News Intel 2Q24 Financial Results

[Kuiva maa]

Here is a question, have we reached a point where what we have is good enough for 95% of the market? If so, there is little incentive for general public to buy the new shiny thing even the relative performance is high. Majority of general purpose software is not limited my modern hardware, and many use cloud based software. It would take few years for software to catch up to the current hardware. What I am trying to say is, Intel (and AMD) have to find different market/target for new CPUs. Otherwise they won't able cover the R&D and manufacturing cost.

Having said that, personally I want more than 16 cores(p-cores) in desktop .

The consumer market for PCs is in a weird spot and has been there for a while now, but it gets increasingly tough.

General purpose computing isn’t quite a thing anymore. There is no “home computer”, people use personal mobile devices for media consumption, communications etc. That corner is iOS/Android domain.

What PC has is certain segments.

Students and professionals still want windows machines (mostly laptops) for productivity. Battery life, size, thermals, screens etc do matter there so in a sense there is always room for better hardware.

Then there is the whole gaming market. With discrete GPUs slowly becoming very expensive, the DIY market is in murky waters but premade desktops by big brands and gaming laptops are still there and they need advanced hardware all the time. If you add that the handheld gaming PC rise, it is evident that gaming PCs become “consolized”. They are devices that are often almost entirely dedicated to gaming (I can think only of render workstations that are widely used on a consumer basis and use similar hardware to gaming PCs).

What Intel should have done (and still can) in the consumer domain is swallow their pride and utilize their unique position. They are the only company that has x86 CPUs, GPUs and their own fabs. They should do what AMD did once upon a time. Accept lower margins and start offering gaming SoCs, semi custom even (for consoles and handhelds), and gaming GPUs (DIY, laptop even). But they need to go all in and not just top their toes like they have been doing for so long. Bring a full product range from top to bottom at half the price or lower of what nvidia offers.

 

[FlameTail]

[Bloomberg] Arm Is Rebuffed by Intel After Inquiring About Buying Product Unit

Arm Holdings Plc approached Intel Corp. about potentially buying the ailing chipmaker’s product division, only to be told that the business isn’t for sale, according to a person with direct knowledge of the matter.

Oh boy.

Note that it's ARM, not Qualcomm.

 

[TESKATLIPOKA]

I really don't know how Intel wants to finance It's foundries.

They already had a $7 billion operating loss during 2023 and at that time they didn't outsource anything(except GPUs) to TSMC except Meteor Lake, which was launched on December 14.
Now they have LNL, which is completely made at TSMC expect the Interposer, but that is only 22nm.
Arrow Lake looks like It will be also made at TSMC.
I don't know what they want to actually make in their FABs. I wouldn't be surprised If they make a loss this year as a whole company.

 

[coercitiv]

I really don't know how Intel wants to finance It's foundries.

Think of Intel as a shark, but not in the sense of a predator, rather in the sense of it's particular physiology: they need to keep moving so that they keep breathing. Intel's momentum is what should keep it alive short term.

As long as their designs sell relatively well in both consumer and datacenter, their foundries get some orders from 3rd parties + some U.S. defense orders + Intel's own server chips, they should be able to keep enough money flowing so that bankers will see a business that can dig itself out. The money they get from the CHIPS Act which IIRC also includes access to cheap loans should be a very potent source of oxygen for Intel in the years to come.

The one catch is their 18A node and their in-house high volume products based on this node: they must deliver both in good health (and on time).

 

[jpiniero]

I really don't know how Intel wants to finance It's foundries.

That's the neat part, they don't.

I think the Board was finally clued into this reality; hence the approval to spin it off.

 

[cebri1]

That's the neat part, they don't.

I think the Board was finally clued into this reality; hence the approval to spin it off.

There is no approval to spin it off. So far they are going to spin it in, which is expected because no one is going to share designs with the foundry if they are not 100% that their IP is safe.

