As pointed out elsewhere, large US chip manufacturers have spent more than $20 billion in stock buybacks. This lack of investment in infrastructure can pretty much be laid at their feet. However, this is still our problem.
(As an aside, let's say we make stock buybacks illegal. What will happen is simply that companies will issue dividends instead. While this removes the ethical problems of stock buybacks, it doesn't change the calculus of internal investment versus investor return.)
Meanwhile, small chip manufacturers (I work with several small fabs), really struggle to get loans, investment, access to hardware, and knowledge of modern processes. There are tools and techniques that I need to use that are common in Asia, not uncommon in Europe, and very rare here. It's not always that it requires more capital. In many cases, the tools used in Asia are better and cheaper than the tools used for "standard" processes in the US. Why is that?
There are PLENTY of people in the US who know how to fab chips. However, almost all of them have been trained in university clean rooms stocked with equipment from a fabrication node that is 40 years out of date (100 mm wafers, 1 micron node). Industry and academic clean rooms (Stanford, Cornell, Georgia Tech... there are ton of them) speak different languages. Because the exchange between industry and academics has generally broken down, we have professors teaching today who have only used fab technology based on that 1980 node, and who were themselves taught by people who only knew technology up to that node. So that node has cemented itself into place. Until recently, you could use old 150 mm lines to bridge that gap when you get out into the real world, but those lines are almost all shut down now. The difference in tooling between 200 mm lines (the current cheap, fully paid for fab lines) and the 100 mm lines universities train people on is tremendous. The philosophy of how to approach fabrication is different. Intel and Qualcomm can afford to retrain everyone, but it's hard for small fabs to find people who know what they're doing. If there's an area the government has failed, it's in managing the grants and programs that were supposed to update university facilities over the last 20 years. We got a lot of newly made equipment for an out of date node.
There are really three things that we're trying to accomplish around chip fab from a government point of view. One, we want to perform chip fab in an environmentally sustainable way. Two, we want to treat workers with respect, including reasonable wages. Three, we want to protect American jobs in this area. We have failed at all three.
Instead of handing over billions to big chip companies, we should institute 0% loan programs for people building and running smaller fabs, tooling companies, and consumable companies in the US (with all the conditions you want on that). We need to take the university facilities programs away from everyone involved in NNCI (National Nanotechnology Coordinated Infrastructure) and get it in the hands of someone who's run a professional modern cleanroom, and we need to hold the universities responsible to prioritize education and training in this space over research with those facilities (there is SO MUCH industry driven research). Instead of tariffs, we need to outright ban materials made in ways that violate our environmental protection laws or labor practices. That needs to be the last step, or we're simply hurting every other part of our economy.
None of those things involve handing money over to the large companies.