WWDC 2026: Filling In The Puzzle (Preview)

My “games” metaphor for WWDC is a jigsaw puzzle, not the traditional bingo card or the modern hotness of a draft. I’m interested in how Apple’s announcements this year will fill in parts of the big picture that I can only really guess at right now.

I hope that by this time on Monday night, I either have some real answers, or enough additional information to bamboozle myself (as well as a skeptical audience) into thinking that I’ve been right all along.

I hereby present an incomplete list of Ponderables. Not all of these are AI-related but you gotta be patient to get there.

(And please forgive the length of this roundup. Alas, I didn’t have enough time to write less.)

The AI Message

Obviously, Apple needs to convince everybody (with an emphasis on the subset of Us that are developers) that the company has, with the assistance of Google plus both hands and a flashlight, located their strategic path forward regarding AI.

Whatever they promise, they’re going to need to lay out a timetable with clear markers throughout. Whatever they deliver, they need to deliver it on time. Whatever they demo on Monday is going to have to be demonstrated live. Firstly during the pre-recorded keynote, and then hundreds of times over the following few days, during in-person briefings with journalists, analysts, and developers. “Trust us, dude” isn’t going to cut it this time.

Gemini

When Apple cut a deal with Google to build Apple Intelligence’s foundation model on top of Gemini, it was the first big shot of good news about Apple’s AI pursuit since…I guess 2024, when Apple finally agreed with us that maybe the next upgrade to Siri should be more ambitious than just making it say “Let me look up that answer on Wikipedia for you…” with a more natural-sounding voice.

But it’s a little weird that we’ve received so few details about such an important partnership. Apple and Google have both mentioned it during a few events and analyst calls. Each time, it was just a short, curt phrase that added nothing and sounded like the residue of many hours of back-and-forth negotiations between the two companies’ corporate communications teams.

We can be confident that iOS 27 won’t have a big yellow sticker on the front of the box that says “Now! Powered by Google Gemini!” But many other questions are in play:

• Will the new Apple Intelligence merely inherit DNA from Gemini? Or is its underlying model going to be more or less Gemini with the serial numbers filed off?
• Are Google engineers acting as advisors and support crew for Apple’s AI team? Or is the work being done by Google engineers operating inside the Apple campus?
• What base reference release of Gemini is Apple working with…assuming of course that this is even a relevant question? Is it something as modern as Gemini 3.5? And will Apple’s Gemini-powered model receive the benefits of future Gemini foundation upgrades (higher speed, lower resource and power demands, more complicated “thinking,” etc.)?
• Is Apple really planning to only use GeminAppleIntelligence in the interim, until their own AI team perfects a 100% in-house foundation model that can replace it?

Siri

How will Apple define the user’s relationship with Siri? Will Apple position it as a discrete conceptual “space” where users go for AI assistance, whether it’s as an app, or as a Presence waiting inside the Dynamic Island to be summoned forth? Or will Apple position Siri like how Google positions Gemini: a fundamental resource that’s somehow everywhere throughout the user experience, making things better in ways that aren’t always obvious?

Is Apple Intelligence’s WWDC 2024 Messaging Even Relevant Now?

As ever, I insist that Apple still has plenty of time to figure out the role AI needs to play on its platforms; its users are still figuring out the role they want AI to play in their lives (if any).

But things might be moving so fast that Apple can’t catch up. Hell, we’ll all be lucky if we get seventeen months of AI being a practical and invaluable contribution to our digital lives before it evolves enough to provoke that inevitable extinction-level event.

At 2024’s WWDC, for example, Apple promised that Apple Intelligence could deliver a few things that no other company could, viz:

• AI features that are immediately useful and relevant, thanks to all of the personal information that you create and transact with on an hourly basis on your iPhone;
• Ironclad privacy that makes it impossible for Apple or any other agency to use your personal data after the transaction is complete, thanks to Private Cloud Compute;
• An AI that empowers all of your apps, across and between all apps, thanks to App Intents

In the intervening two years, Google has introduced:

• Personal Intelligence for Gemini. If you’re already in the Google ecosystem (you sync your iPhone’s Calendar app to Google Calendar, for example), it’s nearly as good.
• Private AI Compute. It’s not as intense as Apple’s Private Cloud Compute (which is built on a whole new OS), but achieves the same results: private data sent to Google’s AI servers is sandboxed away from any other code, and is destroyed when the transaction is complete.
• Android Agent Development Kit. This is a new framework introduced at Google I/O last week. Essentially, any installed app can behave like an on-device server that uses the open Model Context Protocol for agentic control.