 

[Hitman928]

Didn’t really have a good thread for this so putting it here. Report says that Intel is cutting Gaudi 3 planned shipments for next year by 30%. Intel did not offer comment on the report.

[News] Intel May Cut Gaudi 3’s 2025 Shipment Target by 30%, Raising Concerns for TSMC, ASE, and Alchip | TrendForce News

To turn adversity around, Intel launched its latest AI accelerator, Gaudi 3, in late September. However, a report by the Economic Daily News indicates...

www.trendforce.com

 

[desrever]

Didn’t really have a good thread for this so putting it here. Report says that Intel is cutting Gaudi 3 planned shipments for next year by 30%. Intel did not offer comment on the report.

[News] Intel May Cut Gaudi 3’s 2025 Shipment Target by 30%, Raising Concerns for TSMC, ASE, and Alchip | TrendForce News

To turn adversity around, Intel launched its latest AI accelerator, Gaudi 3, in late September. However, a report by the Economic Daily News indicates...

www.trendforce.com

This looks like one -redacted- number being lowered to another -redacted- number. Only way Intel is going to sell 250k Gaudi 3 is if they pay people to use it.

As you already know, Profanity is not allowed in the tech forums.

Daveybrat
AT Moderator

 

[Joe NYC]

This looks like one bullshit number being lowered to another bullshit number. Only way Intel is going to sell 250k Gaudi 3 is if they pay people to use it.

Intel is entering into some sort of AI cloud venture, where Intel will be buying Guidi from itself (via this new venture) and will compete with Intel's other customers, such as AWS.

 

[jpiniero]

This looks like one bullshit number being lowered to another bullshit number. Only way Intel is going to sell 250k Gaudi 3 is if they pay people to use it.

Says ship, not sell.

 

[Hitman928]

Says ship, not sell.

These aren't going into retail channels.

 

[Hitman928]

Intel is entering into some sort of AI cloud venture, where Intel will be buying Guidi from itself (via this new venture) and will compete with Intel's other customers, such as AWS.

Do you have a link for this? I haven't heard about this plan.

 

[coercitiv]

Do you have a link for this? I haven't heard about this plan.

Probably this?

Inflection AI and Intel Launch Enterprise AI System

Inflection for Enterprise, powered by Gaudi and Intel Tiber AI Cloud, helps enterprises tackle critical workloads with AI

www.intel.com

 

[Hitman928]

Probably this?

Inflection AI and Intel Launch Enterprise AI System

Inflection for Enterprise, powered by Gaudi and Intel Tiber AI Cloud, helps enterprises tackle critical workloads with AI

www.intel.com

That doesn't really match what was described, though. . .

 

[Joe NYC]

Do you have a link for this? I haven't heard about this plan.

Here is the tweet and the link:

Intel Debuts AI Cloud With Gaudi 3 Chips, Inflection AI Partnership

Intel has launched Intel Tiber AI Cloud featuring its Gaudi 3 accelerators and said generative AI startup Inflection AI will use the service for a new enterprise AI solution.

www.crn.com

https://twitter.com/x/status/1843369051617399096

 

[TESKATLIPOKA]

Think of Intel as a shark, but not in the sense of a predator, rather in the sense of it's particular physiology: they need to keep moving so that they keep breathing. Intel's momentum is what should keep it alive short term.

As long as their designs sell relatively well in both consumer and datacenter, their foundries get some orders from 3rd parties + some U.S. defense orders + Intel's own server chips, they should be able to keep enough money flowing so that bankers will see a business that can dig itself out. The money they get from the CHIPS Act which IIRC also includes access to cheap loans should be a very potent source of oxygen for Intel in the years to come.

The one catch is their 18A node and their in-house high volume products based on this node: they must deliver both in good health (and on time).

Intel foundry business had a 7 billion operating loss last year.
That was when they made Raptor Lake and in the last quarter also Compute tile for Meteor Lake.
Now both Lunar and Arrow Lake won't be made by Intel foundries but by TSMC, Raptor Lake had the voltage issue, so we can likely expect a bigger loss in 2024 and 2025.
So in the end they will have to pay TSMC big money for wafers compared to 2023, yet still support their foundry from profits of the whole company, which are already pretty small($1.6B last year).

