Re: To little - far too late
"Intel by refusing to invest the required billions in EUV equipment etc has fallen well behind TSMC - the money that should have kept them competitive was spent in shareholder dividends."
This isn't accurate - Intel had a big lead in leading edge ASML equipment. Only the first generation EUV kit wasn't that great. Once the teething issues were resolved, it was TSMC's turn to get equipment from ASML and that meant they caught up. Intel should have still had the advantage at this point but they doubled down on 14nm fabs to cover the short term and waited for more ASML kit. When they had parity with TSMC from an equipment perspective, their 10nm was still failing to deliver at required performance or quality levels. TSMC took a risk and invested in new fabs and lithography equipment leaving them with around 66% of currently installed EUV-capable equipment although that will have dropped to under 50% by the end of 2021 as Intels 7nm fabs (hopefully) start to produce products.
The big issue with leading edge fab capacity is that it takes around a year to finance and order equipment/get approval to build a fab and then 18 months to build it fit it out with equipment (ASML capacity allowing) and 6 months to get products shipping. Intels issues in 2021 are from complacency/being unwilling to acknowledge problems with 10nm in 2017/2018. These were compounded by TSMC doubling down on investment which lengthened queues for equipment.
Biting the hand that feeds IT
