I do believe a good government policy can make a difference in general. But there is almost always a more effective or efficient mechanism than government interventions. For example, the U.S. CHIPS Act of the Biden administration is still far from making your tax money work as it promised. According to the Financial Times (FT), the $84 billion worth of projects were delayed or paused. This is 7% of the $228 billion across 114 larger projects at least $100 million tracked by FT.
For example, TSMC’s Arizona 2nd fab will be delayed to 2027–28 depending on how much of the government’s incentives are received, according to the chairman of TSMC in January 2024. The new fab may produce 3nm chips depending on the demand of regional customers. Sounds OK, right? Only TSMC has started producing 3nm chips since 2022. By 2027, TSMC USA will still be 5 years late, or 2 generations behind. But better than never, right?
There is no doubt the U.S. government can nudge with so much money at its disposal. Intel foundry’s business may have a chance to retake the №1 spot from TSMC. At least for the name of “Make in the USA”. Only if Intel foundry can proper execute 5N4Y strategy. It may better focus if Intel#2 is an independent company instead of a business unit.
A critical caveat is first productionalizing a new node does not mean winging the business. TSMC has been eating Intel & Samsung’s lunches because of the yield & chip performance on the similar node even when TSMC is not the first. Samsung started shipping 3nm chips in Jun. 2022. Whereas, TSMC only started that in Dec. 2022. In 2Q24, 3nm chips contributed to 15% of TSMC’s revenue. BTW, TSMC is still Apple’s favorite chip maker.
OK even if CHIPS act can survive the 2024 presidential election, we should still ask: how might the money, resources, and effort be better used to “lower costs, create jobs, strengthen supply chains, and counter China”?
