....to keep an Indian airbase close for 24 hours.
That only matters if you hit the runways but miss the assets. If you can destroy assets in hangers/tarmac, then they have to move entire squadrons around to get that base back up. These replacements could be from different theater commands, and that isn't a smooth process, especially for a notoriously unsmooth organization like the Indian military/bureaucracy.
So IMO, the real question is, what is the probability that the PLAAF can take out most IAF targets in the open and hardened shelters with its stand-off inventory? This is mostly a task for cruise missiles, if they have penetrators or are accurate enough to fly inside the hangor to detonate. Tomahawks could pull stunts like this in 1991. But this also depends on the availability of hardened shelters for the relevant IAF bases. The Rafales are going to be stationed at Ambala, I think, which does have hardened shelters. But how many assets can the IAF protect by parking them in hardened shelters in this theater? How many of its MKIs will be exposed on the tarmac? Also, the other thing cruise missiles can do is attack targets of opportunity out in the open, this is where 'AI' becomes relevant. If they spot a high value asset, like an AWACS parked on the tarmac, they can prioritize it automatically.
DEAD mission is an extremely difficult task to execute because of existing RWR emitter location inaccuracy.
The AI solution also ties into the SEAD/DEAD issue. We need to remember that in our new (but completely untested) 21st century SEAD/DEAD paradigm, ECM isn't supposed to matter that much anymore. Because if you have an AI enabled drone/missile swarm, then you've actually nullified the opponent's ECM capabilities, because you're no longer dependent on long range network communications anymore.
According to this new paradigm, all you really need to know is the general area where the opponent has deployed their SAM assets, and then you just launch your 'AI' drone swarm in its general direction, and let it do its thing autonomously. If your AI swarm loses contact with the control node due to ECM, that's not supposed to be a problem anymore.
This is a key 21st century capability, which is beyond the
'Network Centric Warfare' paradigm of the late 20th century. Both China and Russia began developing this paradigm as a direct response to 1991. It remains to be seen how they've progressed in this arena. AFAIK, US intel reports generally indicate (and also Praveen Sawhney has commented on this a lot) that the Chinese are advanced in this arena. If it works, it will revolutionize warfare just as much as 1991 did.