To the OP:
Coming up with the idea of utilising excess heat is a good indication - it means you are thinking about the situation. This can never be a bad thing.
The next step is to address the process of doing so and working out things like efficiency, cost, benefit and so on. As indicated above, these numbers will not be encouraging, especially at the scale we are talking about.
Somewhere along the line, there is an important question that is so obvious, you might not even think about asking it - but it is crucial that you not only ask it, but answer it as well. That answer will have a world of further thinking come out of it....
The question is "WHY am I trying to do (this)?"
An answer to the specific scenario given, I would suggest, is that you want to make the most out of the power available. An admirable quest.
The next step - is taking a step backwards and having a look at the broader picture and asking yourself whether there are other ways of looking at the issue ... and what other approaches might be considered. This is where you will find phrases like "lateral thinking" and "thinking outside the box" getting used. It is also where "brainstorming" comes in. (If you've never been involved in or witnessed a brainstorming session, I strongly recommend you take up the opportunity if it comes your way. They can be crazy, fun and surprising.)
One such alternative line of thought is - Can I reduce the power used in the first place? That will reduce the excess heat generated and allow the power that would be otherwise expelled in creating some of that heat to be used directly, without any recovery process required. To get a feel for how good or bad this idea might be, throw some numbers in. Back of a napkin numbers will do for starters.
Let's say we have a resistor dissipating 10W. Compare the effort of trying to recover 50% of that from the waste heat against reducing the power dissipation of the resistor by 50%. The first option hits a brick wall very quickly. The efficiencies are just not there. The second option might require a redesign of the circuit around the resistor. This might be a challenge - but it's not a brick wall.
A practical result from efforts on both of these fronts soon shows which is the best value approach. A recovery system would be lucky to provide a 5% benefit, with the complexity and cost of implementing it and delivering the power recovered in a useful way. A 5% benefit through a redesign may well be fairly easy (in comparison) with 50% and maybe even 90% being a possibility. No recovery system is required and consideration for delivery of the power is unnecessary - it's already available at the existing terminals!
We have now reached a point where the preferred development path - on the basis of the original objective - has been clearly identified. What we then do is assess all the implications of travelling down this path and work out whether this path is a good idea - or if it will lead us into a world of pain because of some incidental consequence.
The addition of an energy recovery system will add bulk, complexity, cost and an increased risk of failure (because there are more things to go wrong). The reduction of power consumption approach will mean the existing power supply can do more - and/or you can reduce the requirements of the power supply. If we opt for the latter, we can have a smaller, cheaper power supply, which could likely be more reliable. The overall heat generation will be less and the measures required to remove unwanted heat will be far more modest. Since heat is the number one enemy of electronics, this is a very good thing. A reduction in size and weight is also very useful - especially in aerospace applications.
However, there is still one hurdle to travelling down our obviously superior path. The long-standing conflict of the engineering ideal against the commercial reality. But this conflict is not just an engineering one - it exists in every industry and there are a lot of factors that can come into play. I won't attempt to go into any details here ... it's a subject all its own ... but you would be hard pressed to find anyone here who's been in the workforce for more than a year or two that doesn't have a tale to tell.