Competition just has too much momentum, and we haven’t executed well enough. So we expect That bottoming. The business will be growing. But we do expect that there continues to be some share losses. We’re not keeping up with the overall TAM growth until we get later into 25, 26 when we start regaining share—material share gains.
Intel CEO, Pat Gelsinger at Evercore ISI TMT
https://wccftech.com/intel-anticipates-further-market-share-…
PieR2,
Competition just has too much momentum, and we haven’t executed well enough. So we expect That bottoming. The business will be growing. But we do expect that there continues to be some share losses. We’re not keeping up with the overall TAM growth until we get later into 25, 26 when we start regaining share—material share gains.
Intel CEO, Pat Gelsinger at Evercore ISI TMT
Hmmm…
This is the first acknowledgement from an Intel (NASDAQ: INTC) official that Advanced Micro Devices (NYSE: AMD) has a technological lead over Intel that I have seen. IIRC, the first step of the twelve-step process to recovery is to acknowledge that you have a problem. So maybe Intel is actually beginning that process???
But the twelve-step process does not happen overnight, and AMD is not standing still, resting on its laurels. So what makes anybody think that Intel is suddenly going to start executing well enough to close the technology and performance gap in one or two product cycles???
Mr. Gelsinger’s projections do not seem realistic.
Norm.
To me, the statement by Intel CEO is stunning
If I am AMD
When your main competitor is throwing the white flag, while admitting that the overall market is growing, this to me suggests AMD is a STRONG BUY
PR2
I see it a bit different. Intel has a HUGE credibility problem right now due to missing too many commitments. Admitting a negative that is already obvious to most seems like an easy way to at least start on the road to recovery. However, they are not out of the woods until they show solid execution on several new projects. I don’t see AMD doing anything different here; they already have this information from impartial review sources.
The 2025 game hinges on Intel hitting their Intel 20A/18A process goals ahead of TSMC. Based on past performance most are betting on TSMC, but it is not a given either way.
Alan
Admitting a negative that is already obvious to most seems like an easy way to at least start on the road to recovery.
To me the new revelation is that INTC is admitting it will not be competitive in the next 2-3 years
If they were close, their Marketing and PR team could have used FUD to still be in play. However, it must be really bad for them to not only state they are not competitive now, but they will not be competitive in the near future. All the talk about their upcoming server chips is now rendered moot. To me this is a clear signal for AMD to make hay…
It is pretty clear that in many markets the AMD Genoa will significantly outperform the Intel Sapphire Rapids in many workloads.
From an old insiders point of view, it is 2024/2025 before the real successors to these designs comes out. Intel has already sent the design of both Granite Rapids and Sierra Forest to the fab for initial checkout runs. It can be anywhere from 1.5 years to 3 years from this point to production silicon. I have no visibility into the next AMD server CPU beyond bergamo, but it is likely a 2024/2025 product.
Server customers are a bit different than the consumer market, in that they will generally spend a LOT of effort evaluating products before they choose what to buy… and then buy a lot of the best one.
Setting expectations low may set you up to pleasantly surprise the customer when they look at actual silicon results, and certainly not surprise them if the results favor AMD.
Intel is doing a lot of work on getting the message out regarding which workloads will excel on Intel silicon. I was surprised to learn there are some workloads that are 50% faster on the crappy old Intel Ice Lake server than the newer AMD Milan server.
In some sense, Intel can stay in the building by helping the customers choose the right product, even if it is from a competitor. I am always pleasantly surprised when a store recommends I go someplace else to buy a particular product as I am likely to return when they have something I want. Granted, it is a tricky proposition and we will have to wait and see how it plays out. AMD has had very good server products for almost 5 years now, but is still at only 15% market share. Intel publicly stated many years ago their goal was to hold AMD to 20%, which is a goal I think they are now giving up on.
Alan
I don’t see AMD doing anything different here; they already have this information from impartial review sources.
Um, they will be doing something very different for the next few years. Zen 4 was not anything difficult or different, but AMD has had to get the mobo and memory makers lined up. Mobos should not have been a significant problem, but AMD had gotten used to compatible AM4 mobos. Even DDR3 to DDR4 was not a major redesign. I won’t go into the differences in DDR5 memory, but it turned out to be a major problem when it came to mobo design. In effect this is why AMD and Intel are basically at parity on DDR5–they share the same mobo makers. (Intel decided to go the extra mile and support DDR4 and DDR5 with Alder Lake. That meant that developers of all flavors could use DDR4 motherboards until it was time to provide benchmark results.
