The three-year iPhone roadmap that cycled through tech blogs this week is a useful artifact, mostly for people who do not read tech blogs. It names four devices, covers 2025 through 2027, and flattens the distinction between what shipped and what leaked. Two of the four phones are already on the market. The other two are doing mobility work in press copy that no shipped phone has ever matched.
If you run a network, sell into one, or plan radio capacity against a device mix you do not control, the roadmap is not interesting as a forecast. It is interesting as a diagnostic. It tells you exactly which parts of the mobile stack Apple is comfortable showing off, and which ones it still cannot solve.
What the Roadmap Says
The version making the rounds is sourced from a YouTube video and a chain of supply-chain leaks. It claims four devices:
- iPhone 17 Pro (2025). Aluminum chassis, vapor chamber cooling, rearranged internals for thermal headroom.
- iPhone Air (2025). 5.5 mm thick, components concentrated near the top, single rear camera.
- iPhone Fold (2026). First Apple foldable, claimed to be crease free and bump free, with a battery between 5,400 and 5,800 mAh.
- iPhone 20 (2027). Bezel-less display, under-screen Face ID, under-screen camera, timed to the 20th iPhone anniversary.
Two of those have shipped or are shipping this cycle. Two are still rumor. The press coverage treats all four as the same kind of object. They are not.
What the Shipped Phones Confess To
The iPhone 17 Pro moved from titanium to aluminum and added a vapor chamber. Titanium is the material a device company picks when it wants to talk about premium. Aluminum with a vapor chamber is the material a device company picks when the sustained thermal load has gotten high enough that marketing has to take second place to physics.
That load is not coming from games. It is coming from on-device inference. The small models that run local summarization, local transcription, and local photo stack work are small enough to fit in memory and large enough to throw heat at sustained duty cycles. A cooling solution that looks like the laptop treatment is the shape of a phone whose chip is doing ML work that was supposed to round-trip to a cloud endpoint a generation ago.
The iPhone Air is the opposite signal. A 5.5 mm enclosure is a phone with less antenna volume. For mmWave, antenna volume is not a rounding error. For 5G Standalone modem performance under marginal signal, surface area matters. The carrier teams already know which retail zip codes got a spike in dropped connection tickets after September. Thinner sells; thinner supports worse.
The shipped phones together are the confession. Apple knows the ceiling on what the current generation can carry. The press release is the marketing. The phones on the shelf are the data.
What Is Still Rumor
The 2026 and 2027 phones are where the coverage stops being news and starts being preorder copy.
Crease free, bump free. Every foldable shipped in the last five years has had a crease. The crease is not a design problem. It is a materials problem. Glass does not bend. The polymer stack that does bend does so by deforming, and the deformation is visible under a side light. Calling the crease solved in a pre-launch rumor is the same move as calling confidence calibration solved in an LLM rumor. The thing that is hard is the thing the press release will not mention.
The 5,400 to 5,800 mAh battery. That range is plausible, because a foldable has a physically larger volume envelope than a slab. What is not yet plausible is how the pack behaves after 40,000 hinge cycles. The spec sheet that nobody publishes is the one that pairs hinge cycle count with battery flex tolerance, and it is the only one that will matter when the first Folds start coming back to carrier stores eight months in.
Under-screen camera and Face ID. Samsung and ZTE have shipped phones with under-screen cameras since 2021. None of them are good. Image quality degrades by roughly a stop, dynamic range collapses in high-contrast scenes, and Face ID wants a dot projector that does not love being shot through a pixel matrix. Apple has not solved optics that the rest of the industry gave up on. It has more quarters to try. That is not the same as a feature.
The Mobility Read
For the operators reading this, the relevant signal in the 2025 to 2027 roadmap is not the form factor. It is the thermal and radio envelope of the devices in market now.
The vapor chamber is in the 17 Pro because on-device inference is now a sustained load, not a burst load. Two traffic consequences follow. First, a meaningful slice of interactions that used to generate short, chatty TCP sessions against a cloud endpoint are now happening on the device. Second, the remaining round trips are more batched, more scheduled, and more background. The traffic mix gets less spiky and more tidal. Capacity planning against the old waveform will over-provision peaks and under-provision nighttime sync windows.
The thin-phone story pushes the other way. A 5.5 mm radio is a worse radio. Support load lands on coverage teams before it lands on product reviews, and it lands in the tickets that never make it to Reddit. If the Fold ships with the edge-of-hinge antenna layout the patents suggest, the support story gets worse before it gets better, because a hinge is a mechanical stress point that passes through the antenna plane.
Nothing in the 2027 part of this roadmap will ship the way it is described this week. That is not a guess. That is the base rate of consumer hardware roadmaps. The thing that will ship is the part that follows physics. The vapor chamber got bigger. The battery got wider. The radio got thinner.
The iPhone is getting thinner. The network carrying it is not.
Sources: iPhone roadmap coverage, Geeky Gadgets (April 2026). All 2026 and 2027 specifications are unverified rumor. Shipped phone specifications reflect press coverage at launch.