Vietnam is planning to add to the non China based chip supply. Vietnam has a plan to build six fabs and 20 packaging and testing facilities by 2050. Which is certainly interesting since a good amount of other countries from the United States to Germany are planning to add more fabs as well. While demand for chips that can assist in the ever increasing AI workload is skyrocketing, you have to wonder at what point the market will become saturated and just how competitive Vietnam will be.
There’s also a question of which chips all these new fabs that are in the works will be making. Not all of them are for AI, but AI is the current corporate buzzword just like “the cloud” was ten plus years ago… so you’ve got to expect a number of them will advertise themselves as making AI chips.
The good thing though is that Vietnam’s time horizon is large. Their goal is to be a leader in semiconductor manufacturing by 2050. 25 years to achieve their goal is a good chunk of time and is much more reasonable than say by 2027. There are however a couple of issues…
In Phase 1 (2024 to 2030) of Vietnam’s plan they want a small fab built by 2030 and 50,000 engineers and university grads available. That is a stretch. While it’s easier to build a small fab than it is a large one, it takes a significant amount of time. It will also take a good amount of time to get both their supply chain set up and to even get the manufacturing equipment. I don’t know if 2030 is doable. I also wonder at the economics of having a small fab, is it going to be profitable or is it going to run on state support? If it does run on state support will its products be subject to tariffs for unfair competition? You’ve got to wonder.
Then we move on to Phase 2 (2030 to 2040). In Phase 2 Vietnam wants two more fabs, 15 packaging and testing facilities, and a whopping 50,000 to 100,000 engineers and university graduates. We’re now getting to pie in the sky levels. 100,000 people is the entire workforce of Intel. You can’t just look at this as a Vietnam thing too… you have to imagine all the other countries also adding to their semiconductor workforces. Taiwan, South Korea, and the United States will all be also adding to the current workforce. Is the world going to add too much capacity and too many workers? Like I mentioned demand for AI chips is astronomical but will that demand continue into the future? That’s certainly a complex question and I wonder if Vietnam has asked itself that.
In Phase 3 (2040 to 2050) Vietnam wants three more fabs with 20 packaging and test facilities. Which is doable. It’s 25 years out, plenty of time to build and train people. But again, will Vietnam be competitive? Their workforce will become with time more costly just as China’s did. The advantages Vietnam has now are likely to erode as it continues to develop. You ultimately can’t just keep paying people pennies for working all day everyday. And high tech jobs are going to require increased pay no matter what, see India.
And demand for chips… god only knows where demand for chips will be in 5 years let alone 25 years. And which chips, the mixture of different types of chips, is even more uncertain. NPUs, GPUs, TPUs, CPUs… even specialist processors for self driving cars! Which chips are going to be in demand and which aren’t. And how many!
So you have an ambitious plan in an uncertain economic / future environment. That’s not the best situation.
You also have to wonder about the political situation. Vietnam is a communist country and while it has a history with China it’s hard to imagine Vietnam supporting Taiwan and the United States in any conflict. Its manufacturing capacity may very well end up supplying China during a conflict. That most likely means a large number of countries will forbid Vietnamese manufactured products. Again Vietnam faces an uncertain future with an ambitious plan.
Who knows though, Vietnam might just pull their plan off.