Yes, I was thinking 3 months, but 4 months could still mean Macs on store shelves in spring 2024. Supply chain analysts have stated that it takes about 3 months from chip availability to laptop retail availability. January plus 3 months = April, and spring ends in June.
Apple's inventory controls and just-in-time manufacturing is better than that. Even more so if just slapping a new logic board into a container that has already been in high volume production for 2-3 years. It doesn't take 3 months to put a laptop or iMac together once you have all the parts in front of you.
If need new jigs, sourcing several new major subcomponents ( screens , logic boards , cases , etc) , but if change as little as possible then the time comes down.
A major part of that 3 months might be to build inventory to handle the initial demand bubble. ( iPhone production goes full tilt in July-Aug-Sept to build stockpile to be rapidly depleted in a three week period late Sept- early Oct.
The bubble for the iMac looks nothing like the bubble for the MBA 13" or iPhone.
As for N3B, TSMC had already stated that they began volume production in 2022 Q4. However, that's presumably for A17, a chip which has a fixed release date of September/October 2023, for a product that ships much, much, much higher volumes.
While technically in "HVP" there has been lots of chatter around N3B that ramp is not progressing as fast as previous generations. Back in Feb-March there were reports of 40K/mo wafer capacity but running at 50% utliization for the 3nm capacity
" ...
DigiTimes reports that TSMC's 5nm fabrication capacity began to loosen in November 2022 as a result of reduced orders from Apple, amongst other partners, with orders for iPhone chips alone having been slashed by 30%. However, the Taiwanese manufacturer has apparently been able to keep its utilization rate at 70% or higher thanks to Apple's thirst for 3nm:
... The foundry will also grow the process output to 50,000-55,000 wafers monthly in March ...
..."
[ That 70% utilization is across all the product lines. ]
There are some other reports that Apple pushed TSMC to 'eat' all the bad dies until the yield rate crosses some percentage point that Apple likes better. [ Instead of paying for wafers (normal mode) , Apple just pays for good dies. That won't get to higher yields quicker , because TSMC is going to control losses; including some equipment go idle and avoiding bigger dies. ]
45-55K/mo isn't going to be overwhelmed by the steady state iPhone demand. There is a bubble in Sept-October but if run from March-June it is doable to build a substantive stockpile to handle the bubble. The only way Apple would get into trouble would be if both the plain iPhone 15 and Pro 15 both had an A17 in them. But as long as it is in the ballpark of 'iPhone' the SoC sales are A16 (on N4) there is plenty of capacity to get some limited M3 sized dies looped into the stream. For M3 (and bigger) Apple likely would have to actually buy the wafers though.
Because the MBA 15" is a new separate product and the MBA 13" is a very high volume seller. The Mac mini isn't a high volume seller so they can afford to wait with it if desired. But yes, they could theoretically stagger those releases too if they wanted.
In the desktop segment, the Mini probably is a relatively high volume seller. The VP of marketing made some comments around WWDC 2023 events to the effect that the "Mini has been quite successful". Unchained from the Intel iGPU limits it is doing
far better than it was in the x86 era.
Apple probably intends for the MBA 15" to be a high volume product. Either take the MBP 13" #2 status away or something very close that is more additive ( get overall Mac growth. Scoop up folks who exited to Windows PC because the MBP 14/16" are to expensive for their budgets. ) . If 4-5 months later throw 'shade' on the MBA 15" there is very good chance going to confuse a substantive number of buyers. Making them choose between SoC instead of screen size. That isn't the intent of the product.
Apple has much bigger competition problems on the desktop products.
For all we know, Apple could release an M2 iMac. 🤪
It is not impossible , but just more than unlikely.
It isn't like they are waiting on the M2 to ramp in production. There have been stories where Apple dialed back M2 family orders earlier in the year. Not good chance either that the iMac is being held up by some radically different screen. ( It might get a tweak so that the peak brightness matches the Studio Display, but pretty good chance Apple just squats on the exact same panel. If aligning backlights with the SD made both of them cheaper to make then they would be interested, but otherwise Apple is likely just looking for lower bill of material costs to offset the increase in SoC costs. )
If Apple was using a M2 they far more likely would have made a very big 'splash' around the exact anniversary of the iMac introduction. Or possibly June ( if waiting to Mini demand surge to subside)
It's not about clamouring for N3E. It's about managing expectations. New M3 Macs in October just seems really optimistic.
Not sure how you square that with N3B being in production since January of this year. At modest accumlation rates Apple could have put together 0.7-1M M3 chips by October and not impacted fabbing A17 volume at all.
To get to a 1M M2 sized dies they would only need around 4K wafers. If have 45-55K capacity that is less than 10% even if tried to do it all in a month. Split over two months it is less than 5%. Split over 3 months then less than 3%.
If Apple did intro in late Oct and shipped in first week in November, then if did a 5% peel in May and June they'd have flow for exiting the fab pipeline in Sept and Oct. The Sept stuff goes to factoring in Oct and ships in Nov. The Oct follows in Dec. Done.
If the A17 stockpiling process started in Feburary then scaling bit very slightly in May isn't going to hurt. ( by May stuff is already coming out of that pipeline before even started the M3 stuff. So even if only did 5% back in Feb you have just filled that 'loss' from a diversion. If Feb was at 10% there is no loss on either. )
The burning question was what was Apple making back in Feb-March timeframe with at least 20K/wafers per month going through. R1 chips? If it wasn't A17 chips it was something!!! And if it was A17 chips then they have got a reasonable sized stockpile before the June iPhone build season even starts.
[ P.S. This does go sideways if Apple just refuses to buy any wafers for M3 until Aug or September ... yeah then it is all 'doomed' because they started too late. The question would be why they would 'need' to do that. ]
