News Intel GPUs - Intel launches A580

[Heartbreaker]

Intel's problem is not a declining future but a declining present. The future is now. Lower margin gamer GPUs is not a good investment when you're hemorrhaging money and will continue almost certainly for several more years. Just compare the silicon needed for between CPUs and GPUs for a given sales price, plus all that essential software work with each big game.

A couple of losing quarters is nothing. Just ask AMD. They weathered years of losing quarters.

It's better to have some short term pain now, than to be increasingly non competitive in the future.

 

[Insert_Nickname]

It's better to have some short term pain now, than to be increasingly non competitive in the future.

I take it you've never met an MBA?

Profit now trumps everything else. Even to the detriment of the company in question.

 

[Heartbreaker]

I take it you've never met an MBA?

Profit now trumps everything else. Even to the detriment of the company in question.

Gelsinger is an Engineer, not an MBA.

 

[Insert_Nickname]

Gelsinger is an Engineer, not an MBA.

Oh, I wasn't referring to Pat specifically. Just a general observation.

 

[DeathReborn]

A couple of losing quarters is nothing. Just ask AMD. They weathered years of losing quarters.

It's better to have some short term pain now, than to be increasingly non competitive in the future.

They did have to pawn/sell the family silver to survive though, I doubt Intel will accept multi billion losses if it means selling Fabs.

 

[jpiniero]

A couple of losing quarters is nothing. Just ask AMD. They weathered years of losing quarters.

It's better to have some short term pain now, than to be increasingly non competitive in the future.

Even with the Chips Act fun bucks they won't be able to afford leading edge fabs for long if this continues. Especially if there are any more delays.

 

[maddie]

A couple of losing quarters is nothing. Just ask AMD. They weathered years of losing quarters.

It's better to have some short term pain now, than to be increasingly non competitive in the future.

Joking here.

So Mr Seer, how many quarters you'll bet you life against? I really think this will be a lot more than a couple of quarters if nothing changes. Their server fiasco, for one, is very far from stabilizing.

 

[Heartbreaker]

Joking here.

So Mr Seer, how many quarters you'll bet you life against? I really think this will be a lot more than a couple of quarters if nothing changes. Their server fiasco, for one, is very far from stabilizing.

So are you going to suggest they abandon the Server business next? Then abandon the Fab business?

 

[maddie]

So are you going to suggest they abandon the Server business next? Then abandon the Fab business?

I'm suggesting that if you need to cut expenses, you cut that which will have the lowest margin and probably the longest time before positive returns.

There is a sense of denial that Intel is in real trouble. The wrong decisions could sink them to 2nd class status. This is possible, as hard to imagine as it seems. Many dominant companies have gone that route.

I'm not disagreeing with you on the benefits of GPUs, just that business as usual time has ended. A smarter attempt when conditions have stabilized is one possibility. Keep basic work going which they have to anyhow because of the iGP needs.

The interview with Jon Peddie on MLID was illuminating. He observed that Intel hired anyone who could spell GPU. Obvious hyperbole, but in the main I believe him. They were stupid.

 

[Heartbreaker]

There is a sense of denial that Intel is in real trouble. The wrong decisions could sink them to 2nd class status. This is possible, as hard to imagine as it seems. Many dominant companies have gone that route.

I get the sense of chicken little from you. One bad quarter and the sky is falling.

 

[DrMrLordX]

I get the sense of chicken little from you. One bad quarter and the sky is falling.

The problem Intel has isn't that they had one (or more than one) bad quarter; it's WHY they had a bad quarter, along with how their competition fared in the same time period.

Intel managed to profit despite a tech deficit, especially in the server room where 14nm shipments represented an outsized share of their volume and revenue. Now that trend has finally reversed. They can't make money off Cascade Lake-SP anymore, and they don't have a replacement that can ship in sufficient volume to satisfy customers. All they have is IceLake-SP which is a). already slower than Milan and b). still not shipping in volumes similar to Cascade Lake-SP at its peak.

