I thought I was buying the dip for SiTime in the deep seek bloodbath. Then I thought I was buying the dip dip this morning. Tomorrow I will buy the dip dip dip.
I agree that Clearwater Forest being pushed out is a huge red flag. If nothing else was wrong with Intel and even Pat was still around, this would tank the stock. After their history of never keeping any dates for the introduction of a new process I'm not sure with what face they can say nothing here to see folks, trust us, 18A is fine ... <Cue the laugh track of all their so called potential fab customers>
Caveat emptor - I'm one of those who bet on AMD crushing Intel when AMD was at 10$, and never completely exited that position (exited 80% of it, oh well), so I might be stuck with perpetual optimism and confusing all of AMD's competitors with Intel... Oh well
I don't understand your take on AMD's MI-line. Contrary to earlier reports, AMD's MI-sales didn't crater in Q4 2024 (you, Dylan and others predicted this) in response to Nvidia's Blackwell. Instead it's at / close to ATH, and Lisa confirmed that it will stay at this level through 1H2025. Lisa also said that 2H2025 will see a greater level of MI-revenue vs. 1H2025, and that we can expect 10b+ MI-revenue starting 2026. I know her to be quite even with her predictions, more towards being understated.
Revenue data for MI-series per TheNextPlatform:
Q4 2023 $415 M
Q1 2024 $610 M
Q2 2024 $1,022 M
Q3 2024 $1,647 M
Q4 2024 $1,746 M
It's not Nvidia level, but I see it as a working money engine and a platform that will get better, and that will have exposure from the edge (assuming UDNA-everywhere), to client and to the datacenter, alleviating some of the lacking scale of software.
I'm more worried about what will happen to X86, will they be able to collaborate with Intel on keeping it competitive, what decisions AMD will make for itself on that, Nvidia's impact with a client APU for Windows -- but that's probably me being stuck in the old paradigm.
Reading through the transcript from the Q&A - I only listened through it - I see that she's actually saying 20b+ of MI-revenue for 2026, or at least at run rate of 20b+ starting Q4 2026.
"Without guiding for a specific number in 2025, one of the comments that we made is we see this business growing to tens of billions, as we go through the next couple of years."
AMD's Instinct sales benefited significantly from the delays at Nvidia in getting their new generation of accelerators shipshape. AFAIK, Nvidia ran into greater problems with the interconnects than anticipated, and a number of key customers (example: Microsoft) then upped their order numbers for Instincts, which AMD was able to fill quickly. The question is now if that somewhat lucky streak will continue now that Nvidia has solved the teething problems of their next generation, and can ship units in quantity.
In some way, Intel not throwing a lot of resources to get their packaging solutions (CoWos, EMIB) up to speed for Clearwater Forest might well be the smarter move. Clearwater Forest will/would have a hard time competing with AMDs many-cores Zen 5c EPYCs in the x86 space. In contrast, client SoCs, especially for laptops, are and will have to be Intel's profit center for the time being. Getting the production of the tiles in Intel 18 and the packaging of these tiles into SoCs ready makes a lot of sense for now.
In the meantime, it is probably high time to force IFS to be rescued, that is formally separated from the rest of Intel. IFS is the only foundry operation in the US that has reasonably state-of-the-art EUV nodes either running or in advanced preparation. Those figleave EUV fabs in the US from TSMC or Samsung don't count or are years away from production. It comes down to national security. The current wrecking crew a.k.a. Intel's Board cannot be trusted to steer the ship that is Intel and IFS clear of the icebergs, as they seem to be intent to hitting as many of them as possible. If Intel's x86 operations would die, it would be sad. If the foundry services die with it, it'd be a national calamity.
Korea is currently abuzz about glass substrates. With Samsung Electronics deciding to handle the glass substrate business in-house instead of through an affiliate, AMD planning to adopt glass substrates in 2028, and Apple reportedly showing interest in glass substrates, would you consider covering this topic?
Glass as a substrate isn't just a buzz in South Korea; it has been "hot" for the last year(s). The question I have is who are the leading players in that field, from foundries to key suppliers? One of them was or is , apparently, Intel (https://www.intel.com/content/www/us/en/newsroom/news/intel-unveils-industry-leading-glass-substrates.html#gs.5vk7g8). If that has continued, could be one of the gold nuggets that should be rescued from the current dumpster fire there.
Samsung going it alone would make sense (in theory), as that Chaebol has parts that work in specialty materials. The caveat of "(in theory )" is there as Samsung's different parts don't have a great track record on getting their joint act together, and they have made flashy announcements like this that seem to serve more as a smoke screen to distract from their problems.
I believe there were some presentations at meetings like "Hot Chips" in recent years on glass substrates, maybe someone here knows more.
I thought I was buying the dip for SiTime in the deep seek bloodbath. Then I thought I was buying the dip dip this morning. Tomorrow I will buy the dip dip dip.
keep up the excellent work sir 'keet
I agree that Clearwater Forest being pushed out is a huge red flag. If nothing else was wrong with Intel and even Pat was still around, this would tank the stock. After their history of never keeping any dates for the introduction of a new process I'm not sure with what face they can say nothing here to see folks, trust us, 18A is fine ... <Cue the laugh track of all their so called potential fab customers>
Good write! Thanks!
