Political Money

When politicians want to boost their countries’ chip industry  there’s not much they can do except throw money at it.

The sort of money politicians like to throw at the chip industry is big money and the only thing the chip industry can do with big money is build fabs.

The problem with that is that a fab is more of a liability  than an asset if it can’t be filled. Running a fab at at a low utilisation rate is a fast way to go bust.

So it’s a positive that the US Chips Act’s $300 billion is not achieving much.

In May 2020 TSMC announced plans to build a fab in Arizona with construction starting in 2021 and production on 4nm scheduled to start in 2024. Production has now been put back to 2025.

The second fab was announced in December 2020, with production on 2nm scheduled for 2025. Production has now been put back to 2028.

A third TSMC fab for Arizona was announced but no date for the start of  construction was set though production there is now slated for the late 2030s.

Intel planned to build two fabs in Ohio scheduled to start production  in 2025. Now, completion of the construction has been postponed to 2026–2027 and production postponed to around 2027–2028.

Intel also planned to spend €30 billion on two fabs in Magdeburg Germany with construction set to start in 2023, This has now been put back to 2025.

Samsung planned to build two fabs in Taylor, Texas and started construction in 2022 with initial production set for 2024 but it is now being said that production will not start until 2026.

These are all to be advanced node fabs and presumably the companies involved know they wouldn’t be able to fill them if they got completed on the original schedules.

If companies know they can fill a fab, they’ll build one, so political money will make little difference as to whether a fab gets built or not.

Where political money can be usefully spent in the chip industry is in improving process technology. This is what the Japanese did in its 1970’s VLSI project, Sematech did in the 80s and the EU did with the Jessi and Medea projects in the 80s and 90s.

Getting ahead with the tech is always a profitable thing to do in the chip business and giving political money to the R&D people and the manufacturing equipment people usually delivers a good return.