Earlier quoted context omitted.
You seem to think that this is a publicity stunt (rather than a sincere offer of heating services). I don't think it is.
Ignoring the technical difficulty of actually using the excess heat for reliably and consistently heating a building, it's hard to believe that the economics of connecting and servicing single cabinets randomly located all over the landscape could work out.
I kinda thought the same.
But assuming it can be properly encrypted (hard) what's the difference to swapping in a new server compared to swapping in a new gas bottle? User pays for the initial infrastructure.
Pull the old one out, pop the new one in, 60 seconds, any blue collar worker can do it, if it's setup properly. It would even be automatically checked remotely it was done 100% correctly.
Once every 3 years? Probably cost $50 a pop? Not necessarily crazy.