Add to that fact that what is arguably Intel's most competent CPU product (Alder Lake) was not sufficient to fight off surprising losses in the client computing space. They actually saw a bigger dip there than in datacentre, despite Intel's client product portfolio being relatively much stronger than datacentre.

Intel will survive in one form or another, but if they can not reverse this trend quickly, they may become a very different company soon.

 

[Heartbreaker]

Intel will survive in one form or another, but if they can not reverse this trend quickly, they may become a very different company soon.

One losing quarter is not a trend. If they have a couple more losers, then we can talk trends.

 

[DrMrLordX]

One losing quarter is not a trend. If they have a couple more losers, then we can talk trends.

Their revenues have been in decline for awhile. That's the trend. And it suddenly got a lot worse.

 

[Heartbreaker]

Their revenues have been in decline for awhile. That's the trend. And it suddenly got a lot worse.

Lower revenues are a fact of life for Intel now that it has a viable competitor again. Intel was making monopoly style money for a decade, while AMD was asleep. They aren't going to have it that easy again in the foreseeable future (if ever). But they were still making healthy profit.

It got worse this quarter because economic conditions have changed, and PC sales are tanking.

But you don't react to short term market conditions by killing off vital sections of your business to save a few bucks.

How many losing quarters did AMD sustain in the decade before they turned around?

You don't sacrifice your long term future, to make the balance sheet look better in the short term. Intel can weather many more lean or even losing quarters to maintain R&D for it's long term future.

 

[NTMBK]

I'm suggesting that if you need to cut expenses, you cut that which will have the lowest margin and probably the longest time before positive returns.

There is a sense of denial that Intel is in real trouble. The wrong decisions could sink them to 2nd class status. This is possible, as hard to imagine as it seems. Many dominant companies have gone that route.

This misunderstands the whole purpose of the GPU business. Intel need to keep increasing their silicon volumes, so that they can pay the ever increasing cost of new cutting edge fabs. Intel Foundry Services is one way to do that, but another way to do that is with Intel GPUs. They might not be as high margin as server CPUs, but if they can fill fabs then they're worth chasing.

Failing to get sufficient volumes to fill their fabs would be a long term catastrophe for Intel.

 

[jpiniero]

This misunderstands the whole purpose of the GPU business. Intel need to keep increasing their silicon volumes, so that they can pay the ever increasing cost of new cutting edge fabs. Intel Foundry Services is one way to do that, but another way to do that is with Intel GPUs. They might not be as high margin as server CPUs, but if they can fill fabs then they're worth chasing.

Failing to get sufficient volumes to fill their fabs would be a long term catastrophe for Intel.

And yet they are fabbing GPUs at TSMC...

 

[NTMBK]

And yet they are fabbing GPUs at TSMC...

In the short term, yes. They need to get their design teams up and running, their drivers ready, their distribution and channels to market sorted. Might as well do that on TSMC for a couple of generations, then once Intel processes are ready (probably around 20A?) switch to using internal.

This was always a long term plan.

 

[jpiniero]

In the short term, yes.

What you said was the original plan. But you know, 10 nm. It's kind of a catch-22 because they'd get more rope if it was fabbed internally but the products would be even worse.

 

[DrMrLordX]

But you don't react to short term market conditions by killing off vital sections of your business to save a few bucks.

I would agree with you there. It's important not to be dismissive of the threats facing Intel in the immediate future, though, especially since there is much more in play than a "short term market condition".

 

[maddie]

This misunderstands the whole purpose of the GPU business. Intel need to keep increasing their silicon volumes, so that they can pay the ever increasing cost of new cutting edge fabs. Intel Foundry Services is one way to do that, but another way to do that is with Intel GPUs. They might not be as high margin as server CPUs, but if they can fill fabs then they're worth chasing.

