Apple Made a Foldable iPhone. I Have Three Questions.

Apple held its September event yesterday at 18:00 BST, and for once I came away from an iPhone launch genuinely interested in an iPhone I have absolutely no intention of buying.

The iPhone Duo is real. Apple has finally made the foldable phone it spent years allowing everybody else to beta-test first, and I actually think it looks rather good. It folds out into a 7.6-inch display, has a 5.4-inch display outside, runs the same A20 Pro as the new iPhone 18 Pro models, supports proper side-by-side multitasking and will even get Apple Pencil support later this year. It starts at £1,999, because of course it does.

There were also the iPhone 18 Pro and Pro Max, new Watches and new AirPods. I could go through all of those feature by feature, tell you how many megapixels each camera has and repeat Apple’s battery-life figures, but Apple already has a website for that and I don’t think anybody needs me to make a worse version of it.

What I actually finished the event with were three questions, beginning with the fairly obvious one: where the hell is the iPhone 18?

There is an iPhone 18 Pro. There is no iPhone 18.

Apple announced the iPhone 18 Pro and iPhone 18 Pro Max today. It announced the iPhone Duo. It did not announce the ordinary iPhone 18.

That is not me missing something in the keynote. There simply isn’t one.

Reporting around the launch suggests Apple has split what used to be the annual iPhone release into two parts. The expensive phones come first — 18 Pro, 18 Pro Max and Duo — while the normal iPhone 18, 18e and the next Air are expected in early 2027. Apple itself has not announced those phones or committed publicly to that timetable, so for now we have the slightly absurd situation where Apple sells an iPhone 18 Pro without selling an iPhone 18.

I can understand some of the logic. Apple’s iPhone range has become enormous, and launching every possible variation simultaneously would mean explaining a base phone, an e phone, an Air, two Pros and now a foldable during the same presentation without the whole thing turning into somebody reading a particularly expensive Argos catalogue.

Splitting the cycle might make more sense operationally. From the customer’s side, it is still weird.

For years, September has meant the new iPhone. Not necessarily the one you personally want and not necessarily one worth upgrading to, but there was an extremely understandable cadence: number goes up, normal one appears, Pro one appears alongside it. This year, the number has gone up only if you’re willing to spend Pro money.

If you own an ordinary iPhone 15 or 16 and have been waiting for the ordinary iPhone 18 as your upgrade, Apple’s answer today is essentially to keep waiting. Maybe moving the ordinary models to spring ultimately produces a cleaner release schedule, and I can see an argument for not dumping six phones into September every year. It is nevertheless difficult not to notice that the first half of this new schedule consists entirely of the expensive ones.

Apple has not cancelled the ordinary iPhone, as far as anybody knows. It has just made the ordinary customer wait while the premium range goes first.

Then there’s the Duo

The Duo is the bit I actually found interesting.

I have never been desperate for a foldable phone. They are cool pieces of engineering, but most have existed in the category of technology I enjoy looking at in a shop and then very happily leave in the shop. I don’t particularly want to spend laptop money on something I can accidentally close around a grain of sand.

Apple’s version is annoyingly compelling.

The passport-ish size when closed makes much more sense to me than simply producing another enormous slab. The inner display looks genuinely useful rather than just phone but wide, particularly with two apps running beside each other. Apple has clearly thought about how iOS should adapt to the different orientations, and being able to prop the thing up for video, FaceTime or StandBy without carrying another stand around is exactly the sort of hardware-software integration Apple tends to be good at.

Then I looked at the controls.

On Apple’s own Duo pages, Touch ID lives in the side button and Camera Control is another physical control placed around the edge of the device. Apple also describes the Duo as comfortable to use one-handed while closed.

Which hand?

I have poor dexterity in my right hand

I should be precise here because poor dexterity and simply being left-handed are not the same thing.

I can use my right hand, but not with the same control or reliability as my left. A design which favours controls on the right side of a phone may be mildly awkward for somebody who just prefers their left hand; for me, anything involving precise repeated presses or coordinated movement can be more difficult.

When the Duo is closed, I suspect most people will work something out. Fingers wrap around phones, Touch ID can recognise different fingers and the closed device is relatively small. I am not going to declare a phone I have never touched inaccessible based on product photography.

Opening it is the more interesting part.

You now have a 7.6-inch device built around a hinge. Its defining interaction is physically moving two halves of the phone relative to each other, and Apple naturally demonstrates that with two hands. If your right hand has poor dexterity and your left hand does most of the precise work, how comfortable is that actually going to be?

Can you open the hinge against a surface with one hand? Once it is open, can you rotate it so the hardware controls are more accessible from the left without the interface becoming awkward? What happens to Touch ID in those orientations? How much grip strength does the hinge need? Can you reposition it without relying on the kind of coordinated movement Apple’s promotional material takes for granted?

Apple is, to its credit, very good at software accessibility. iOS has AssistiveTouch, Voice Control, Switch Control, Eye Tracking and a fairly extensive collection of ways to avoid physical buttons.

That helps with the interface. It does not solve the hinge.

