Fabbing at TSMC is an embarrassing but great decision. The design side of Intel is the revenue/profit generating side. They can't let the failures of their foundries hold back their design side and leave them hopelessly behind AMD and ARM. Once they've regained their shares of the server market or at least stabilized it, they can start shifting some of their fabbing to Intel Foundry Services, who are going to really suck at the beginning. But no one else is going to take that chance on those foundries if not Intel's design side. The foundry side will need that stream of business while they work out their processes.