Failing to get sufficient volumes to fill their fabs would be a long term catastrophe for Intel.

That seems to be the opposite of a successful strategy. Let me reframe what you wrote.

Lets build fabs and then try to do competitive designs able to maximize their use. What I understand from your post is that Intel has too many fabs and/or the wrong (not enough leading edge) ones, so they need to make work to keep them utilized.

Do your brilliant designs and plan for the fab space needed to produce it. Not build the fab space and then look for designs to use it.

Is build it and they will come really what Intel should be doing? Maybe they should be cutting fabs.

Intel being a fab company as some have suggested, will find this decision impossible to make, until external factors become too stronge to resist. I hope it does'nt reach that point, because if it does, they will have to be in a really precarious spot, with their survival at stake.

 

[Aapje]

The real issue I see is that Intel seems to have grown complacent over time and can't seem to get turn their plans into reality as well as the competition. Intel risks to be 2nd rate at manufacturing and 2nd rate at chip design.

 

[NTMBK]

That seems to be the opposite of a successful strategy. Let me reframe what you wrote.

Lets build fabs and then try to do competitive designs able to maximize their use. What I understand from your post is that Intel has too many fabs and/or the wrong (not enough leading edge) ones, so they need to make work to keep them utilized.

Do your brilliant designs and plan for the fab space needed to produce it. Not build the fab space and then look for designs to use it.

Is build it and they will come really what Intel should be doing? Maybe they should be cutting fabs.

Intel being a fab company as some have suggested, will find this decision impossible to make, until external factors become too stronge to resist. I hope it does'nt reach that point, because if it does, they will have to be in a really precarious spot, with their survival at stake.

The R&D for a new silicon manufacturing process costs billions of dollars, and so does purchasing all the equipment. And each generation gets more and more expensive. There's a reason why IBM, Texas Instruments, Globalfoundries, AMD, STMicro, etc have all given up on making cutting edge chips- because it's too expensive. You need enormous volumes to amortize those costs, and the volumes required increase every generation. That's why the industry has consolidated to just TSMC, Intel and Samsung.

If Intel wants to keep in the fab game, they need a new way to keep increasing their volumes. The PC market isn't growing any more, and their server sales have peaked. They need to find a new market- hence GPUs.

 

[Heartbreaker]

I would agree with you there. It's important not to be dismissive of the threats facing Intel in the immediate future, though, especially since there is much more in play than a "short term market condition".

I also acknowledged they have permanently reduced revenue due to AMD being competitive. But until this quarter there were still profitable.

The reduced revenue and profits will continue for the foreseeable future, if not permanently, but the quarterly loss is almost certainly a short term market conditions (which could be a few quarters).

Intel became fat and lazy in their monopoly years. Now they have to learn to operate leaner and more competitively, which will take time, effort and financial pain.

But that's more of a systemic issue across the board, than an issue dealt with by cutting out major functionality, like GPUs.

 

[maddie]

The R&D for a new silicon manufacturing process costs billions of dollars, and so does purchasing all the equipment. And each generation gets more and more expensive. There's a reason why IBM, Texas Instruments, Globalfoundries, AMD, STMicro, etc have all given up on making cutting edge chips- because it's too expensive. You need enormous volumes to amortize those costs, and the volumes required increase every generation. That's why the industry has consolidated to just TSMC, Intel and Samsung.

If Intel wants to keep in the fab game, they need a new way to keep increasing their volumes. The PC market isn't growing any more, and their server sales have peaked. They need to find a new market- hence GPUs.

I understand why they want to do what you claim, but I stand by what I wrote. They still can succeed, but its a high risk gamble to do what you suggest, plus, I feel I've seen this movie before.

 

[igor_kavinski]

Intel offering Core i7-12700K/i5-12600K alternative CPU prizes to Arc Scavenger Hunt winners - VideoCardz.com

Anyone feel like screaming into a pillow?