The yearly M-series release cadence

[Bodhitree]

I remember hearing at the time of the M1’s release that one of the reasons that Apple ditched Intel chips was because they weren’t supplying frequent-enough product updates. It’s interesting because looking at it a few years later, Apple has managed to establish a yearly or nearly-so update schedule. But has it really helped them in sales?

 

[senttoschool]

Of course it has. If not, they would stop doing it.

 

[Bodhitree]

Well, it seems to me the effect of a yearly release is quite difficult to separate from the other effects of ARM processors. But perhaps they just want to time new products as a release before the holiday season, that would make sense to give people an extra reason to give tech products for Christmas.

 

[!!!]

The problem I have is how they keep staggering the M_ release for certain models. Reminds me of "The new Mac mini is certainly coming" when they just abandoned the minis for years upon years. At least the M4 seems to be mostly consistent barring the Mac Pro.

 

[Pressure]

The problem I have is how they keep staggering the M_ release for certain models. Reminds me of "The new Mac mini is certainly coming" when they just abandoned the minis for years upon years. At least the M4 seems to be mostly consistent barring the Mac Pro.

I agree with this.

The desktop line-up continually gets left behind. It's annoying having to wait up to 6 months with uncertainty and then they do an M3 Ultra while the rest of the products are M4.

 

[Bodhitree]

First you think, they do it for the holiday sales, and then you see something like this…

https://forums.macrumors.com/threads/new-macbook-pro-nears-mass-production-but-four-bigger-upgrades-expected-next-year.2467800/

Perhaps we will still see an M5 announcement with new MacBook Airs and Pros this year, but maybe not.

 

[maflynn]

Perhaps we will still see an M5 announcement with new MacBook Airs and Pros this year, but maybe not.

I dunno, the window is closing quick, if we don't see anything in the next couple of weeks, it highly unlikely. I think mid-october is the latest they announce things.

As you can see from the buyers guide, it seems only the MBP is at end of its projected life span. I don't know if we'll see the M5, but with each passing week, its becoming less likely.

Apple has managed to establish a yearly or nearly-so update schedule

My guess is that cadence will slow down, I don't think annual updates is sustainable in the long run, but then I'm no chip expert.

 

[senttoschool]

First you think, they do it for the holiday sales, and then you see something like this…

https://forums.macrumors.com/threads/new-macbook-pro-nears-mass-production-but-four-bigger-upgrades-expected-next-year.2467800/

Perhaps we will still see an M5 announcement with new MacBook Airs and Pros this year, but maybe not.

They clearly do it because it gets people to upgrade.

The people on internet forums complaining about how upgrades are too fast and feels obsolete too fast are clearly not the mass audience.

The new chips get hype, marketing, fomo. Corporations are also more likely to upgrade their heavy users with ever more powerful computers.

I dunno, the window is closing quick, if we don't see anything in the next couple of weeks, it highly unlikely. I think mid-october is the latest they announce things.

As you can see from the buyers guide, it seems only the MBP is at end of its projected life span. I don't know if we'll see the M5, but with each passing week, its becoming less likely.
View attachment 2560806


My guess is that cadence will slow down, I don't think annual updates is sustainable in the long run, but then I'm no chip expert.

The only way it slows down is if the iPhone series also slows down.

They all share R&D and the same basic building blocks. I don’t see why it would slow down in the foreseeable future

 

[throAU]

I remember hearing at the time of the M1’s release that one of the reasons that Apple ditched Intel chips was because they weren’t supplying frequent-enough product updates. It’s interesting because looking at it a few years later, Apple has managed to establish a yearly or nearly-so update schedule. But has it really helped them in sales?

Almost got it...

The reason(s) apple left intel weren't the frequency of updates, it was the mediocrity of the updates.

Intel reliably pushed out an "update" every year but when that update is same core count (4 for about a decade), 100mhz boost in clock, higher power consumption and some new magic instructions to be able to claim significant performance improvements (but only if the software was rewritten to make use of said extensions) - they aren't really useful updates.


You just need to look at Mac market share in the past 5 years to see how it has helped. Instead of being an oddball machine people tolerate to use macOS, Macs are now class leading performance and nothing can touch them on battery life.


edit:
as far as an M5 MacBook Pro goes, I'd be expecting one in November.

The M4s are still great, but Qualcomm are not standing still, and Apple doesn't want to be comparable to them. They want to maintain that significant performance advantage. Its a huge selling point for the platform at the moment and missing an update will perhaps lead people to doubt their commitment.

