I'm too lazy to do internet/twitter searches to back up my numbers, but my vague recollection is as follows. Please do internet searches to confirm/refute/clarify my claims since: (a) I think my numbers are only approximately right; and (b) I'm just a random guy on the internet so what I claim should not be taken as being reliable (not because I'm lying, but because I have seen only a subset of the information available).
1. Starlink terminals costs a few thousand dollars to make, and are normally sold to customers for $500 or $600 plus a monthly subscription fee. The expectation is that, over the course of a few years, the profit made on the monthly subscription fee will pay for the loss on selling the hardware. There is precedent for this type of business model, e.g., "get a free/cheap smart phone when you sign up to our £40/month plan for a minimum of 24 months".
2. The monthly subscription fee (based on my probably out of date recollection) is usually something like $120 for a stationary installation for a family (limited number of users); $150 for an family (limited number of users) who like to roam (e.g., fit the Starlink terminal to an RV/caravan, or transport it between main home and holiday home); a few thousand dollars/month for a commercial installation that might have hundreds/thousands of users (e.g., a cruise ship that uses Starlink to provide internet access to all its passengers).
3. Recently, twitter comments revealed that, each month, about 500 of the 25,000 Starlink terminals currently in Ukraine are destroyed due to war-inflicted damage, and need to be replaced. Since those 500 terminals won't be gradually paying off the hardware-sold-at-a-loss via monthly subscriptions, the loss of those terminals is very expensive for the manufacturer. For example, if it cost $2000+ to make each of those terminals (which is my *hazy* recollection), then 500 of them costs $1 million. With this sort of expensive attrition rate, it seems reasonable for SpaceX to request that a donation to fund the cost of a Starlink terminal might be for a substantial part of the manufacturing cost rather than the sold-at-a-substantial-loss price normally charged to consumers.
4. A recent comment on twitter indicates that individuals in Ukraine who buy a Starlink terminal (rather than getting one via a donation from a large organization) pay only $60/month subscription charge. I assume the majority of these individual purchases will be used by small teams of mobile troops, so $60 is a substantial discount on the $150 normally charged for a roaming family.
5. A recent tweet from Elon indicates that at least some Starlink terminals in Ukraine are being hooked up to a network infrastructure that can supply internet to about 10,000 or 20,000 people. I *think* the tweet stated the network infrastructure might be mobile phone cell towers to supply connectivity at town scale, but my recollection of the details is hazy. (I'm *guessing* that ethernet-based LAN might be preferred for providing a military base-scale communication network but I'm not a military guy.) I'm *assuming* that this sort of use case is at the high end of what the several thousand dollars/month subscription fee is intended to cover.
6. My *assumption* is that the Ukrainian military will use a subset of Starlink terminals for town-scale or military base-scale communication, and a subset for smaller scale mobile-army-troops activities. However, rather than configuring a specific subset of Starlink terminals for one use case and another subset of Starlink terminals for a different use case, it is logistically simpler to just configure all the Starlink terminals for the most general-purpose (and hence most expensive) monthly subscription service.
7. According to recent Elon tweets, SpaceX engineers have been investing a lot of effort into hardening Starlink infrastructure to withstand hacking/jamming attacks from Russia. This represents not only the salary cost of those employees, but also an opportunity cost, since those employees could otherwise be working on profit-making projects for SpaceX. For example, if this anti-hacking/jamming work ends up delaying the maturing of Starship by a few months, then the opportunity cost is basically the profits that could have been made by launching Starship to deploy commercial satellites multiple times during a few months.
8. There have been some apparently contradictory claims about who is bearing the cost of supplying Starlink to Ukraine. These boil down to claims like "It has cost SpaceX $XXX to provide Starlink to Ukraine" and somebody else saying, "Hang on a minute, [name of organization] has provided $YYY funding for supplying Starlink to Ukraine" and then some people claim "This means Elon is lying since it really costs SpaceX only $XXX - $YYY = $VeryLittle", while another interpretation could be that it costs $XXX + $YYY to supply Starlink to Ukraine. I have no idea which of these contradictory claims about cost are true.
The take-away point is this. If you assume costs of $600 per Starlink terminal and $60/month subscription charge, then you get a relatively low cost of supplying Starlink to Ukraine. If you assume the higher manufacturing costs per Starlink terminal and the non-subsidised typical subscription-fees-depending-on-the-use-case costs plus the effort invested in preventing hacking/jamming attempts, then you get a much higher cost of supplying Starlink to Ukraine. And that just based on what I have read. It is almost certain that my knowledge of relevant information is incomplete, so treat the above with a grain of salt. At the moment, I think it is too early to confidently claim whether Elon is a good/bad guy in this current controversy.