closet to take the reins of an Intel in a current market state completely unfamiliar to Gelsinger at any time in his previous Intel experience. It's not working out. (Understatement.)

Gelsinger started with some assumptions about Intel that he had that he's had to jettison, painfully I would say, simply because Intel isn't the same company it was, and because Intel has never had sustained competition of the kind AMD is providing today (even with TSMC's obvious technical superiority over Intel's own FABs, it's AMD's design and engineering prowess that bring the value of TSMC's advanced foundries to the fore.) AMD beat Intel before, with K7/A64/Opteron (leapfrog tech advances), but in those days Intel was still paying companies like Dell not to sell AMD CPUs and related products, and getting away with it--which Intel cannot expect to do these days for a number of incontrovertible legal and market conditions. But the main reason that AMD slowed significantly post Opteron was because AMD had no Act II or III or IV to move into after Opteron, which was a pity, imo. Today, AMD is not making the same mistakes but is executing a prolonged, multiyear plan for technology that will keep Intel in the laggard column for a very long time, it appears. AMD's own wonderful talent of near-flawless execution of its roadmaps those plans is something Intel no doubt admires, if not outright envies.

How did this happen? The same way that AMD eclipsed Intel with the k7/A64/and Opteron. AMD came out of the blue to challenge and beat an arrogant, snoozing Intel that was very impressed with itself. For the past decade prior to AMD's climb of the last five years, Intel has been resting on its laurels and raking in the dough from the market position Intel is best suited for--from a position of a high-end x86 CPU monopolist. Having beaten AMD at its own game essentially, and even licensing x64-64 from AMD (after failing to inspire the markets to move to Itanium and Rdram and reject AMD's x86-64), and then moving to a 64-bit x86-64 CPU and bus known as Core 2, Intel began snoozing again, and raking in the dough, again, convinced that AMD was now forevermore located as shrinking flotsam in the Intel rearview mirror... Amusing when I think about it, but amusing also because the sentiment is so accurate.

The interesting thing for me is to see how Intel is struggling here--as AMD is a dramatically moving target which is so far managing to stay ahead of Intel technically as the years roll on! The 64 Thousand Dollar Question is this: will Intel ever again eclipse AMD? In just a couple more years, the question will be answered. Having your own FABs I thought at the time was the ticket for AMD in the k7/A64/Opteron era. But as AMD discovered and as I learned, owning your own FABs can be disastrous for a company if suddenly you cannot operate them at capacity selling your own products! So AMD divested itself of its FABs by doing a life-saving deal with Global Foundries! Contrary to being an EOL move of desperation (well...that's really what it was...), AMD then had room enough not only to survive but to continue on and to mature into the juggernaut of a company that AMD had always worked hard to achieve!

The final AMD advantage it has over rival Intel is the fact that at no time since its founding has AMD ever been an x86 high-end CPU manufacturer! AMD doesn't even know how to think like one... When Intel was raking in the dough, AMD was burning the midnight oil and scrambling and doing so on one-tenth of the resources of Intel. And there you have the qualitative difference between the two companies. It is significant, in my view.

 

[rsmith222]

Why? So AMD can rail us like they did with socket 939?

These companies are not our friends, never forget that

I prefer the FX 9590 as an example. At least the 939 stuff was good tech.

The FX 9590, on the other hand, was a predatory product designed to fleece enthusiasts who didn't follow tech closely. It wasn't good enough to be called half-baked, and that includes how it caused certified motherboards from ASRock to catch fire.

Vega is another good example. AMD promised to the moon but instead of getting that we got 'Polaris forever' in order to prop up the thoroughly mediocre consoles of the time. Vega's IPC was identical to Fury X's, after everyone waited so long. Then, AMD came out with Radeon VII which was a too-small die pumped too high in clocks to pad AMD's margin.

At least the company finally decided to make good CPUs again. We're still waiting for it to compete with products like the 4090 in performance.

 

[rsmith222]

Intel completely deserves to be exactly where they are and worse after all the ridiculous and underhanded **** they did to competitors like AMD over the years.

The public deserves behavior like that for enabling the invention called the corporation to do what it does.

Corporations are designed to be amoral and the humorous claim is that that's the path to morality (I.e. profit for the rich at the expense of the planet and most of its people, although some of them get shiny beads in the process.)

 

[yRaz]

I prefer the FX 9590 as an example. At least the 939 stuff was good tech.

The FX 9590, on the other hand, was a predatory product designed to fleece enthusiasts who didn't follow tech closely. It wasn't good enough to be called half-baked, and that includes how it caused certified motherboards from ASRock to catch fire.

Vega is another good example. AMD promised to the moon but instead of getting that we got 'Polaris forever' in order to prop up the thoroughly mediocre consoles of the time. Vega's IPC was identical to Fury X's, after everyone waited so long. Then, AMD came out with Radeon VII which was a too-small die pumped too high in clocks to pad AMD's margin.

At least the company finally decided to make good CPUs again. We're still waiting for it to compete with products like the 4090 in performance.

Vega is a horrible example, it was a compute monster and a large reason nVidia created the 1080ti. To this day, AMDs workstation cards are still based on Vega while their gaming GPUs are a completely different architecture.

And the thing about bulldozer is that it sold horribly, it was almost the end of the AMD. While there some unfortunate customers, anyone who didn't buy a 3770k at the time was a fool.

Now, I'd like pose a question. Does AMD even need a 4090 competitor? The 7900xtx is outselling the 4080 and the 4070ti, which is slower than than 7900xt, has cause the xt into a "blow MSRP situation". You can regularly find the 7900xt at $30-50 US below MSRP but they tend to sell out quick at that price.

Now, AMD has an interesting solution to 4090 that it could use if it so desired, it's chiplet design. It would be almost trivial for AMD to make a 4090 competitor but nVidia has already saturated the market with the high end and the sales volume likely isn't there.

But we are also left with absurd level "new old stock" on both sides. AMD nor nVidia has any inclination to make lower end cards or drop prices until their old stock has moved. And there is A LOT of it left to move.

Going back to intel/AMD. Nothing could be worse for the industry than Intel failing to make the AMD fanbois happy. We are not these companies friends, we no longer "customers" we are "consumers" and never forget that