Why not just use AWACs to control drones remotely?
Firstly, it doesn't have to be a choice, you can quite happily have both.
However, even if you do use AWACS or dedicated airborne drone controllers, having a twin seat J20 that can also pilot drones would still be a massive advantage.
Lag and vulnerability would be the key differences.
In air to air combat, split seconds can be the difference between victory or defeat. That means you want to keep the distance between drone and controller as close as feasible, and ideally within line of sight so you don't need to rely on satellite relays, which both adds lag, and also introduces potential bandwidth bottlenecks, not to mention presenting systematic weak points an opponent could target to render your entire drone fleet useless using ASAT weapons.
With an AWACS, you have the dilemma of having to trade off between lag and security.
The AWACS or even separate, dedicated airborne control drone centre would also be high value targets that the enemy would be willing to pay a steep price to kill, since doing so would render a large number of drones useless.
Stealth fighters also make it that much harder to effectively protect your AWACS.
A twin seat J20 would at worse only have marginally worse kinetic performance compared to the baseline J20, but would still be an extremely deadly air dominance fighter.
Being stealthy, they are inherently far harder to detect and engage by the enemy compared to an AWACS, and given the expected high performance of the jet, there is little guarantee any potential adversary fighter could get the better of it even if they were able to bring the J20 to battle.
And even if the enemy can find a J20S, bring it to battle and shoot it down, they are at best only taking one drone out of the fight.
The second part of vulnerability comes in the form of jamming. The further the distance, the weaker the signal, and the more susceptible that signal is to being jammed.
The direction the signals are traveling in can also be significant in determining how effective that jamming is.
In a very simplistic example, say you have one scenario where the fighter is controlled by a friendly fighter in LOS with it flying a couple hundred miles behind the UAV, compared to an example where the same UAV is being controlled by an AWACS a thousand miles away using a satellite relay.
Now, it would be possible for an enemy to either try and blanket jam your communications relay satellite, since they would be able to detect and pinpoint it, and/or, they can direct jamming signals downwards, targeting your UAV by either using their own satellites to emit/retransmit jamming signals or bouncing signals using the atmosphere.
The point is, the enemy jamming signals would either be able to brute force overload your satellite receivers, and/or be coming from the same direction as your own command signals to your UAV, thereby massively complicating your counter jamming operations.
If you have a fighter tailing the drone and commanding it, the command signal is going to be coming from directly behind it, and it would be far harder, if not impossible, for your enemy to be able to get jamming signals to reach your drone from that vector. That immediately gives your drone a massive counter jamming advantage as it can perform simple directional analysis on incoming signals and effectively ignore anything not coming from the direction of its controller. Depending on the design, it could potentially even position its receiver antenna such that it doesn't even 'hear' much of the enemy jamming signals to start with.
