Well, that was fast.
The M3 chip, introduced last October, is already yesterday’s news. We’re living in an M4 world, courtesy of the surprising announcement that the new iPad Pro is powered by Apple’s next generation of chips.
Even if you don’t care about iPads, this announcement will affect the trajectory of the Mac and iPhone in quite a few ways.
The first question to answer is: why so soon? The iPhone gets a new chip generation once a year when the new iPhones arrive. Apple likely wants the M series on a similar cadence, since the two chips are really variations of one another. The rush to the M4 breaks this cycle.
A new chip architecture
The answer is complicated and has a lot to do with the vagaries of chip manufacturing. To simplify: TSMC, Apple’s chipmaker, introduced a new 3-nanometer process last year, which enabled Apple to ship the M3 (and A17 Pro) chips and boast about them being the first major chips to be on a 3nm process.
Apple
The problem is that, even then, the manufacturing process that built those chips was a dead end. TSMC was recalibrating its chip manufacturing and moved from its old “N3” technology to the new (and very different) “N3E” version. The new version is not compatible with the previous method of designing chips, which mandated a wholesale redesign. When Apple says that the M4 is made using a second-generation 3nm chip technology, this is what it’s talking about.
Basically, Apple used the M3 generation to get to 3nm before everyone else, but it knew that it would need a new design for the system TSMC was building toward. Hence the M4–and presumably the A18–were redesigns Apple absolutely had to do.
Why launch the M4 on the iPad Pro rather than a more popular product, like a MacBook Air? Chances are pretty good that at the beginning of a chip cycle, the volume of chips coming out of the factory will be low. The iPad Pro doesn’t sell in remotely the same numbers as a MacBook Air, allowing Apple and TSMC to ramp up deliveries.
The future of iPhone chips
What does this mean for the future of the chips in the next-generation iPhones coming this fall? Last year, the iPhone 15 Pro featured a 3nm A17 Pro chip based on the older TSMC process. The non-pro models used the older A16 processor from the previous-generation process.
It seems unlikely that Apple will want to keep any iPhone on the older chip process, so I’d be surprised if Apple shipped two different chips with the iPhone 16 Pro and iPhone 16. Instead, I’d expect an A18 chip that powers all new iPhones.
However, it’s important to note that the M4 chip that Apple announced on Tuesday comes in two varieties. iPad Pro models with less storage and RAM also have one fewer performance CPU core. It’s entirely possible that Apple will differentiate the iPhone and iPhone Pro with variations of CPU and GPU cores, and even RAM. But I can’t imagine the company building more than a single iPhone chip for this new cycle, given that everything needs to come over to the newer chip fabrication process.
The future of Mac chips
So, what happens next on the Mac? The M3 generation thus far extends only to the iPad, MacBook Air, and MacBook Pro. Bloomberg’s Mark Gurman has suggested that there’s one yet as-unreleased M3 chip (the M3 Ultra, presumably) still to be released.
This means we might be in for a first-time Apple silicon experience: New Macs on an older generation chip released after the newer generation has been introduced. (Presumably a Mac Studio with M3 Max and Ultra options?)
Foundry
But when you look at it closer… I’m not sure if there’s a need for concern. First off, all that we’ve seen from the new M4 generation is the base M4 chip. There’s absolutely no doubt that an M3 Ultra (and for that matter, the current M3 Pro and Max) is far more powerful than the little M4, even if it is from the next generation of Apple Silicon.
And then there’s the big question: how big a leap forward is the M4, anyway? In all of Apple’s promotion of the M4 on Tuesday, it kept comparing it to the M2–since that’s the chip in the previous model of iPad Pro. Which, fair enough–every device is different, and comparing the iPad Pro to a MacBook Air is a little weird. (But it’s also potentially a bit self-serving if Apple has reasons to not want anyone to compare the M3 and the M4.)
While the M-whatever naming scheme makes us want to think of each Apple chip generation as a monolith, the truth is that Apple chips are a collection of different components, including CPU cores, GPU cores, Neural Engine cores, display and memory controllers, and a whole lot more. And not every component gets a major upgrade with every generation.
For example, it seems pretty clear that the GPU cores in the M4 are pretty much the same ones as in the M3, and the CPU cores are only slightly updated (to add some additional AI speed-ups from the M3. Apple tweaked performance by increasing the overall maximum CPU core count of the M4 from eight to 10, by adding two more “efficiency” cores, which use less power than the four beefy “performance” cores.
Given all of that, it’s possible that the base M4 won’t actually be a whole lot faster than the M3. It’s even possible that the two bonus CPU cores were added in order to prevent the M4 from running slower than the M3! TSMC’s new process may be more future-proof and the start of a whole new generation of manufacturing processes, but it may actually be a bit less dense than the previous process–and as a result, this new generation might be more of a sidestep than a big step forward.
Regardless, it behooves Apple to get off of TSMC’s old process and onto the new one as soon as possible, which is why (as Bloomberg’s Gurman has reported) Apple intends to turn the entire Mac line-up over to M4 by the end of next year. Given that the M4 generation has started surprisingly early, I’m wondering if the end of the M3 era will come even sooner than expected.
