My question was for after they've filled all the carrier slots plus whatever extras they need for land-based roles like reserves and training. Its a good idea to keep producing at least some airframes, perhaps a minimum single-digit figure, even after those two use cases are filled just to preserve institutional knowledge to allow for surge capacity whenever a new carrier is finished, especially if such timelines have to be accelerated due to global events.
Based on my (admittedly napkin maths), it seems to me that unusually few (perhaps even only one?) "J-35 surge for new carriers" would be needed overall, keeping in mind how J-XDS first flight was before J-35 even officially entered service. I note that this is potentially the fastest direct replacement cycle we've seen amongst premier fighter jets in recent history, or perhaps the history of flight.
Due to the limited space on carriers, I suspect J-XDS would directly replace J-35 with the only bottleneck being production rate and integration maturity. Airframes are designed to last several decades and the J-35 has features (that are not without material and opportunity cost) that make it financially inefficient to procure if they're going to be pushed out of their intended carrier role before even a decade has passed. This is unlike J-15 since its efficiency as a missile truck cannot be matched by the smaller J-35, so they fit separate roles and can have ongoing parallel production.
In fact, this has been my primary concern with the J-35 programme. Certainly, its J-35A derivative will likely be very fruitful for the PLAAF given the cost of J-20 and the expected costs of J-36 and J-XDS (land variant) along with the potentially limitless air fleet cap, but the limited space on carriers pushes for a smaller production volume and higher investment per individual airframe; ergo, there's no point in having a few J-XDS lead an only slightly larger amount of J-35, might as well make the entire CAP carrier fighter allocation J-XDS. Its possible J-XDS (naval variant) will reach mass production before "Type-005" (the 5th carrier) is completed and commissioned, meaning that only 4 carriers-worth of J-35 (+ reserves and training) would be produced. Given the expected production rate of >100x J-35 family airframes per year from SAC due to their new factory, this could be achieved within just a few years (possibly before 2030) even with the majority of production being for the J-35A variant. We could be looking at a very short production run for the J-35.