 

[Alameda]

The problem I have is how they keep staggering the M_ release for certain models. Reminds me of "The new Mac mini is certainly coming" when they just abandoned the minis for years upon years. At least the M4 seems to be mostly consistent barring the Mac Pro.

Most likely is that this is how long things take. Each new chip version, I believe, depends on a new manufacturing improvement from TSMC, and the advanced Max chips require substantially more effort to build and package.

To be honest, the rate at which Apple has put out new chips is stunning. They’re really, really good at it.

 

[maflynn]

To be honest, the rate at which Apple has put out new chips is stunning. They’re really, really good at it

The cadence has to slow down, the rate of innovation will slowdown just as it has for the iphones, we're seeing less and less new features.

With that said, it appears that the M5 is going to be an impressive improvement.

 

[Alameda]

The cadence has to slow down, the rate of innovation will slowdown just as it has for the iphones, we're seeing less and less new features.

With that said, it appears that the M5 is going to be an impressive improvement.

There are people in Cupertino who have the five year roadmap. And a great deal is based on what TSMC can deliver, and how fast. And that is based on what TSMC’a suppliers can deliver.

For sure, most of what we’re getting from M1 through M4 is performance. Improved process technology will continually improve performance. And smaller transistors can reduce power consumption as well.

THEN, the software teams create software feature roadmaps based on what the added processing power can offer.

 

[deconstruct60]

edit:
as far as an M5 MacBook Pro goes, I'd be expecting one in November.

The M4s are still great, but Qualcomm are not standing still, and Apple doesn't want to be comparable to them. They want to maintain that significant performance advantage. Its a huge selling point for the platform at the moment and missing an update will perhaps lead people to doubt their commitment.

X2 Elite systems are just reference system demos now.

X2 Elite isn't going to ship until 2026. So if Apple ships M5 Macs in 2026 they are late how?
Pretty likely all that will get in January CES 2026 is demos from system vendors that also are not shipping.

More likely it is spontaneous , random tariff of the week that is larger contribution to the delay of the higher average priced Mac product line up. ( Windows laptops aren't going to swim upstream of that either. ) Also getting deeper into "fast enough for most folks" zone . Finally, there is more competition for the leading edge wafer slots at TSMC. the "spend money like drunken sailors on shore leave" AI crowd are throwing money at TSMC too. Starting with the iPad Pro is 'smaller' than the Macs while juggling the leading edge iPhones (which apparently has two different dies at this point). .

 

[OptimusGrime]

I dunno, the window is closing quick, if we don't see anything in the next couple of weeks, it highly unlikely. I think mid-october is the latest they announce things.

The last couple of years they’ve announced new macs in late October. Delivered in November.

 

[!!!]

Most likely is that this is how long things take. Each new chip version, I believe, depends on a new manufacturing improvement from TSMC, and the advanced Max chips require substantially more effort to build and package.

To be honest, the rate at which Apple has put out new chips is stunning. They’re really, really good at it.

The iMac does not use the Max or even pro versions. With the mini/studio, They'll release the MBP with the Pro/Max chip anyways. I get that they will only do occasional Ultra updates, that's understandable given the small production volume, but it just feels off to me when I saw the iMac being sold with an outdated processor compared to the other lineups.

 

[Falhófnir]

Sounds like this month we will get the M5 launched (though not the M5 Pro & Max until maybe January again?) in a new iPad Pro, Vision Pro, and maybe the base 14" MacBook Pro? AFAIK the Mac lineup has never done better in sales than it's doing now, I'd attribute that at least in part to the M series chips being such a step change in what even the entry level computers are able to do.

 

[Bodhitree]

The iMac does not use the Max or even pro versions. With the mini/studio, They'll release the MBP with the Pro/Max chip anyways. I get that they will only do occasional Ultra updates, that's understandable given the small production volume, but it just feels off to me when I saw the iMac being sold with an outdated processor compared to the other lineups.

I agree. It’s like the focus is on the MacBook Pro as being the greatest profit centre, and the iMac, MacBook Air, Mac Mini and Mac Studio are kind of left to be updated in bits and pieces.

 

[throAU]

The cadence has to slow down, the rate of innovation will slowdown just as it has for the iphones, we're seeing less and less new features.

I don't think the cadence has to slow down any time soon, and in terms of processing power, even without new features there is PLENTY of scope for improvement in either raw power or power per watt.

