All the figures for markups and cost estimate requirements are standard DPA stuff. They may even make the rough outlines publicly available, given everyone works to the same numbers.
Detailed figures are unlikely to ever appear in public because a) they're not the sort of thing that can be released and b) you have to have something to compare to if you want to know if something is 'cheap' or not. (Not that I'd have such things anyway, but I'm familiar with how this kind of program tends to work so can guess at stuff).
But a basic breakdown would be the actual UAV itself, the avionics, and the base-station.
The UAV itself is relatively cheap. This is the bit people see and think is just a big RC plane, which is indeed what it is to some extent. No idea on the cost, but it wouldn't be a particularly high number, low 6 figures probably if that. Mostly because it's an 'aircraft' with all that involves in test and certification, rather than a toy.
The 'avionics' isn't just the bits that fly the plane (and I kind of doubt we had flight computers 100 years ago!), it also includes the sensor suite and any other electronics on the platform.
Of this, the SAR is going to be the big cost item. Getting the antenna array, and the onboard processing small and light enough to fit a UAV is not a cheap job. Mid - 7 figures wouldn't be a surprise for the SAR. The other sensors e.g. IR are going to be cheaper, but not 'cheap'. While in many cases the concepts already exist the specific solutions tend to be new to make them smaller/lighter/better to actually get the project to meet the requirements.
The basestation cost could be almost anything, depending on how complex it is. Assuming a self contained vehicle with other facilities a high 6/low 7 figure sum wouldn't be a surprise.
One thing that people often overlook when considering the costs is that most of the time, the component order quantities are tiny. Most commercial equipment is ordered in quantities of 1000's. Military kit tends to hit 10's, if you're lucky. And the cost per component is therefore dramatically higher. And this assumes you can actually get the stuff in the quantity you want, many suppliers can't cope with supplying 10 of something if they normally sell at an MOQ of 500 - so you either can't buy, or buy stuff you don't use. This affects all sorts of normal type bits, resistors, screws, processors - anything where you'd usually tend to buy in bulk rather than one or two. You can reduce the impact by a standard parts list across multiple programs, but it never goes completely away.
Then you hit the need to buy 'special' MIL-SPEC bits, either semiconductors with the right temperature range (at least a 10x markup from the supplier due to different process/packaging/low volume), or special versions of components - it's not a shock to spend £250 on a connector that would cost £5 for the normal version, just because you *need* the ultralight, robust version.
Then you trip over the manufacturing problems of low quantity - you either can't use the 'cheap' method e.g. you machine something from billet because it's cheaper than tooling for pressing/casting/folding, or you end up with horribly expensive parts that would normally be cheap eg. plastic or rubber parts, or extrusions, that normally would be penny's or a couple of quid in mass production, but with £30K of tooling over a run of 15 start to get expensive. Same with PCB layout/manufacture and all the rest.
And of course you tend to be designing for a 20-25 year service life, at full performance, which is a different ballgame from the 5 year design life of most equipment. Makes a big difference to how things are done, and how much it costs.
Then of course there's all the QA, documentation trail etc. etc. which isn't cheap.
Plus of course the 'program' costs include the NRE costs i.e. development, which is a big fat lump of cash, and looks horrendous spread over 50 units. On 100000 units it wouldn't even show up.
None of this is exactly a surprise - find out how much a development run of 10 prototype Mondeos or Nokia phones costs and you wouldn't believe it either. Or just work out how much an XBox would cost if there were only ever 50 made.
I can see why people think all this stuff is a rip-off, and can't understand why it's so expensive, but the simple reality is that high performance, low production quantity bespoke kit is always expensive. It's not a con, it's not a conspiracy, it's not sloppy engineering or management (apart from *cough* changing customer requirements *cough*), it's simple reality.