Some orders from 3rd parties + some U.S. defense orders won't cover the operating cost of the foundries and cheap loans are just a short term solution and an additional mid(long)-term burden unless they can fix their node competitiveness and there is still the question how much would the banks be willing to lend them.
18A node could help a lot, but even If It's a good node can the wafer orders for their server chips be worth 6-7 billion $ per year?

P.S. Intel would need to order 240,000-280,000 wafers costing $25,000 per wafer for $6-7 billion revenue for their foundries. Even with a 500mm2 die It would mean they need to sell 27-32 million dies. I think that's not realistic.

 

[coercitiv]

Intel foundry business had a 7 billion operating loss last year.

Is Intel foundry selling wafers to the the rest of the company at market prices? What loss did Intel as a whole company have last year?

 

[TESKATLIPOKA]

Is Intel foundry selling wafers to the the rest of the company at market prices?

I don't think so, I just made a very very optimistic scenario. So in reality they would need to order even more wafers and sell more server CPUs.

What loss did Intel as a whole company have last year?

$1.7B profit and foundries had $7B operating loss.

edit:
Let's not forget that Intel's biggest revenue comes from client, yes, from the CPUs, which will be made by TSMC. The more they sell the more they will order and pay to TSMC.


edit2: Now that I look they really had only $762 million income before taxes. State paid them back $913 million.
When they had $7.7B profit in 2022, the government still gave them back $250 million. Ridiculous.

P.S. I have to move to the US, start a company so the state will pay me back instead of me paying taxes.

 

[jpiniero]

Is Intel foundry selling wafers to the the rest of the company at market prices?

Yes... but the 10 nm prices are likely very low.

 

[TESKATLIPOKA]

Yes... but the 10 nm prices are likely very low.

Actually, they can price wafers however they want, because It still is a single company.
It depends on which business unit they want to look better. Foundries with smaller loss or the other groups with higher profit, of course revenue will stay the same.

@gdansk didn't think about that, what you said is true considering this is a publicly traded company.

 

[jpiniero]

Students and professionals still want windows machines (mostly laptops) for productivity. Battery life, size, thermals, screens etc do matter there so in a sense there is always room for better hardware.

Vast majority of the PC market is Corpos and their Corpo machines.

 

[gdansk]

Actually, they can price wafers however they want, because It still is one company.
It depends on which business unit they want to look better. Foundries with smaller loss or the other groups with higher profit.

It seems like pricing it however they liked could be fraud to investors. Especially if it can be shown they did so to make one part of their company look more attractive.

I really hope they aren't doing this but have some justifiable metric for determining the price of their intracompany transactions.

 

[jpiniero]

Actually, they can price wafers however they want, because It still is a single company.
It depends on which business unit they want to look better. Foundries with smaller loss or the other groups with higher profit, of course revenue will stay the same.

True but Intel's never even tried to pretend they could get foundry customers on 10 nm past IDM 1.0's collapse. And they have a ton of capacity.

 

[coercitiv]

P.S. I have to move to the US, start a company so the state will pay me back instead of me paying taxes.

Once a business grows in scale past a certain point, very odd things happen - like a country fighting to NOT collect 13 billion in taxes

PS: rest assured, either one of us would pay every penny.

 

[TESKATLIPOKA]

Once a business grows in scale past a certain point, very odd things happen - like a country fighting to NOT collect 13 billion in taxes

PS: rest assured, either one of us would pay every penny.

Yeah, that one was hilarious, but I think It was because Ireland feared that many other big companies would also move out If they taxed Apple, because likely they would be next. Ireland probably could lose more than gain from these taxes in long term.

Not surprising, just look what was Ireland's HDP growth in the last 20 years, the growth was astonishing 174% or 8.7% per year. And It was not because of their economy.