AssistiveTouch can replace a button. It cannot unfold a phone.

That is what makes the Duo interesting from an accessibility perspective. A slab phone has physical requirements too, but a foldable makes one of those requirements central to the product. It is not enough for iOS to become accessible once somebody has already managed to get the hardware into the right shape.

Maybe this is completely fine in practice. I have not held one, and perhaps the hinge is light enough to manipulate against your body or another surface without much trouble. There may also be orientation behaviours or accessories which make the whole thing much easier than it appears in Apple’s demonstrations.

I would just like Apple to show that.

“Comfortable to use one-handed” is quite a claim for a device whose defining feature involves unfolding it.

Apple also basically Sherlocked Bevel

The Watch part of the event produced another familiar feeling.

Apple Watch Series 12 and Ultra 4 now have Readiness, which uses activity, vitals and sleep to produce a score intended to tell you how prepared your body is for the day. Apple has also introduced Health Age, which tries to place various health measurements in the context of your chronological age. Apple explains both in its announcement of the new health and fitness features.

If you use Bevel, that probably sounds familiar.

Bevel already has Recovery, Sleep and Strain scores, a real-time Stress score, Energy Bank and Biological Age. The app’s entire appeal is taking the large pile of health data an Apple Watch already produces and turning it into something more useful than individual graphs.

In other words, Bevel has been Sherlocked.

Not entirely. Bevel does much more than produce a recovery number, and Apple adding first-party versions of a couple of its ideas does not erase the rest of the app. If anything, Apple putting Readiness into watchOS validates the basic premise: most people do not wake up wanting to inspect their HRV graph. They want to know how knackered they are and whether going hard at the gym today is a stupid idea.

This is also what platforms do. Third-party developers find a useful way of working with the platform, the platform owner sees that people like it, and eventually some version turns up as a system feature. There is a reason Apple has supplied the industry with its own verb for this.

I do not actually object to Apple adding Readiness. If the Watch is already collecting all this data, making better use of it in the operating system is an obvious improvement.

What interests me is where Apple draws the hardware line.

Apple ties its version of Readiness to the new health-sensing hardware in Series 12 and Ultra 4, including much more frequent heart-rate and HRV measurements. That may make Apple’s metric better than anything it could have produced from previous Watches, and with health features there is a perfectly reasonable argument that Apple should not publish a number unless it is confident in what that number means.

But Bevel already produces a useful approximation from data older Watches can collect.

So why does Apple’s version need to be a cliff?

Why not give an older Watch a less detailed Readiness score based on the information it has, while explaining that Series 12 produces a better one because its sensors collect substantially more data?

There may be a perfectly good reason. Apple may have validated its algorithm specifically against the newer sensing system and decided that anything less would be misleading.

The interesting part is that we have seen this pattern before.

On my iPhone.

My Mac can do this. My iPhone can’t.

I have been using the new Siri AI features for about two months on my Mac mini M4, slightly ahead of the normal release cycle because I’m running Apple’s public beta.

I cannot use them on the iPhone sitting next to it because that phone is an iPhone 15.

Not an ancient iPhone. Not something I found in a drawer with a 30-pin connector attached. An iPhone 15.

Apple Intelligence supports the iPhone 15 Pro and Pro Max but not the ordinary iPhone 15. On the Mac side, support goes all the way back to M1.

That makes the contrast particularly obvious for me because I have already been living with the newer Siri. I know what it is like when it has the extra context and capabilities, because the public beta put those features on my Mac before the final release. Then I pick up my phone and immediately drop back to the older experience.

Same Apple Account, same ecosystem, same assistant, completely different capabilities.

There is a technical reason. The ordinary iPhone 15 has an A16 and 6GB of RAM, while the 15 Pro has an A17 Pro and 8GB. Apple has said from the beginning that running its models on-device depends on memory, bandwidth and Neural Engine performance.

I am not claiming Apple has a secret switch somewhere labelled:

if poor_person_phone: siri_stupid()

The hardware requirements are real.

What bothers me is that Apple chose the hardware too.

The base iPhone 15 having 6GB of RAM while the Pro had 8GB was not an unavoidable fact of nature. Apple decided how much memory went into each model. Nobody surprised the company a year later by inventing machine learning. Neural Engines had been part of Apple silicon for years, and Apple obviously knew it was investing heavily in AI before Apple Intelligence appeared on a keynote slide.

A limitation can be technically genuine now and still come from a product decision made earlier.

The Mac comparison makes that especially obvious. Apple is not simply reserving AI for whatever shipped this year, because an M1 Mac from 2020 still qualifies. Apple silicon Macs started with enough headroom that the software could grow into them.

The base iPhone 15 did not.

That is still Apple’s problem.

Bevel makes the question harder to avoid

This is why the Bevel comparison matters beyond the usual joke about Apple Sherlocking an app.

Bevel is a third-party developer working with the health information Apple exposes from hardware people already own. It cannot add a new sensor to your Watch. It cannot decide that the answer to a difficult engineering problem is to sell you next year’s chip. It has to look at the data available and work out what useful experience can be built on top of it.

