News Intel 4Q25 Earnings

[DavidC1]

wait itsn't that a good thing? Are these server products selling well?

Their DC did increase in revenue over last year, but a mild increase of 9%. 9% is normally very good, except compared to what the Mag 7 are getting.

So, they built too many fabs for the IDM 2.0 strategy, but without having customers, so they had to scale down the plans, yet they are having supply issues, because lot are still on older Raptorlake product, but that's due to moving most of the wafers to server supply, yet despite that they only got 9% increase in revenue for DC/AI. The company is somehow full of contradictions.

It does not make sense since Intel is at the same time saying they are in the biggest shortage in years, and Intel is still saying it as of now. They anticipate Q1 to be the worst point of the current shortage.

So the big question is where did all of Intel capacity go? Intel used to serve a LOT more demand and used to generate a lot more revenue with the existing capacity that's in place.

All of the 14nm and 10nm capacity that used to generate 25-30% more revenue more CPUs. All that capacity is still good for Intel 10 and Intel 7 today. And Intel says this is the capacity that has the biggest shortage.

I'm only taking what they are saying at a face value. I think it was during the lockdown era when they had supply shortages, but then the demand crashed shortly after that and somehow they are in this situation again. Prior to that had to do with fabbing the Apple modem pulling resources out of client division but they lost that deal.

 

[jpiniero]

Their DC did increase in revenue over last year, but a mild increase of 9%. 9% is normally very good, except compared to what the Mag 7 are getting.

So, they built too many fabs for the IDM 2.0 strategy, but without having customers, so they had to scale down the plans, yet they are having supply issues, because lot are still on older Raptorlake product, but that's due to moving most of the wafers to server supply, yet despite that they only got 9% increase in revenue for DC/AI. The company is somehow full of contradictions.

Probably eating a lot of wafers for Emerald Rapids for not a big price.

Could be worse.

 

[jpiniero]

Prior to that had to do with fabbing the Apple modem pulling resources out of client division but they lost that deal.

Intel was the one who bailed on that because of 10 nm.

 

[DavidC1]

PCMs suck balls since there not scalable.

It's not just that, they made subpar decisions.

First, they branded the 16/32GB modules under the name "Optane Memory", but they were HDD accelerators, with whole list of caveats such as locking your system out totally if it failed. Instead, it should have lived up to the name and acted as "Slow Memory" complementing DRAM as an extension. It's not like the infrastructure to do this didn't exist. Their datacenter SSD had a software called Memory Drive, which did precisely that. That would have had lot more value than an HDD accelerator and more justified the price. RAM isn't just for performance. At certain point there's a hard line where you have compatibility issues or the performance plummets. The Memory Drive + Optane Memory would have worked significantly to alleviate this at a much lower cost.

Second, they should have worked with AMD to make sure the PM modules work with them. It should have been the goal to make the PM DIMMs fully compatible with DIMMs, where the only difference is the underlying technology, and to get it recognized easy as certifying other DRAM modules. While it seems contradictory to work with a competitor, it would serve to prop you up during times when your own product is uncompetitive and you would partially offset that by the competitor doing well and buying your other products. That would increase volume, which would allow them to gain more real world experience, lower cost. It would have been perfect TODAY.

Third, the goal should have been moving to PC as quickly as possible. 128GB modules for $300.

The big issue with Intel is that they are so into trying to optimizing 1% margin, they lose the sight of big picture. Work with everyone, competitor or not, which would serve to reduce the valleys and peaks of revenue and stabilize the business. They at least did better with the WiFi business, but mainly because it was so ubiquitous in the first place, people would often ditch Marvell and AMD-branded ones to install Intel's. But same could have been applied to Optane.

Without this mindset changing, Intel's pitfalls will continue in the future. Novalake, Unified Core, it doesn't matter. Are the fundamentals fixed where they have a laser-focus on 1% margins/revenue and proprietary interfaces to lock out competition? Without changing this, the long term trajectory for Intel is still a decline, eventually into bankruptcy. This includes Lip "No products under 50% margin" Bu Tan.

 

[Joe NYC]

This includes Lip "No products under 50% margin" Bu Tan.

Delivering 37.9% gross margin, going down to 34.5% gross margin next quarter. When, we are led to believe, will be the most severe shortage.

While we are in middle of memory shortage, the memory makers are printing money, experiencing their best profitability in a decade.

Intel, while in middle of shortage, is guiding down revenue and profitability. It just does not add up.

 

[poke01]

let me guess nvidia and apple

 

[DavidC1]

Delivering 37.9% gross margin, going down to 34.5% gross margin next quarter. When, we are led to believe, will be the most severe shortage.

While we are in middle of memory shortage, the memory makers are printing money, experiencing their best profitability in a decade.

Intel, while in middle of shortage, is guiding down revenue and profitability. It just does not add up.

What happens and how bad will it get if the RAM/SSD prices crash demand then?

 

[Saylick]

You know, they say hindsight is 20/20, but now I call theml, Intel "bad timing" Corporation.

