Sabitlenmiş Tweet
Journal of a technologist. Daily musings and ramblings, mostly on semiconductors, aviation and general chaos in life.
English
Salvor Hardin
1.9K posts
Salvor Hardin
@mayorhardin
Dabble in nanometers, thrive in Gbps
Terminus Katılım Mart 2025
123 Takip Edilen150 Takipçiler
@rennyzucker True but their latest statement was they will hit their year end goals by end of Q2. Assuming that was a 70% functional yield, then it does mean it’s stable at this point. However, yield is a distribution. So 70% is the mean with tail end yield being their limiter.
English
@mayorhardin That is all factually correct. Whether or not one can take David at face value anymore is up to them.
I personally think it’s a stretch to say 18A is currently stable considering they said 6%+ monthly yield gains several months ago though.
But valid to view that optimistically
English
I don’t understand this type of guy.
Intel doesn’t currently have a leading edge foundry… they don’t even really have an HVM internal fab… there’s one product in actual risk production on 18A and it’s a ridiculously priced laptop SoC. Outs are what 3-5kwpm? At 65% yields?
Adam Watson@WatsonAdam42920
@rennyzucker Intel has the process IP, process engineers, chemical engineers, industrial engineers, chemical supply chain, process tools, fab complexes, land, etc…. Let me know if you see Apple post jobs reqs and starts buying IP.
English
@rennyzucker Look all I’m saying is, 18A is a much better process they’ve ramped in a decade. It will meet its wafer starts by end of this year. And Zinsner did say that 18A is going to be their workhorse. So even if they are working on 14A, 18A&P have shown signs of being stable.
English
@mayorhardin I totally agree that for a product to have meaningful SKUs that it needs to have a process that actually works.
We totally agree there.
English
@rennyzucker So there’s this thing called time and progress. What was true a year ago, might not be true a year later because progress happens.
English
@mayorhardin What he said originally before the White House invested 10bn
English
@rennyzucker It’s definitely not a million dice a year. If 18A isn’t HVM, you wouldn’t be seeing the SKUs PTL has, you won’t be seeing WCL launch later this year. None of those would actually happen. For any process to have meaningful skus, it needs to have a process that actually works.
English
@mayorhardin Good die? Good dice? Good dies sounds grammatically incorrect but English is weird (Latin and French are weirder)
English
@rennyzucker LBT’s first call was a year ago when 18A wasn’t HVM. LBT himself changed his opinion based on recent calls. So you can either stick to what happened a year ago or what they did say recently. Which one would that be?
English
@mayorhardin Nobody on the planet, not even Zinsner, thinks 18A itself can be profitable… it’s an audition for 14A. LBT basically said in his first call that if they can’t show the world that 18A works at massive scale that the company is dead
And he was right
English
@rennyzucker That’s exactly my point. 10nm didn’t work and we could see, 18A is doing the exact opposite of 10nm. If that’s not proof enough of 18A not being dog shit, what else do you want?
English
@mayorhardin Yes because Intel 7 worked out swimmingly we should totally comp it to that
English
@rennyzucker Substandard yield is definitely not true. Compare it with Intel’s 10nm where Icelake barely had any parts available. In fact ICL was only launched with limited skus. That is definitely not the case with 18A.
English
@mayorhardin As a process guy you should be able to infer this by my 3-5kwpm risk production at sub-standard yields comment.
Genuinely curious if you think that profiles as a viable foundry partner for external customers considering you made a living here.
English
@rennyzucker Every leading edge node was at the same position as 18A is. TSMC N7 didn’t hit peak wpm until 2020. It entered HVM in late 2018. 18A’s ramp is not the profit limiter. It’s the yield loss on Intel 3/4/7.
English
@mayorhardin 18A is a leading edge process obviously but it’s not in high volume manufacturing (HVM) and thus not a viable foundry process… which explains its multi-billion dollar quarterly loss making profile
English
@rennyzucker I am not saying this an an INTC stock guy but as someone who made a career in chip design.
English
@rennyzucker What do you think 18A is if it’s not leading edge?
English
@BenBajarin Pat Gelsinger used to talk about $100/share in his early days back at Intel. We are getting to see it now and it just feels right.
English
Genuinely didn't think I'd see this in 2026. $INTC north of $100.
English
@PatrickMoorhead More like change in reality, right? Because 2021 power crisis did happen!
English
Texas one of the solar leaders. Yup. Narrative violation.
Craig Lawrence@clawrence
Solar providing 68% of electricity in ERCOT right now is nuts. Crazy weather for May. Cool, dry.
English
In the era of AI, Intel will find it very difficult to hide its secrets...
English
