Intel making chips at TSMC makes sense for them, but not really for TSMC, if they can already fill out every wafer possible for existing customers who compete with Intel they have little reason to help Intel both because it won't make them more money and it will make life more difficult for all their long term partners who with more competitive chips will steal sales from Intel. Also Intel have repeatedly tried (and failed) to bring on customers and compete as a foundry business (more to utilise their under utilised capacity than to be a genuine player though).
Intel are just in such a weird place, if 10nm can't work they have a bunch of insanely expensive fabs, billions in R&D and AMD without any of that cost is producing 7nm pretty much a year before they can produce something similar.
They still have the new completely unused 8billion fab... wherever they built it, being wasted. It was intended for 14nm but they struggled to fill capacity at other fabs so never spent the money to fill it with equipment as it wasn't worth it. They planned to fill it with 10nm equipment when 10nm was ready..... and that has meant it sits there even longer waiting to be used.
What senior guy at Intel can say, you know that 8 billion fab we built they decided to only use with the new process.... well we want to go use TSMC instead. Even if it's the right decision that's the kind of thing no one wants to say and no board member wants to hear.
You can't proclaim to be the best foundry company in the business, with the best nodes offering incomparable chips then next generation start using your rival for major production of your most important chips. They do actually use TSMC as they made modems there for quite a while though the latest 5g stuff is supposed to be being made on Intel 14nm. It's one thing to make a special RF based chip on a process better optimised for RF, it's another to make basically your entire CPU lineup at TSMC, the very chips your industry leading process is supposed to do better.
But again, even if they wanted to, even if someone brought that to the board, at this point why would TSMC do it.
At this point Intel's best shot might be the Global 14nm plan. As in, licence 7nm off TSMC or global and fill up that unused fab with equipment and pay through the teeth to have TSMC guys come and get it ramped up asap. Then pay them 5billion to keep the deal silent and just say they got it working finally
In terms of faith I have in Intel's current 10nm claims, lol. As you say I think end of 1h, if there are realistic rumours that Intel have finally got okay yields going then I can just about believe a production ramp could have half decent volume moving into early 2020, but until they get something going and there are rumours of decent enough yields for a full ramp then everything is just a guess/best hope for them. If end of h1 or early h2 start bringing rumours of more issues then that xmas 2019 date will slip yet again.
