Mining on a SoC that fits in a security camera must be the definition of futile. How in the world does one of these ever "win" to earn?
You statement is true in the wider sense, you can have mining botnets of computers and maybe some high-end phones, but IP cameras are really-really weak as far as compute power go.
Of course they most likely aren't doing that, but it's not out of the realm of possibility (and thus a missed opportunity to make this debacle peak cyberpunk).
Non-techy people are willing to look so so far the other way in trade of convenience. Showing them "facts" about how much electricity their cameras are using would not impact the vast majority of the users if they feel like the service they are receiving is good enough.
(Might be easier to discover it on the design/procurement/supply chain side. Perhaps stock players would find out when investigating their investments. A high-profile brand putting crypto miners on consumer hardware is newsworthy, and news move stocks, so there's incentive to leak the story.)
So comparable to a smartphone under load. That's a pretty good power budget, given that most of the components you mentioned (except RAM/SoC) are going to be operated intermittently.
> sensor
Serious question: you mean image post-processing and encoding here? I thought CMOS sensors themselves have negligible power demands?
> heaters
A crypto miner is just that - an overcomplicated heater. If you include one, you don't need the other.
> How much computing power do you really think you can get with that?
Not much, though an ASIC would get a nice multiplier on it. I'd guesstimate it to be comparable to a cryptominer running in the browser of an unsuspecting regular person - who will typically have a cheap, underpowered machine that can barely lift the OS and the browser without hanging. In recent years, at least some people thought you can make a profit with it, because we've seen such web cryptominers deployed around the web.
My point isn't that it's happening - I fully expect the cost of sourcing a miner ASIC and redesigning a camera around it to be much larger than the mining it would bring in any reasonable timeframe. But I think that on first look, it's possible, if someone tried hard enough, to build a sneaky security camera like this, that would be able to eventually yield some profit if deployed wide enough.
It would be horribly inefficient however a network of tens or hundreds of thousands of low resource units can absolutely make money since the costs are zero.
Also on top of all of that most IP cameras are designed with tightly defined specs. Trying to run a miner on them would absolutely requires stopping the video feed which would make the "attack" (which is still useless) easy to detect.