Anyway, if you want to put together a Zen 4 Raphael system, fine, wonderful. But if you don’t want a lot of arrows in your back, wait until after Thanksgiving or better after Christmas. Otherwise, I suspect you will be waiting for a non-beta AGESA for your chosen motherboard. For those who have not had the pleasure of working on the bleeding edge, the AGESA is the AMD part of the motherboard BIOS that does things like setting memory timings. If you don’t have a good entry for your new memory in the AGESA, you can run the memory at “standard” timings, which today is likely 2133 instead of 3600. Then spend a couple of days changing one or more settings. See if the system runs stably at the speed you expect. Wash, rinse, repeat. (DDR5 speeds will be more like 4800 bog standard, 6400 overclocked per DRAM vendor specs. This article will probably convince you that you never want to hand-tune DDR5 memory: https://www.tomshardware.com/how-to/overclock-ddr5-ram)
Like I said a few weeks back. This is Intel’s Titanic moment.
Gelsinger’s statement was the official sounding of the klaxon’s for the upcoming impact which will be the next 2 years.
The start of the impact will be the canceling of the ARC consumer GPUs.
The ongoing impact will be Raptor’s lake failure to be anything more than competitive at twice the power draw and thermals, which will result in significant laptop share going to AMD.
The ongoing impact will also spread into the server market as Sapphire Rapids and then Emerald rapids win in only certain use cases and AMD continues to scoop up significant market share.
Keep in mind the earlier post by Physician quoting Lisa Su as stating AMD doesn’t expect supply constraints for for Zen 4.
I am not kidding about a Titanic moment. The most probable future timeline is looking very rough for Intel. I do expect Intel to survive the iceberg of the next two years but the question is in what shape will they be after the hit.
The start of the impact will be the canceling of the ARC consumer GPUs.
Intel has already canceled about 6 programs. The most visible was the Optane server memory. With the introduction of CXL on Sapphire Rapids it appears third party solutions can fill this need at much lower cost.
Intel is still at around 120,000 employees. I expect them to continue to slim down.
Alan
Marc,
I am not kidding about a Titanic moment. The most probable future timeline is looking very rough for Intel. I do expect Intel to survive the iceberg of the next two years but the question is in what shape will they be after the hit.
I’m not persuaded that it’s as dire as you portray, if only because Advanced Micro Devices (NYSE: AMD) does not seem to have access to enough capacity to supply the broad market for typical business and home computers – which undoubtedly represents well over half of the x86 market, so Intel (NASDAQ: INTC) can continue to turn a decent profit on the low end of the business. On the other hand, that segment of the market does not generate anywhere near the margin of the higher end so Intel will have to spend more severely constrained R&D resources very judiciously going forward.
Norm.
It’s easy to focus on the X86 piece of the business, but Intel is making a hard turn. They need about 2 factories to meet their X86 demand, which is services by the Israel and Arizona factories running the Intel_7 process. The newest factory Intel has in production was built about 10 years ago. I have historically said about three factories, but Intel has given the market for one of those to AMD.
Intel currently has six leading edge factories in various stages of construction, and two more on the drawing board for about $100B in capital expansion. Intel has mitigated the huge risk this capital expansion creates through both the chips act and the deal with Brookfield capital.
It is easy to see a mess based on the issues with the current Xeon products and roadmap, however if the foundry business is able (unable?) to fill up those factories the Xeon problem will seem somewhat small.
Alan
I was curious where Intel’s margins have been trending
CCG - Client Computing Group
Q2 2021 - $10.3B rev - 39% op margins
Q2 2022 - $7.7B rev - 14% op margins
Datacenter
Q1 2021 - $5.5B rev - 38% op margins
Q1 2022 - $4.6B rev - 5% op margins
Note, I could not find gross margins by segment, I didn’t look too hard. Either we are about to have a significant chip market downturn, or Intel is getting kicked in the nuts.
Either we are about to have a significant chip market downturn, or Intel is getting kicked in the nuts.
I think it is both.
Alan