From all accounts wafer availability looks like its the constraining factor for AMD (and by implication a key factor in Intel's performance).
All roads lead to TSMC ....
All roads lead to TSMC ....
there's a need for another foundry like TSMC. Samsung is not cutting it.
I saw two articles that convinced me of this.
First, I saw that Germany asked Taiwan to ease the semiconductor shortage's impact on the auto industry.
https://www.reuters.com/article/us-taiwan-autos-chips/german...
Then, I saw that Germany is sending a warship to the South China Sea.
https://asia.nikkei.com/Politics/International-relations/Ind...
UMC is also exposed to geopolitical risk. Samsung too.
There's a lot of demand for semiconductor fabrication right now, and I don't think that's going away. Why concentrate all of the capex right on China's doorstep? I just don't see the wisdom of that approach.
This was never intentional. When TSMC was founded, their business model was nuts from a Western perspective: why would a company produce chips for someone else, the money is in the design and the brand, not the manufacturing. The market for this started when companies like Intel and TI wanted to derive some profits from unused manufacturing capacity.
The zeitgeist of American business over the past 40 or so years has been that brand is what people buy. Manufacturing is expensive, low-margin, and is best outsourced. Thus, there's been a slow, steady march towards design here, build somewhere else. Not just in the silicon space, but in everything from food processing to plastic tchotchkes.
That is a perfectly valid model, if there is a lot of competition in the manufacturing space. But now TSMC is so far ahead of the competition, and nobody expected that. Sometimes being the contrarian is where the profits are.
Yeah, that sounds like more of the Koolaid the business schools were selling.
Remember all the TV brands that were made in the US? Now they are bought and sold in China. Not the products, but the brand names.
What I'm referring to as "that approach" is the below suggestion from MangoCoffee, to whom I replied.
> there's a need for another foundry like TSMC.
this podcast with SemiWiki Founder Daniel Nenni. https://www.youtube.com/watch?v=O4DgXtxkZNg
good discussion on how we go from IDM (real man have fab) to the foundry business model.