This is getting off topic, but some of you have got things seriously backwards if you think China’s strategy is going to be about spamming suicide drones against American and American-pawn targets in the second island chain.
The most likely outcome from a head-to-head military clash between China and the US will have the US take on the role of Russia today, reduced to taking pot shots at Chinese civilian infrastructure soft targets from long range, because Chinese A2AD and offensive strike capabilities will wipe out US forces in the first island chain very quickly and will make strikes against Chinese military targets prohibitively expensive and with limited chances success.
What China will need to do is strike deeper and further to push the US so far back that the number of missiles and systems they can use to try to strike of soft civilian targets on the Chinese mainland is reduced to such a small number that it is no longer cost prohibitive for the PLA to intercept them effectively indefinitely. That is essentially the minimum threshold for China to be able to achieve a minimal cost victory against the US to make direct armed conflict a good choice for China.
Once that happens, you will see US appetites for direct conflict plummet.
This is why the PLAAF have been favouring heavy, long range fighters, but also why they won’t go full 5th gen.
Once they become available, I can see the PLA get into loyal wingmen drones in a pretty big way, both for AG and AA. For AA, they will probably also go both high and low end, with dark sword like high end loyal wingmen intended to take on enemy manned fighters, while FH97As take the low-end missions for anti-missile and drone work.
This is also why the AA version of the FH97A drone have gone with maximum number of miniature missiles approach. I don’t think these loyal wingmen are meant to go after enemy manned fighters, rather they are intended to take out enemy cruise missiles, UCAVs and suicide drones. It’s orders of magnitudes more cost effective to fly a loyal wingman to shoot a tiny missiles at a cruise missiles than needing to use a HQ9 missile.
FH97A drones will probably rely heavily on AI and be deployed in large numbers compared to friendly manned aircraft and will be able to do most of the engagement autonomously since they will be taking on targets with minimal to no evasive or defensive capabilities.
Dark Sword class high end drones will paradoxically probably use a lot less AI since we are still too far off from that level of tech. Instead you will likely have the back seater in a J20S take control directly to take on enemy manned fighters.