I don't buy this. It means that the M4, which was a surprisingly big leap I thought, is going to sit exclusively in an iPad for.... how many months? There's no reason for that.
If there is both a A18 and A18 Pro SoCs which are also on TSMC N3E that are 'hogging up' the fab production line, then there would be a good reason for that.
[ And Intel is about to release Ultra 200 (Lunar Lake) which is on TSMC N3B so it isn't like those fab machines have less to do either. Even if those SoCs ship in products in October, a large wave of those dies need to be in production now. (N3B has a long 'bake' time. ). The CPU+GPU tile might be smaller than a M4 but likely Intel will need more than several million of them. So the amount of N3 family dies that will ship this year has a pretty good chance of doubling. ]
Similarly M4 in iPad Pro is as much to blunt Surface ( and other) X Elite tablets with the screen just as much with compute. Apple's display controller hasn't been leading the pack in performance. Apple openly stated that they needed a substantive upgrade to drive the iPad Pro screen. M3's display controller tech works just fine on what will very likely be the same set of Mac screens that will come on next iteration ( in part to control costs).
Apple just did a 'short cycle' refresh on the MBP 14/16". Lots of good reasons not to do another one now. The MBA just got a refresh. So even less reason there.
The Mini is coupled to both the M4 and M4 Pro. M4 Pro is coupled to the MBP 14/16.
The Mac Pro .... Apple's iteration cycle for last decade has been 3+ years. And the 'short cycle' on the M2 Max in the MBP 14/16 likely only puts more pressure on the Studio/Pro to soak up more M2 Max dies for a longer period of time. For the maximum cost SoC of Ultra there is pretty good reasons for Apple to get on a n+2 iteration cycle: M2 -> M4 -> M6 ( or odds ). If Apple went every 2 years for Mac Pro it would be much faster than they have been doing the last decade or so. The market size of the Ultra Studio/Mac Pro is not big enough to churn out a new SoC every year.
Additionally, the much larger ( 4x + ) dies eat up even more fab capacity. And there currently is no "hand me down" product to throw the Ultra into. Nobody (AMD , Intel , Nvidia , etc) completely tosses large dies/packages in a single year. [ Some folks will hand wave at the "AI servers" are coming , but that is substantially a one trick pony. Once do the initial roll out, is Apple going to throw away deployed severs on a yearly basis? Probably not. If Apple is not explicitly charging for AI servers, then more unlikely. More so it would help with keeping something like a n+2 cycle steadily chugging along. It will soak up some older large SoCs, but it will have a limited capacity to do so. ]
After initial demand bubble for the the A18 , then it would not make as much sense. October/November would be dragging on the Mini and would be feeling more 'competitive heat' in the upper end laptops. Discounts on the MBA would keep them moving until 2025. ( **)
If there's no hardware I suppose the logical time for nearly the entire Mac line-up getting M4 / Pro / Max / Ultra / Megabastard would be September to take advantage of all these AI features when the software drops. But that also just doesn't sit right.
The entire line up? Probably not. Bigger than Ultra SoC? Even more so probably not. The faster the "kill off and move on" pace of the SoCs, the less likely the package will get bigger (and even more costly to make).
** And if Apple is going to reduce the roll the N3B M3 is going to play in "hand me down" systems ( e.g, skip iPad Air) then MBAs soaking up more of those for longer would be a way of clawing out better returns on the SoCs via volume. ( all the more so if flip the MBP on a year iteration so those M3 gen dies don't contribute as much also.)
Plus, during the gap where there is M4 MBP 14" versus M3 MBA , Apple can upsell more folks into buying the MBP 14". (better margins. Decent chance that uplift in margins substantially pays for the discounts applied to MBA to keep them moving. So no net hit on aggregate margins. ).