Semiconductor Capacity Construction Costs
The fab is expected to cost between $23.35b and $31.14b over the next few years as the fab is fully built out, and the company has enough land at the site to build three additional fabs (P4 – P6) with a combination of NAND, DRAM, and foundry capacity going forward, however it seems Samsung is facing the same inflationary cost increases that we as consumers have been facing, despite the fact that Samsung affiliates are some of the major suppliers to the project. Samsung C&T (028260.KS) has increased its contract price by 34.4% against the original contract, Samsung Engineering (028050.KS) has raised its contract price by 40.0% and Samsung Heavy Industries (010140.KS) has raised its contract price by 24.1%, putting the increase for just these three affiliates up by 35.1% in aggregate, or an additional $778.5m US, with all three citing increases in raw materials and labor costs as the reason behind the increases.
While the construction and the equipment installation continues, some of the ongoing contracts have been pushed out by as much as a year, which we expect means that Samsung is focusing on finishing specific P3 projects and delaying the build-out of others in this large fab. Intel (INTC) is also facing a similar situation as it plans out its new fab in Magdeburg, Germany, as the cost has risen from 18.07b US to 21.26b US, a 17.65% increase, with the company indicating that it has delayed the start of construction for the project, which was to begin in the 1st half of 2023, even with the $7.23b promised by the German government toward the project.
While none of this is good for the global economy, it does have one positive outcome, and that is to slow the capacity increases that have been planned for the semiconductor space, which we expect would cause an over-supply situation in 2024 and 2025. While the delays will inevitably move back into the queue as overall consumer demand begins to stabilize, at least the build-outs will be a bit more staggered, although China is still a wild card that will likely see less semiconductor planning changes as the battle with the US government over semiconductor equipment continues. China has little choice but to build capacity as it sees both external supply and demand decrease. While China might not be able to build advanced node capacity, they can add more mature nodes for internal consumption and wait until the global climate is ready to allow them back, at which point we expect they will dominate the generic semiconductor space, as they have done in both the LED and display industries.