A two-inch cube with 10,000 hand-soldered connections. Why would anybody be skeptical about that?
To be nitpicky:
Roughly half the connections are in the controller, which is about 4 litres in volume.
The other half are in the Ecat proper, which including the outer shell is about 8 litres in volume.
Also, not all the connection are soldered.
What had seemed unlikely to me was not the number of connections but the (tentatively provided) power density of the reactor core: 5kW in 100cm^3. But this is 50 watts per cubic centimetre, which is in the ballpark of a candle flame. The much more amazing thing would be that 80 percent of this power is in the form of electricity (if Rossi has done the measurements correctly).