Intel 16 customers are of under-rated importance. Intel needs DUV external contracts to keep those fully depreciated DUV machines occupied as they move most of their internal chips over to EUVIntel 16 is their cheaper process based on old tech. 16nm, not 16 Ångstrom!
Not cancel per se, but Falcon Shores has been relegated to an internal chip with no intent to release to market. The goal is to focus on Jaguar Shores instead, which will be designed as a whole rack platform to better compete with other AI-dedicated server solutions. I honestly think it will be too little, too late. Who knows how long it will take for Jaguar Shores to be ready... and by then the key market players will have settled in.
Fully integrated rack isn't that important in launching a product, its only good for scaling deployments, just shows they have no confidence in Falcon Shores. Who knows when they will have Jaguar Shores out.Not cancel per se, but Falcon Shores has been relegated to an internal chip with no intent to release to market. The goal is to focus on Jaguar Shores instead, which will be designed as a whole rack platform to better compete with other AI-dedicated server solutions. I honestly think it will be too little, too late. Who knows how long it will take for Jaguar Shores to be ready... and by then the key market players will have settled in.
Yea but this also depends on how much customers they have for older processes. It's all smoke and mirrors as claiming Intel 4 had external customers. Yes they did, with Ericsson, which is chump change for Intel.Intel 16 customers are of under-rated importance. Intel needs DUV external contracts to keep those fully depreciated DUV machines occupied as they move most of their internal chips over to EUV
Falcon Shores should have not been cancelled in the original hybrid form, so one critical explanation is that it was bad enough that it wouldn't have found success even if launched in the original format.Falcon Shores was supposed to be a reworked product to better compete in the market, so yeah, I agree with your assessment.
they forgot to do baby steps first.CWF delay is cruel, hybrid bonding is hard.
Precisely, in no way should you make a beyond-ret 3D package without starting on simple stacks, just as AMD did with Milan-X/X3D before MI300.they forgot to do baby steps first.
it's like the anti-AMD mentality.
More like anti-TSMC mentality. Intel is behind, so they need to take aggressive steps to catch up. But an aggressive roadmap is also a risky roadmap, and in the world of semiconductors where you play Russian Roulette and only find out if you blew your brains out many years after you pulled the trigger, the odds ain’t good.they forgot to do baby steps first.
it's like the anti-AMD mentality.
This thread. Specifically what @Hitman928 posted.Er, what are you reading? They did a $128M loss this quarter.
It was 6.79B according to Intel Corporation Consolidated Statements of Income and Other Information . And even If I ignored It the operating loss would still be 4.7B, a big difference from a year ago when they still made 93M from operations.GAAP shows the 18.7B restructuring charges from Q3 2024 hence massive loss. Non-GAAP without that is 600M loss.
DCAI is so telling:
I can't shake off the impression they have too much pride to just copy the good AMD stuff, even if it would help them get more competitive faster.they forgot to do baby steps first.
it's like the anti-AMD mentality.
They're behind in nodes, big dawg, not really packaging.More like anti-TSMC mentality. Intel is behind, so they need to take aggressive steps to catch up.
They do copy the good stuff (reluctantly), DMR+1 is just a bootleg Venice.I can't shake off the impression they have too much pride to just copy the good AMD stuff, even if it would help them get more competitive faster.
Not an impression, it is exactly what's happening. As an organization, they are too proud. Their corporate culture to deal with defeat is to plan over-complicated schemes to "leapfrog" the competition that fail to hit their target most of the time. They are the perfect example of a decaying empire ruled by shortsighted elites who only experienced winning through the sheer force of their starting position (the glory of old). They never stop to look at the underdog in the mirror.I can't shake off the impression they have too much pride to just copy the good AMD stuff, even if it would help them get more competitive faster.
[...]
Intel gives off the impression of trying too hard to find some super awesome superior technology to leap-frog AMD
What kind of capacity problems do you expect in 2025?Seems like they are all in on PTL. They probably won't have capacity for CWF anyway in 2025.
... and my concern about Lunar Lake isn't about it being a good product, it is if Intel can make any money selling it.It won’t. Intel is thoroughly screwed. They've got a singular good product (Lunar Lake).
H2, and many here expected Q2/Q3. I think I am going to have to take a "I'll believe it when I see it" approach from here on out.It was end of 2025 previously?
Intel's "Tick - Tock" approach was well founded IMO. It was based on a steady, incremental, change one thing at a time approach.More like anti-TSMC mentality. Intel is behind, so they need to take aggressive steps to catch up. But an aggressive roadmap is also a risky roadmap, and in the world of semiconductors where you play Russian Roulette and only find out if you blew your brains out many years after you pulled the trigger, the odds ain’t good.
It won’t. Intel is thoroughly screwed. They've got a singular good product (Lunar Lake).
Did AMD actually announce any of this? I thought Zen6 2027 was a rumor.Everyone was so disappointed by AMD announcing the late introduction of Zen 6 and the relatively small single core performance improvements expected (although core counts could double).... but this may well be the future for all companies.
When did that happen?Everyone was so disappointed by AMD announcing the late introduction of Zen 6
Average Joe doesn't check which product is better. He just buys a laptop or PC.Still isn't stopping them from making $8B in the quarter.
Intel's strategy in the past few years was to flood the channel with chips and flush AMD out. With their increased reliance on TSMC and their dire financial situation, this is no longer sustainable: they either sell the new gen chips at reasonable prices or they lower their volumes. They can still flood the market with Raptor Lake, although this can also backfire and eat into MTL/ARL sales - where the real source of healthy revenue should be. That's the catch with using TSMC to achieve node parity, it exposes any weakness in product design and removes pricing flexibility.Average Joe doesn't check which product is better. He just buys a laptop or PC.
Then there is still the fact Laptops or PCs with Intel are in greater number.
There's also the question of whether the market will continue to absorb Intel 7 products at the same rate.They can still flood the market with Raptor Lake, although this can also backfire and eat into MTL/ARL sales - where the real source of healthy revenue should be. That's the catch with using TSMC to achieve node parity, it exposes any weakness in product design and removes pricing flexibility.
It's also the question of which manufacturer would replace that volume. AMD aims mainly for premium and Qualcomm too.There's also the question of whether the market will continue to absorb Intel 7 products at the same rate.