It took a stupid amount of money to get on the treadmill (TSMC--$200 million in 1987), takes a stupidly larger amount of money to go to the next layer of the treadmill ($100 billion+ over the next 3 years), and, if you fall off, you aren't getting back on the treadmill.
2) The lead times on the equipment are measured in years
As the podcast points out, TSMC has bought AMSL's production of machines for the next 5 years. Those are lithography machines that nobody can buy for any amount of money for the next 5 years.
To compete, you need to set up not just the fab but the infrastructure supplying the fab.
The podcast Acquired.fm had a recent episode covering TSMC’s history and strategy that I found really informative (and entertaining!): https://www.acquired.fm/episodes/tsmc
Any kind of research and development project is a calculated risk that the effort will yield a worthwhile result. All of the top contenders poured a lot of money and high-caliber talent in their next generation fab R&D and TSMC's approach paid off the best and Intel's (some say more aggressive but riskier) approach turned out to be a dead end causing them to have to scramble to recover. But it was always a gamble that could have turned out badly for any of them.
Apple uses TSMC, which means they get tons of revenue and sell tons of chips.
That turns out to be a key advantage when learning how to make newer chips.
Taiwan also has put a ton of money into this field. A lot of support from the country plus support from some of the biggest names in the industry leads to TSMC.
TSMC didn't always dominate. Intel was definitely better just 5 years ago. But Intel tripped up and GloFo gave up.
Not a good singular article but try searching for all fab-related keyword combos for the past year or two, eg: smic, tsmc, asml, china, trade, ban, etc.
I have a difficult time undestanding it myself but found the Asianometry YT channel being able to put things into the greater overall context on why things are the way they are.
on the high end, we have Intel, Samsung and TSMC. TSMC is the only foundry that is pure play. TSMC have no side business or any business that compete with its customers.
Apple/Samsung is the example of that dynamic. Apple leave Samsung after Samsung come out with their own smart phone that compete with Apple. their relationship have taken a nose dive after that.
TSMC win on trust. They don't compete with their customers. they are a foundry and that's all they do.
Samsung is a major supplier for Apple. SoC fabrication, screens, RAM, NAND. Samsung is a conglomerate of many companies. They’ll happily sell you components while competing in the smartphone market with you.
