Putting STM on a wafer basis, and saying plainly that the headline claim survives only in softer form, is what made the rest of the map credible for me. One date your 2028 arithmetic runs into first: Tower's May filing backs the 2027 contracts with a larger wafer commitment for 2028, prepayments due by January 2027. That's customers paying up front for the year your base case says risks overcapacity. If it lands on schedule, do you read that as the inputs changing at 3.2T, or as insurance rather than demand?
Interesting that the scarce asset is the capacity to fabricate the PIC, and the wafer under it. I'd be curious whether packaging and bonding know-how is as scarce as wafer capacity. In other industries the unglamorous steps, like bonding and test, are often where know-how quietly accumulates.
Putting STM on a wafer basis, and saying plainly that the headline claim survives only in softer form, is what made the rest of the map credible for me. One date your 2028 arithmetic runs into first: Tower's May filing backs the 2027 contracts with a larger wafer commitment for 2028, prepayments due by January 2027. That's customers paying up front for the year your base case says risks overcapacity. If it lands on schedule, do you read that as the inputs changing at 3.2T, or as insurance rather than demand?
Interesting that the scarce asset is the capacity to fabricate the PIC, and the wafer under it. I'd be curious whether packaging and bonding know-how is as scarce as wafer capacity. In other industries the unglamorous steps, like bonding and test, are often where know-how quietly accumulates.
https://semiconductor-today.com/news_items/2026/sep/aeluma-scat-150926.shtml
This could be an interesting development in relation to InP