They make a decent chip with Pantherlake, at a time when the prices in memory and SSD will likely limit it's impact. And they got their dGPU out months after crypto boom stopped. And if Xe4 rumors are accurate, then perhaps after AI market crashes they would be significantly behind competition, to the likes of B390 vs 890M.

Had they continued their Optane lineup, they could have benefitted from the AI push for their datacenter lineup. Heck, they could have got enthusiast Optane DIMMs out for PCs, and have a slow memory + fast memory tier. The DIMMs with 100-300nS latency is actually fast enough to be RAM for vast majority of use cases.

For sure. I know people make fun of AMD's GPU division for never missing an opportunity to miss an opportunity, but one can argue it applies to most of Intel. To list a few complete blunders off the top of my head:
- Pantherlake looks good, but it's releasing at a time when memory prices are through the roof, thus stifling demand
- GPU accelerator market is growing rapidly due to AI boom, but Intel was too little, too late with investing in their own GPU development.
- Selling NAND memory is exceptionally lucrative right now, but Intel sold off their memory division to SK Hynix in 2020.
- Intel completely missed the timing for EUV adoption and double downed on DUV multi-patterning and cobalt interconnects, which let TSMC catch up and overtake them.
- They took forever to pivot to a chiplet architecture, which let AMD blindside them in the server market with higher core counts and Intel is still bleeding market share.
- Intel didn't want to pursue Apple's SOC win for the iPhone because they underestimated the volume of sales it would have, and thus the financials didn't pencil out. ARM takes over the mobile chip market.

All in all, Intel's culture is not one of agility. They try to pivot, but by the time it happens they've missed the boat. Imo, Intel's stock price rise over the last 6 months is driven purely by hopium and a belief that the USG will never let them fail.

 

[Joe NYC]

For sure. I know people make fun of AMD's GPU division for never missing an opportunity to miss an opportunity, but one can argue it applies to most of Intel. To list a few complete blunders off the top of my head:
- Pantherlake looks good, but it's releasing at a time when memory prices are through the roof, thus stifling demand
- GPU accelerator market is growing rapidly due to AI boom, but Intel was too little, too late with investing in their own GPU development.
- Selling NAND memory is exceptionally lucrative right now, but Intel sold off their memory division to SK Hynix in 2020.
- Intel completely missed the timing for EUV adoption and double downed on DUV multi-patterning and cobalt interconnects, which let TSMC catch up and overtake them.
- They took forever to pivot to a chiplet architecture, which let AMD blindside them in the server market with higher core counts and Intel is still bleeding market share.
- Intel didn't want to pursue Apple's SOC win for the iPhone because they underestimated the volume of sales it would have, and thus the financials didn't pencil out. ARM takes over the mobile chip market.

All in all, Intel's culture is not one of agility. They try to pivot, but by the time it happens they've missed the boat. Imo, Intel's stock price rise over the last 6 months is driven purely by hopium and a belief that the USG will never let them fail.

Another potential one we may talk about in 5 years:
- Intel paying 2x price for High NA EUV machines, while those machines were not needed for another 5 years. It slowed down potential growth of Intel foundry, and made Intel costs too high to compete.

 

[Saylick]

Another potential one we may talk about in 5 years:
- Intel paying 2x price for High NA EUV machines, while those machines were not needed for another 5 years. It slowed down potential growth of Intel foundry, and made Intel costs too high to compete.

Right... SemiAnalysis estimated that a single pass of High-NA EUV to be 2.5x the cost of Low-NA EUV, so it's only worth using it on layers that normally would require 3 or more passes of Low-NA EUV. Also, Intel said they are designing 14A for both multi-pass Low-NA EUV and single-pass High-NA EUV, which implies that High-NA is currently a nicety but not a necessity.

 

[desrever]

Intel now makes most of their money from client sales. They are going to be killed by high ram prices destroying client demand for the next 2 years.

 

[branch_suggestion]

For sure. I know people make fun of AMD's GPU division for never missing an opportunity to miss an opportunity, but one can argue it applies to most of Intel. To list a few complete blunders off the top of my head:
- Pantherlake looks good, but it's releasing at a time when memory prices are through the roof, thus stifling demand
- GPU accelerator market is growing rapidly due to AI boom, but Intel was too little, too late with investing in their own GPU development.
- Selling NAND memory is exceptionally lucrative right now, but Intel sold off their memory division to SK Hynix in 2020.
- Intel completely missed the timing for EUV adoption and double downed on DUV multi-patterning and cobalt interconnects, which let TSMC catch up and overtake them.
- They took forever to pivot to a chiplet architecture, which let AMD blindside them in the server market with higher core counts and Intel is still bleeding market share.
- Intel didn't want to pursue Apple's SOC win for the iPhone because they underestimated the volume of sales it would have, and thus the financials didn't pencil out. ARM takes over the mobile chip market.

All in all, Intel's culture is not one of agility. They try to pivot, but by the time it happens they've missed the boat. Imo, Intel's stock price rise over the last 6 months is driven purely by hopium and a belief that the USG will never let them fail.

Intel and Qualcomm are both missing the same boats and are doomed to poverty TAM hell.

