@Medved: I think UVC LEDs are, and will always be a niche for good reasons. That monochromatic UVC lithography is probably a niche to things like supercomputers or military technology or other specifics which don't pass the 100K units per year, and they're capable of paying such things, but they probably invest their research to use anything cheaper than the LEDs instead of developing the LEDs, so they will probably extinct as an industry failure.
Also, silicon components are dirt cheap nowadays, which means they use cheaper process like blue light and their research will go into make the process more efficient and cheaper than experiment with a risky, expensive and unknown thing that can cause massive losses, so again such LEDs will extinct and whoever needs them will need to manufacture them for theirselves.
You are wrong, the short wave UV LEDs are used mainly for the mass market semiconductor manufacture.
The military is not using such advanced geometry nodes, for a good reason: These components are way less reliable and military need to cut the processing costs per function that much as the consumer goods do (as the military has way less number of units made, so the extreme development, complicated by the reliability requirements, would never pay off and would delay the availability of the devices for their users). The purpose of using the more dense technologies is to reduce the production cost, noting else. Te thing is, although smaller transistor nodes are more expensive per unit area of the chip, you may pack way more functionality on that unit of area. So for the same function you get on way smaller chip, the wafer holds way more of them, so bottom line the same function is available for cheaper. And that is the only drive for the more advanced nodes.
Traditionally FE lasers (very bulky devices, so need an optical cable, but that passes barely 1% of the input radiation power, so the lasers had to be of very high power) were used for a single peak exposure, the seemingly very expensive LEDs are actually a cost reduction thing, because they are way cheaper than the laser tubes. Not the light source itself, but the lito unit itself gets way simpler so cheaper with LED vs the laser as a light source. The thing is, the LED is small, lightweight so could be used without the hefty optical cable, you may modulate its output as you desire, all that makes mainly the mechanics way simpler (we are talking about a machine positioning its few kg head in um precision within few 10's ms; having the head some kg of the optical cable and its support lighter makes the thing quite easier to handle, while some k$ for the LED is quite nothing compare to what you save on the mechanics of that machine; plus the laser is not much cheaper anyway).
But unlike with the lasers or discharge tubes, the LEDs have the inherent advantage of becoming way cheaper per unit once you ramp up the production. So for the few 1000's units requiring UVC sold world wide the discharges are clearly the cheapest. But if the numbers go to 10+M units, the situation will be very different.
Yes, today it cost $1k to make an UVC LED. But it is mainly because you need barely few 100's of them world wide, worth 1M$. But the point with LEDs is, making them in 10M/year would cost lets say $10M production cost. But that dilutes to $1 per device, you may need ~3 of them, so $3 per disinfection unit. Compare that to the $30 a typical UVC germicidal lamp cost. Yes, it pays off only if you sell the 3M disinfection units per year and you must pay few $100M in advance for the development. And that is the main reason, why theere are no $1 UVC LEDs on the market yet, there was simply no such big market for these devices. And it is the biggest unknown for the future at the moment: The virus may trigger such market, or may not. So a big question mark, if it is worth investing the M$ into the required development now or just let the development go slowly with minimum short term investments.