You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
It would require TSMC to have issues as well TSMC only passed Intel as a tech leader due to Intel fumbling the ball for like 3-4 years and stagnating they can have tech leadership back imo but not domination like TSMC.
Fab process development and chip manufacturing aren't my in fields of expertise, but I'm having trouble believing that TSMC has to stumble for Intel to meet or exceed them. I also don't believe that the only reason TSMC passed Intel in process development because Intel stumbled. Would Intel have been the best foundry for the iPhone even if 10nm was on time and met objectives? I doubt it.
I think this is the least possible option, because I can't see Intel giving up the x86 ISA, and I think it will take a leap beyond x86 to achieve this objective.
Well, at least Intel did not sue WJL when he left Intel for TSMC. Also, I believe Intel has (or, at least, had) a policy that those, who left the company but decided to come back within 3 months, get fully restored in their seniority and related benefits.
I was asking for actual evidence of this supposed TSMC excessive restriction on labour mobility. Where is it ? How is TSMC worse than anyone else ? Not liking TSMC isn't evidence.
But here's a problem: every generation of the M series in the past has been single-process. Unless the future M series shifts to chiplet. If this is true as stated, Intel handles the low-end M, TSM handles Pro, Max, and Ultra. This means Intel could theoretically also make high-end ones if yields permit. Because it's agnostic, just like Intel's current CPU core tiles.
I was asking for actual evidence of this supposed TSMC excessive restriction on labour mobility. Where is it ? How is TSMC worse than anyone else ? Not liking TSMC isn't evidence.
WJL moving from Intel to TSMC - no lawsuit
WJL moving from TSMC to Intel - lawsuit
It is sort of evidence. Obviously, the are some nuances but still...
Intel has never sued any of its executives who left for another company. TSMC did it and not once either.
I guess that’s a good excuse for not breaking even, as making IFS viable could be beneficial to the valuation. I remember DZ saying on multiple occasions that a properly run fab company should be valued at 1–2× its book value.
Should we focus on future cash flows instead of break-even timing?
Most the finances (Margins, Cash flow, NPV) for IFS are bad until they get >1 Fab worth of external customers paying full market price. 2029 timeframe.
Until then, IF IFS is sold, someone will be paid to take it (unless everything is written down like the DRAM companies did 10+ years ago). And IFS will have lost >60B in cash and net income since 2021.
WJL moving from Intel to TSMC - no lawsuit
WJL moving from TSMC to Intel - lawsuit
It is sort of evidence. Obviously, the are some nuances but still...
Intel has never sued any of its executives who left for another company. TSMC did it and not once either.
After a few discussions, It looks like the issue is all about TSMC Taiwan and Wei Jen. They think he took files and downloads (lots of people leave companies and do not take files.). They are legally warning Intel about this (this is a pretty common next step by lawyers). LBT is saying they will not use IP (I dont think they will).
All of this was avoidable by Intel ... very strange. And now LBT has doubled down on supported Wei Jen.
Most the finances (Margins, Cash flow, NPV) for IFS are bad until they get >1 Fab worth of external customers paying full market price. 2029 timeframe.
Until then, IF IFS is sold, someone will be paid to take it (unless everything is written down like the DRAM companies did 10+ years ago). And IFS will have lost >60B in cash and net income since 2021.
After a few discussions, It looks like the issue is all about TSMC Taiwan and Wei Jen. They think he took files and downloads (lots of people leave companies and do not take files.). They are legally warning Intel about this (this is a pretty common next step by lawyers). LBT is saying they will not use IP (I dont think they will).
All of this was avoidable by Intel ... very strange. And now LBT has doubled down on supported Wei Jen.
According to Daniel's article, all Wei Jen took was his personal notes:
"The legal issues stem from Lo's alleged deception regarding his retirement announcement and the handling of his personal notes from TSMC."
So, at this point it is not clear if there was anything for Intel to avoid. It could be just TSMC abusing their influence/control over Taiwanese government.
Cadence was a far easier problem space. It looks to me like Intel has only three remotely potential vectors to reach a $1T market cap:
1. Become a foundry that is a peer-level competitor to TSMC. (The not-TSMC opportunity doesn't look like enough.)
2. Become the client and server CPU of choice, displacing AMD, Qualcomm, and cloud computing internal chips.
3. Develop leadership AI chip and system-level product lines which displace a portion of Nvidia's marketshare, and that of cloud computing internal chips.
All three are long shots. Intel is unlikely IMO to succeed in any of these options to get to $1T by 2030, especially option 1, which is my favorite.