Apple has a different set of choices.

Sometimes new hardware is genuinely required. My iPhone 15 cannot unfold, and no amount of clever software is going to give it the Duo’s second display. An older Watch cannot suddenly sample a sensor more frequently than the hardware allows.

The murkier question is whether a feature needs to disappear entirely when the old hardware cannot provide the best version of it.

Bevel can produce recovery information from existing Apple Watch data. Apple’s new Watch can produce a more sophisticated Readiness score from better data. Those two facts can both be true.

Likewise, perhaps Siri AI cannot run locally on an ordinary iPhone 15 at a speed Apple considers acceptable. That explains why my phone cannot run the full implementation. It does not automatically answer whether there could have been a smaller model, more selective cloud processing or some reduced version of the experience.

Apple may have good answers to all of these questions. My issue is that “requires X or later” collapses several very different situations into one neat compatibility note.

Sometimes something genuinely cannot work.

Sometimes it could work badly, and Apple has decided badly is worse than not at all.

Sometimes Apple could build a reduced version but has decided that doing so is not worth the engineering effort.

And sometimes the clean hardware cutoff is also extremely convenient for a company whose business involves selling the hardware on the other side of it.

Those are not the same thing.

Supported, except for the interesting bits

This is the bit of modern Apple software support I find increasingly difficult to judge.

Traditionally, asking how long a phone would receive updates gave you a fairly useful idea of its lifespan. If an iPhone kept getting iOS, it kept getting the new iOS.

That is less true now.

My iPhone 15 runs the current operating system. I can visit Apple’s page for that operating system and watch demonstrations of its headline features. Then I hold down the side button and get a different Siri because my phone does not meet the Apple Intelligence hardware floor.

An older Apple Watch can receive the current watchOS while newer Watches get a health layer which becomes one of the major reasons Apple gives you for caring about that release in the first place.

There is nothing inherently wrong with that. Software gets more demanding, sensors improve and eventually old hardware cannot keep up. I would be more suspicious if Apple claimed every feature could run identically forever.

What feels different is how quickly “supported” can stop meaning “gets the things Apple is advertising”.

An iPhone 15 does not feel old.

A Watch being replaced today certainly did not feel old yesterday.

If major software features increasingly depend on hardware that is one or two generations old, then the question “how many years of updates does this get?” stops telling me what I actually want to know.

What I want to know is how many years the device will remain part of the version of iOS or watchOS Apple talks about on stage.

That is a much harder thing to put on a spec sheet.

It should also change how we look at hardware when it launches. Six gigabytes of RAM did not make the iPhone 15 a bad phone in 2023 and it does not make mine a bad phone now. It still does basically everything I bought it to do.

But in hindsight, six gigabytes also meant Apple’s mainstream phone had very little room for the software Apple itself was already building.

When the eventual base iPhone 18 arrives, I am going to care a lot more about that sort of headroom than I used to.

Not just whether it is fast today.

Whether Apple is building it to participate in whatever iOS becomes three years from now.

Three questions, really

That is what I ended up thinking about after the event.

Where is the ordinary iPhone 18?

How well does a foldable iPhone actually work if your right hand has poor dexterity and your left has to do most of the precise work?

And when Apple says the interesting new software needs newer hardware, how often does “needs” mean physically impossible, how often does it mean Apple does not like the degraded experience, and how often does it mean Apple simply chose not to build one?

The iPhone 18 Pro looks extremely good, although I’m probably not the person Apple is trying to sell it to. I’m more of a base iPhone guy: give me the regular model, the sensible price and the features that make it a good everyday phone. The Duo is one of the most interesting pieces of hardware Apple has made in years. Siri AI, from actually using it early on the public beta on my Mac rather than judging it from a keynote demo, is considerably closer to what Siri should have become years ago.

I’m not anti-Apple here. I use a Mac every day, my phone is an iPhone, and I watched the event because I like this stuff. Apple is still very good at making hardware which makes me want to spend ten minutes figuring out how they engineered it even when I have absolutely no intention of buying it.

That is why the gaps stand out.

I want the ridiculous £1,999 folding phone to exist. Technology would be considerably more boring if companies only made sensible things I could personally justify buying.

I also want the ordinary phone.

I want to know whether the folding one was designed around people with limited dexterity before the accessibility software gets involved.

And when Apple builds something like Readiness after developers such as Bevel have already demonstrated what can be done with existing hardware, I want to know why Apple so often seems to start from the newest device rather than asking what version it can bring backwards.

Not because every limitation is fake.

Because “supported” is becoming a much less useful description of what an Apple device can actually do.

My iPhone 15 cannot run Siri AI. My Mac can, and because I’m on the public beta I have already been using it long enough for that mismatch to stop feeling theoretical. Bevel can already take Apple Watch data and tell me something useful about recovery. Apple can now do something similar itself, provided I have the right Watch.

Maybe every one of those boundaries has a sound technical justification.

I would still quite like to know where Apple thinks the boundary between can’t and won’t actually is.