Thank you for a great article, as always!
Caveat emptor - I'm one of those who bet on AMD crushing Intel when AMD was at 10$, and never completely exited that position (exited 80% of it, oh well), so I might be stuck with perpetual optimism and confusing all of AMD's competitors with Intel... Oh well
I don't understand your take on AMD's MI-line. Contrary to earlier reports, AMD's MI-sales didn't crater in Q4 2024 (you, Dylan and others predicted this) in response to Nvidia's Blackwell. Instead it's at / close to ATH, and Lisa confirmed that it will stay at this level through 1H2025. Lisa also said that 2H2025 will see a greater level of MI-revenue vs. 1H2025, and that we can expect 10b+ MI-revenue starting 2026. I know her to be quite even with her predictions, more towards being understated.
Revenue data for MI-series per TheNextPlatform:
Q4 2023 $415 M
Q1 2024 $610 M
Q2 2024 $1,022 M
Q3 2024 $1,647 M
Q4 2024 $1,746 M
It's not Nvidia level, but I see it as a working money engine and a platform that will get better, and that will have exposure from the edge (assuming UDNA-everywhere), to client and to the datacenter, alleviating some of the lacking scale of software.
I'm more worried about what will happen to X86, will they be able to collaborate with Intel on keeping it competitive, what decisions AMD will make for itself on that, Nvidia's impact with a client APU for Windows -- but that's probably me being stuck in the old paradigm.
Reading through the transcript from the Q&A - I only listened through it - I see that she's actually saying 20b+ of MI-revenue for 2026, or at least at run rate of 20b+ starting Q4 2026.
"Without guiding for a specific number in 2025, one of the comments that we made is we see this business growing to tens of billions, as we go through the next couple of years."
I think it will not happen until the arrival of UA Link, UDNA rack level systems and improved software stack until 2026.
AMD's Instinct sales benefited significantly from the delays at Nvidia in getting their new generation of accelerators shipshape. AFAIK, Nvidia ran into greater problems with the interconnects than anticipated, and a number of key customers (example: Microsoft) then upped their order numbers for Instincts, which AMD was able to fill quickly. The question is now if that somewhat lucky streak will continue now that Nvidia has solved the teething problems of their next generation, and can ship units in quantity.
In some way, Intel not throwing a lot of resources to get their packaging solutions (CoWos, EMIB) up to speed for Clearwater Forest might well be the smarter move. Clearwater Forest will/would have a hard time competing with AMDs many-cores Zen 5c EPYCs in the x86 space. In contrast, client SoCs, especially for laptops, are and will have to be Intel's profit center for the time being. Getting the production of the tiles in Intel 18 and the packaging of these tiles into SoCs ready makes a lot of sense for now.
In the meantime, it is probably high time to force IFS to be rescued, that is formally separated from the rest of Intel. IFS is the only foundry operation in the US that has reasonably state-of-the-art EUV nodes either running or in advanced preparation. Those figleave EUV fabs in the US from TSMC or Samsung don't count or are years away from production. It comes down to national security. The current wrecking crew a.k.a. Intel's Board cannot be trusted to steer the ship that is Intel and IFS clear of the icebergs, as they seem to be intent to hitting as many of them as possible. If Intel's x86 operations would die, it would be sad. If the foundry services die with it, it'd be a national calamity.
And with that, I am off my soapbox.
Will Qualcomm a good stock to buy after started selling ARM CSS to hyperscalers??
I recommend to check out Claus Aasholm's most recent posting here on Substack regarding Intel's inventories. Interesting stuff!
I'm buying AMD. May be I'm a bit too early, but man they delivered on client and DC CPU side. X3D chips selling too good, also EPYC carried DC segment
It may go down to 95 short term, but if it goes higher it's a 180 stock. That's the dilemma for me, and it's an easy choice
Korea is currently abuzz about glass substrates. With Samsung Electronics deciding to handle the glass substrate business in-house instead of through an affiliate, AMD planning to adopt glass substrates in 2028, and Apple reportedly showing interest in glass substrates, would you consider covering this topic?
Glass as a substrate isn't just a buzz in South Korea; it has been "hot" for the last year(s). The question I have is who are the leading players in that field, from foundries to key suppliers? One of them was or is , apparently, Intel (https://www.intel.com/content/www/us/en/newsroom/news/intel-unveils-industry-leading-glass-substrates.html#gs.5vk7g8). If that has continued, could be one of the gold nuggets that should be rescued from the current dumpster fire there.
Samsung going it alone would make sense (in theory), as that Chaebol has parts that work in specialty materials. The caveat of "(in theory )" is there as Samsung's different parts don't have a great track record on getting their joint act together, and they have made flashy announcements like this that seem to serve more as a smoke screen to distract from their problems.
I believe there were some presentations at meetings like "Hot Chips" in recent years on glass substrates, maybe someone here knows more.
This probably isn’t his field.