> This requires a magically flexible definition of "market" at very best.
The definition of a market is the set of goods or services which are fungible substitutes for one another.
The only thing that's hard about this at all is determining how fungible they have to be in order to be considered substitutes, e.g. two restaurants that are 3000 miles apart are obviously not substitutes for each other, but what if they're 30 miles apart? That one could go either way depending on how aggressive you want to be.
But it doesn't really matter how the coin flips like that turn out because the things that are close to the line aren't the markets where consolidation is causing major trouble, and conversely causing some local shops to not merge with each other so it stays unambiguous that the local market isn't consolidating is likewise not a big issue.
> Do non-profit organizations count for this rule?
Yes. If a non-profit has more than 15% market share then it has to be broken into multiple non-profits that each have less than 15% market share, the same as anyone else.
> Imagine a serious condition getting its first drug treatment approved and the company saying "Only every seventh patient is eligible."
You're assuming that there are still patents. If there are then that obviously doesn't count during the patent term because that's what patents are, but then you count only the patented invention and any sort of tying is entirely prohibited. You shouldn't be able to patent a connector and use that to monopolize the market for replacement razor blades.
Whereas if there aren't patents then you don't have this problem to begin with because you get a dozen different companies all producing the new medicine right away.
> operating systems
Operating systems are easy. Microsoft would be atomized because they have too much market share and then a dozen different companies would have the full rights to make derivatives of Windows. None of them would ever be allowed to have majority market share again, so now it's in everyone's interests to standardize APIs so that the same software will run on each company's Windows distribution, and the same APIs would end up being supported by (or consolidated with) non-Windows operating systems as well.
The OS then becomes fungible as intended and most likely ends up being open source since there's not a lot of margin in fungible commodities. And software licensed to allow the entire public to modify and redistribute it has an unlimited number of suppliers and would never exceed the market share threshold.
> utilities to a specific house or even city
If you're a utility company and you know you're going to get broken up if you have more than 15% market share, what do you do?
The sensible thing to do is to build your infrastructure differently. You're digging up the road once, but instead of installing one big pipe, you install a conduit with 10 smaller ones inside it. Then you sell off nine out of ten to nine other companies so that you each have ~10% market share and if one of you ends up with 14% and another 6% you're still not over the threshold.
Then if a natural disaster damages the infrastructure, in theory they could each employ their own repair crews and each send one to each damaged location, but that seems pretty inefficient when there is another option. You have multiple independent repair companies and the utility companies pay whichever one gives them the best rate for a given job, which will tend to be the one already sending someone to that site, i.e. the same repair crew usually repairs all 10 damaged pipes and gets paid by all 10 utility companies. Meanwhile a different repair company puts in the best bid for a different site.
The status quo takes the assumption that everything has to be vertically integrated, but it doesn't, and it's a lot easier to have competition when each new entrant doesn't have to reproduce the entire supply chain themselves.
> spy satellites, combat aircraft, armored fighting vehicles
You're listing things that are concentrated markets because the government is the only one that buys them and then there is only one seller because there is only one buyer. And even then the way the government does that is stupid. If they're the only one who wants armored fighting vehicles then they should take competing bids to design one and then take the design they paid for and take competing bids to manufacture it. The contracts to make these things should be going to the same companies that make trucks and locomotives.
Likewise they should break Boeing up and then have the various pieces of it put in bids to design, manufacture or assemble combat aircraft.
> EUV lithography
Nothing about this should be a monopoly. It should be something that dozens of companies know how to do. It consolidated because the government has been allowing companies to merge with competitors and suppliers for decades.
> chemicals that are difficult or dangerous to synthesize
These are the sort of things that should have 50 suppliers rather than 50,000. Nothing about it requires a monopoly. Indeed, allowing the market to concentrate is dangerous, because what happens if that one company's facilities are damaged or they're the only ones who know how to do it and the bus to the company retreat goes over a cliff?
> literally any new product category
So someone invents e.g. 3D printers. How does the time it takes to litigate a case in court compare to the time it takes for multiple companies to be making the new category of product?
There is no need or incentive to file a case if it would be dismissed as moot by the time a decision would be rendered. And if a "new" product category is still a consolidated market multiple years later then that's actually a consolidated market and it's time to break them up.