My company is designing a new UAV and we would be interested by using the Myxa A0 ESC.
Our preliminary study show that we are going to use a motor with a maximum power of 500W and a power of 200W at hovering.

We would like to use the Myxa A0 for a better integration but the Myxa product page specify the maximum power of the A0 at 400W and the A2 maximum power at 850W. Is the hardware different between the A0 and A2 ? Or it’s just a thermal management problem ?

The difference between A0 and A2 is that the latter is equipped with a heat sink, which is a part of the enclosure. You can study this interactive 3D model here to understand the construction:

The heat sink significantly reduces the thermal impedance of the device, allowing it to dissipate more power, hence increasing the operational envelope. The applicability of A0 for your application largely depends on its heat constraints: are there going to be any other heat-radiating components near Myxa? Is the volume well-ventilated?

If the parameters of the system are known, you could take the above linked 3D model of Myxa (without the enclosure) and perform a FEM simulation to understand the thermal dynamics. Alternatively, you could just get one Myxa A0 and test it on a bench.

So we will run some thermal simulation and test it on a bench.
Do you have a heat map of the Myxa A0 ? I saw in this post that you have a temperature map but I couln’t find an estimation of the maximum heat diffuse by the different components.

Greetings Hadrien! When dealing with simulations you always have to find a tradeoff between the amount of details in your model and effort you want to put into the simulation vs accuracy of the result.
In Myxa ESC the components that produce most heat are power mosfets, heating of all other components can be ignored. Also in some simulations it may be tricky to model the PCB itself reliable, so if possible - it should be excluded too.
I used this approach and just put heat directly into the housing radiator surface keeping in mind that crystals inside mosfets will be warmer than the thermal interface by the known amount which depends on power(your usual thermal resistance stuff).
The second very important factor is convection(airflow). It is often measured in pretty weird units that are hard to relate to real environmental conditions. You may try to use this table Convective Heat Transfer Coefficients Table Chart
But convection is really what dictates the final power you can get. We had one design where just by placing the ESC in slightly different place we achieved 2.5 increase in operational power level.
So the best way, I believe, is to use model linked above to perform your own simulations with your drone model and convection factor(which you will have to guess, maybe).
Heating power is another questionable factor. In general, there are at least 4 approaches to get it, but the simplest one is to get it from estimated efficiency of Myxa(which is around 95%) and your operational power level. Say, it is 500W. So, you will have to dissipate (1-0.95)*500 = 25W. Of course this is not the exact number, but this is a good estimation of what you will have to deal with.

I will be pleased to see results of your simulation. You can publish it here or just e-mail me to alexander.sysoev@zubax.com