On sells a variety of components -- some are completely commodity and are produced in super cheap fabs in Malaysia. Some are mid-tier and are (or were) made at fabs in Phoenix or Pocatello. And some are high-end and come out of the Gresham fabs that they bought when LSI went fabless.
There's value in producing the commodity stuff. Obviously they're not going to enter a bidding war with TSMC for cutting edge manufacturing equipment to make commodity voltage regulators. But they still invest in fab technology at the lower end.
I don't think ONN is particularly special in this space. They had plenty of competitors and they worked to distinguish themselves on cost, depth of their product catalog, and ability to execute.
For that matter, I wonder how a government-subsidized fab could compete with an old overseas fab making lower end parts that has been long-ago paid for?
Automotive chips for example tend to be produced on older nodes, not fancy 14nm and below at TSMC. I'm not talking about SoC used in infotainment but all the micro controllers all over the car.
There is really zero incentive for a chip maker to expand capacity of old nodes since they are obsolete in some sense. It doesn't pay to upgrade to a new node without a product that demands something from the new node, and even then it's still an obsolete node. Think 130nm, 90nm, or even 45nm, the world could use more capacity at all of those but would you really want to invest heavily in any of those?
I suspect the end game for most of the industry is commodity priced chips at every "last node" where that term means the last node that doesn't require X, and X is a technology change like immersion lithography, multi-patterning, material change, or EUV. With that kind of environment nobody can afford to increase capacity much.
I'm not arguing for government funding here, just pointing to what I suspect is one of the problems.