There are currently three foundries of logic chips such as CPUs and GPUs with a not completely ridiculous claim to be on the cutting edge: TSMC, Samsung, and Intel. Intel has largely been failing at staying on the cutting edge lately, but they were the leader for decades. This number has steadily decreased over the decades as each new process node cost more than the previous, leading one foundry after another to conclude that they didn't have the volume to justify further development costs.
Now SMIC would like to join the party. They don't really have an international business case for it, but they do have the backing of a major government. Chinese dictator Xi Jinping wants to have everything made in China, rather than relying on foreign countries that sometimes get hostile when China tries to rob them. His plan is called Made in China 2025, and that means that China wants the capability to build everything from CPUs to GPUs to DRAM to NAND on their own--and similarly for many other industries.
You need a foundry to produce it, and SMIC is it, at least for logic chips. There are others that produce DRAM, albeit largely by stealing designs from Samsung, Micron, and Hynix. With enough funding from a government, you don't actually need to be profitable. And SMIC can fabricate chips for China that they can't get foreign companies to build over IP concerns or embargoes or whatever.
I don't know how much of their own development SMIC is doing, as opposed to stealing stuff from foreign companies, though it surely takes some amount of in-house development even to take a stolen design and make it work. But either way, they're way behind the curve. They recently started production on their 14 nm process node, about four years after TSMC and Samsung and five years after Intel. They hope to move to 7 nm in two years or so, still well behind the competition.
Except that they're not calling it 7 nm like everyone else does. They're calling it their "N+1" node. After N+1 comes N+2, naturally. From the looks of it, N+1 might be comparable to one of TSMC or Samsung's earlier 7 nm nodes, while N+2 is more of a 7 nm EUV node. Considering how dubious the nanometer nomenclature of process node names has become, getting away from it is hardly a bad thing.
Companies like Apple, AMD, and Nvidia aren't going to start producing chips at SMIC anytime soon, let alone a company like Intel or Samsung that prefers to fab their own chips. Rather, SMIC is really only targeting China's domestic market.
Considering that their development of new process nodes is only happening at all because of lavish funding from the Chinese government, SMIC is in a precarious situation. Their only hopes of survival are:
1) continued funding from the government of China, regardless of how much money they lose, or
2) a large enough market of producing chips for use only in China because those chips cannot be produced at TSMC or Samsung and then imported
In particular, SMIC can't really produce chips for foreign companies or produce chips for Chinese companies to export, as they just won't be competitive with what you could get from TSMC or Samsung. Even companies that want to build a chip on an older, cheaper process node generally want that node to be very mature, which SMIC's 14 nm isn't. That's the problem with being several years behind the competition, and one of the reasons why so many dozens of foundries have stopped developing newer nodes over the decades.
Would SMIC be able to catch up to their foreign competitors if their government continues heavily subsidizing them for decades? Likely not. At minimum, they're going to have a much harder time buying needed equipment then their competitors. China doesn't have a viable alternative to ASML (which produces some key equipment that foundries use), for example. It will be interesting to see how this plays out, and whether SMIC can ever be anything more than a backup plan to produce things that China cannot import or refuses to import for political reasons.