 

[QuickyDuck]

18A is hurting Intel's margin badly

 

[adroc_thurston]

I know people make fun of AMD's GPU division for never missing an opportunity to miss an opportunity

It's just client proles being permabutthurt that Radeon no longer exists to force NV pricecuts.
Silly critters.

Intel and Qualcomm are both missing the same boats and are doomed to poverty TAM hell.

Ironically, both are parasites.

 

[inquiss]

It's just client proles being permabutthurt that Radeon no longer exists to force NV pricecuts.
Silly critters.

Ironically, both are parasites.

What do you mean by parasites? Only do well when others are doing well?

 

[adroc_thurston]

What do you mean by parasites?

Means they leech off tightly gatekept IP/patent portfolios.
Qualcomm makes modems and everything else stems from that. It's why auto has been the only diversification thing at QC that did not detonate catastrophically.

 

[DrMrLordX]

18A is hurting Intel's margin badly

Mind explaining how? Q4 numbers didn't even include any 18A products.

 

[DavidC1]

All in all, Intel's culture is not one of agility. They try to pivot, but by the time it happens they've missed the boat. Imo, Intel's stock price rise over the last 6 months is driven purely by hopium and a belief that the USG will never let them fail.

And what happens when USG fails?

The "too big to fail" is a big mistake. Let. Them. Fail. Otherwise they will continue to make those mistakes. But prevent that from happening again. The GM and Chevy executives were flying in from private jets when they were getting bailed out by the US Government.

Sure, it hurts like hell to let that happen. But that's how you prevent things from happening again. What happens when you coddle a kid? You'll get a spoiled adult out of it. The world is harsh, and not fair either. Deal with it. But now I'm not sure there's any long term solutions to this. The whole system is now extremely unstable.

 

[jpiniero]

Read the 10-K... Client ASPs were flat for the entire year. Server ASPs decreased 4%.

 

[QuickyDuck]

Mind explaining how? Q4 numbers didn't even include any 18A products.

bleak margin looking forward

 

[reaperrr3]

One thing about is that does not make sense is that Intel is making it sounds like there is a gap. As if the production output disappeared.

If the capacity was diverted, it means that new wafers "in" were diverted, and all of the wafers that were currently in process just continued through the fab and to packaging. They did not disappear.

So even if new wafer starts shifted in H2 2025 to server, the old wafers (Raptor) prior to switch would still be going out of the fab throughout Q3, Q4. Yet, the client revenue in Q4 is down.

Well, we don't know the exact delay from wafer to finished product for Raptors.

If "25H2" meant as early as early July, and if the shift was substantial, and with Xmas driving up consumer demand and most client production going to mobile due to OEM contracts, desktop might've gotten hit disproportionally.

But yeah, might be a mix of some truth and some convenient excuse.

 

[Doug S]

I find it very hard to predict what's going to happen this year, because of the RAM market. I'd expect that the consumer side of both AMD and Intel won't do that well, but if your average AI server has more ram per cpu than other servers, that might hurt the server side too.

Everyone (well not everyone but a LOT of would be buyers) will be looking to stretch the life of their current hardware a little longer, and that effect will magnify the higher DRAM/NAND pricing gets.

If you were planning on upgrading a bunch of laptops to migrate off Windows 10, suddenly it is cheaper to keep the old ones around and pay the Microsoft tax for extended support. If you were going to upgrade a bunch of servers to be faster/more efficient suddenly it is worth paying for more electricity to run the less efficient servers. Everyone's TCO calculations have seen major, material changes over the past few months. Those numbers are only going to get further out of whack favoring the "keep our old stuff longer than originally planned" side of the TCO calculation.

There are always some who have no choice but to suck it up and pay the going rate, but those who have a choice are going to make very different choices in 2026 than they had previously projected, and that's going to hit both Intel and AMD's client AND (non AI) server sales.

 

[Doug S]

Delivering 37.9% gross margin, going down to 34.5% gross margin next quarter. When, we are led to believe, will be the most severe shortage.

While we are in middle of memory shortage, the memory makers are printing money, experiencing their best profitability in a decade.

Intel, while in middle of shortage, is guiding down revenue and profitability. It just does not add up.

Of course they're guiding down. PC/server demand for 2026 is going to get absolutely crushed outside of the AI space, and Intel doesn't play in the AI space. That's a much bigger effect than whatever production shortfalls they may have while they ramp 18A.

 

[adroc_thurston]

PC/server demand for 2026 is going to get absolutely crushed outside of the AI space

Servers are healthy and PCs are barely down.
Intel just got caught with their pants down wrt i7 capacity.

 

[DavidC1]

Servers are healthy and PCs are barely down.
Intel just got caught with their pants down wrt i7 capacity.

Cause DRAM prices haven't fully caught on yet. It usually takes about a quarter. And SSD prices are going up too. Let's see how Q1 is.

Higher prices have a nonlinear effect on volume. 10% increase in prices is greater than 10% difference in volume.

It's even worse than that. Intel released a dedicated crypto mining chip just after the crypto mining boom crashed.

So somehow they have no foresight of any sort. I don't know if this is a) they are all stupid b) it's too chaotic c) even malice/intentional.