The J20 as it is is not suitable as a carrier fighter. Even if you do bother to make the necessary structural reinforcements and add in a completely new landing gear design, some of it's basic design choices would make it a les than ideal carrier fighter. It is too long and large for one, and the main landing gears are too far forward. Those are the sort of things that are hard if not impossible to change without it being easier to just start fresh. Thinking of what the PLAN would be doing and who they might use their carriers against, I really don't see the PLANAF seeing a carrier version as a particularly high priority. If there is a war with the US, the PLAN's carriers would is all likelihood not feature. Just too many eggs in one basket and they would tie down too much of the PLAN's assets to protect even if that was possible. I would see the carrier fighters operating from land bases while the escorts are retasked with protecting mainland installations from cruise missiles and air strikes and proving escorts for any invasion force headed for Taiwan. TBH, I think the USN would love to see J20s on carriers, where they are all bunched up and could be taken out with one attack far easier than if they were based on mainland bases. I can see the PLANAF operating J20s, but they will be based at land bases and used to hunt enemy AWACS and tankers and intercepting incoming strike packages instead of going carrier hunting. Hell, AShW is probably one of the worse ways you can use the J20, right up there with CAS. A carrier battlegroup is probably one of the best suited forces to counter stealth fighters and bombers. You have a dearth of very powerful AESA radars, both ship borne and air borne, all netoworked via high bandwidth datalinks and backed up with serious processing capacity to analysis all the sensor data with the different escorts being spaced out. You would be hard pressed to come up with a better set up to counter stealth. Remember that stealth is as much about re-directing incoming radar energy as it is about absorption. What might work against a single radar could be a disaster against a group of them, whereby your stealth shaping ends up redirecting the radar energy from on enemy ship right smack bang into the array of another. The kinds of weapons needed to take out ships and carriers also makes the J20 unsuitable as you need massive warheads, and a lot of them to kill ships or carriers. The J20 cannot carry such weapons internally, and even if it could, with layer defences or a fleet, as soon as you release the weapons they will be detected and still subject to multiple layers or interception. So you still need saturation attacks. To get close enough to bypass some layers would mean you are putting the J20 itself within range to be engaged. It's just nit a good idea. The J20 would be far better used as escorts to forge ahead of a strike package and sweep things like AWACS and enemy fighters away before your conventional strike package and open up a corridor for them. As for where the J20 will be first deployed, well I think the J10 deployment could be a good indicator. They will almost certainly base the first units inland so potential enemies don't have a chance to get an idea of what it's signiture might be like. So Chengdu would make sense as that is close enough to CAC to have quite on site assistance if anything crops up.