Right now for example my m4 max can't games i want to play as fast as i want with all the features, and it certainly can't do it long on battery. I still need to wait for it to render things. I still need to wait for LLMs to respond. I still can't run LLMs as large as i would like. The fan can still be annoying under heavy stress, Etc.

no new features required to justify new chips/continued release cadence, just better execution of what we already have - there's huge scope there for improvement.

Don't get me wrong - i'm not complaing, and love my m4 max. But thinking we're reaching some sort of end goal with M series chips is.... just not a thing.

 

[pshufd]

I don’t see the need for an annual cadence for any of the chip companies and they could probably save a decent amount of money going slower. Same thing with macOS.

It makes for a nice cadence for OEMs on the Windows side but we have had early, mid, and late for a long time and it’s probably good enough.

 

[throAU]

I don’t see the need for an annual cadence for any of the chip companies and they could probably save a decent amount of money going slower. Same thing with macOS.

Time doesn't stand still.

The fabs are continually advancing to be the best they can be and longer term having your stuff running on a smaller process means either more widgets per wafter or better chips.

You also need to keep your CPU architecture teams doing the R&D and real world testing - or the competent ones leave and go work somewhere else; the ones who don't leave do a lot of sitting on arse and their skills become stale and even more irrelevant. And then you're starting an uphill battle against one or more competitors who built a way better product while you were sleeping.

See: Intel.

 

[maflynn]

See: Intel.

Intel is a good example, in that they tried and failed to artificially adhere to a tick-tock cadence and that failure was the impudence for apple going to ARM

 

[Bodhitree]

It seems to be necessary to refresh chip architecture teams once in a while. Intel managed it once when they went to the first Core chips, but people get stale eventually. Of course the node also has a lot to do with it, but Apple’s improvements don’t come entirely from one or the other. A smaller node gives room for more circuits, bigger caches, a wider bus, more sophisticated branch prediction and so on. Architecture improvements are not independent of node size.

 

[throAU]

Intel is a good example, in that they tried and failed to artificially adhere to a tick-tock cadence and that failure was the impudence for apple going to ARM

Nah.

They didn't try. That's the point. They slept through the 2010s on a manufacturing process that was never going to work. Before the 2010s they slept way too long on p6 architecture relying on their fab division to save the day. IPC improvements between sandy bridge and say... ice lake, if you take special purpose instructions which are rarely used out of the mix are minimal.

Their design department crapped out mostly the same thing with minimal work with clock pushed another 100mhz higher (due to minor fab improvements) and called it a day. For more than a decade.

This is why they're toast today.

Apple meanwhile has spent the past 15 years relentlessly iterating on their CPU designs and the result is that we have devices with unbeatable battery life and performance per watt. The pace of Apple development is where intel was at in the 1980s-1990s. when they became unstoppable. Ditto for AMD since 2016 with their Zen/Epyc/Threadripper lineup.

If apple don't maintain that, someone else will. You NEED TO SHIP NEW PRODUCT. You can't have design teams playing with designs on paper or in the lab or you simply won't know how they scale (both in terms of ability to manufacture, and performance in the real world), run real world software, adapt to change in demands, etc.

 

[pshufd]

Time doesn't stand still.

The fabs are continually advancing to be the best they can be and longer term having your stuff running on a smaller process means either more widgets per wafter or better chips.

You also need to keep your CPU architecture teams doing the R&D and real world testing - or the competent ones leave and go work somewhere else; the ones who don't leave do a lot of sitting on arse and their skills become stale and even more irrelevant. And then you're starting an uphill battle against one or more competitors who built a way better product while you were sleeping.

See: Intel.

You have slippery slope problems with your arguments.

If time truly doesn't stand still, then you'd be better off with 6-month cadences.

I'm running on a Lunar Lake thin and light right now. Good performance, runs cool and quiet and great battery life. CPU performance is a factor but Apple making a faster M5 or M6 isn't necessarily a primary reason for a laptop choice today. Cost, other functionality, display, compatibility are areas that Apple will be competing on in 2016.

 

[throAU]

If time truly doesn't stand still, then you'd be better off with 6-month cadences.

There are practical limits.

Retooling a fab and developing the next gen process takes time. Re-couping your investment on the existing design takes time.

If those things can be done faster, they would.

Year N+1 is not aimed at people running hardware from year N. It's aimed at the market as a whole who on average (at least people who buy new to make money with the hardware) replace 1/3 of their fleet annually on a 3 year cycle, and buy N+2.