And while there’s a hot rumor floating around that maybe iOS 27 will let you generate Shortcuts based on a prompt…welp, Google demoed a feature that lets users create entire widgets that way, and install them on phones, desktops, and a watch. And Sparks (as yet untested by me), which appears to be kind of like OpenClaw, except it’s easy enough and safe enough for the average user, it requires zero configuration, it works across all devices, and it doesn’t require you to beg, borrow or steal a Mac Mini and keep it running 24/7.

I’m leaning heavy on Google examples because Google I/O was just a couple of weeks ago and I have far more experience with Gemini than I do with any other AI. The point is that 2024 to 2025 was an awful time for Apple to be essentially standing still on AI.

An Xcode For The AI Era

I’m in no way qualified to judge Xcode’s AI tools; I’ve built a few little projects but I’m strictly a tourist.

From this perspective, it seems like Apple’s been working hard to give devs the AI tools that have graduated from Convenient to Critical over the past couple of years. It also seems like it’s been a bit piecemeal.

So I’m keen to see if Xcode is ripe for revolution. Google’s message to developers at I/O this year was that its latest tools are agentic as a snake in a rathole…

(I couldn’t come up with a way to complete that sentence. I panicked. Apologies.)

…and that Gemini can break a project down and delegate sub-tasks to hundreds of sub-agents all the way up and down the development cycle, thus acting more like a project manager than as a coder.

Again, I’m incapable of judging such a claim on my own…but that’s where Google is publicly setting the bar.

Different Pitches for Different AI Customers

The purpose of every OS feature is to help Apple sell more hardware. AI helps them sell more stuff to the average consumer by making devices more helpful, and delivering features that users are increasingly coming to expect.

But! As anybody who’s tried to buy a Mac Mini or Mac Studio recently knows, Apple’s selling Macs to AI researchers, AI developers, and AI vanguard users by the bucketload. I’m keen to see how hard Apple pushes that segment of its customer base. If Apple crafted a message that they’re in the same tier as Nvidia as a maker of AI hardware, they’d certainly get away with it.

New Hardware?

At this point we’ve been trained not to expect any major new hardware announcements at WWDC. At best, Apple fills a couple of vacant minutes in the runtime by revealing a performance upgrade to a Mac or MacBook that’s valued by developers.

This year, we could see two kinds of exceptions. The minor one would be the announcement of a new CPU that underscores Apple’s authority as a top-tier maker of AI chips. Annnd maybe a couple of new Macs to wrap around those chips.

…Or Very New Hardware?

The iPhone Fold would be the major one.

(I know, there’s a rumor that the name of the folding iPhone will be something else entirely. I won’t use it here. I believe in the old superstition that we give power to ancient evils by invoking them by name.)

Rumors of an end-of-2026 debut are running hot. Other rumors say that the iPholde is encountering multiple — but solvable — manufacturing issues.

Apple will follow the example of the first iPad, Apple Watch, and the Vision Pro: an announcement/reveal, followed by a retail release some months later. They have to. As with those other devices, Apple needs to give developers the tools and the simulators they’ll need to support the new device.

(This schedule also helps Apple control the messaging once the FCC paperwork is a matter of public record. Though the paperwork might not necessarily reveal more than Apple’s imminent release of Some Type Of Phone.)

Also? This isn’t the sort of device that Apple (or carriers) can camouflage inside a shell when they take it out and about for real-world testing. Apple will want the public’s first peek at the iPholde to be in a slick, expensive video with a bangin’ soundtrack. Not via a furtive photo taken on a subway with a cheap Samsung phone’s digital zoom.

If Apple is confident of a late-2026 release date for the iPholde…sure, they might choose to take the wraps off of it at WWDC, and release a new version of Xcode that devs can start building iPholde apps with straight away.

I doubt it, though. The other similarity between the iPholde and those other devices is that it’s a whole new thing. Apple will almost certainly choose to tell the device’s story at its own event, free from distractions, like they did with the Watch and the iPad.

Still…they did launch the Vision Pro at WWDC, didn’t they? Hmm.

Speaking of the Vision Pro: everybody’s going to be picking apart all of the week’s session videos for some vague detail or another that we can convince ourselves betrays Apple’s plans for smart glasses. Good luck to us all!

Conjunction Junction

iOS/iPadOS and MacOS (yes, I insist upon the initial capital M) used to be such distinct entities that one might have worried that Apple was content to allow the Mac to gradually fade away.

Apple’s been nudging those platforms closer to each other every year. First came interoperability between devices. Then came continuity of operations, and then consistency of experiences.

And now we might be at the start of a new era: Conjunction. My iPad Pro has a freaking menu bar, for the love of God!

I’ll be looking to the WWDC keynote for signs that this trend is accelerating or receding. I’m not rooting for any particular outcome. I’ve seen both Android and Windows pull off this “one UI for every surface” thing successfully. But it’s an interesting trend.

Back to the iPholde. Apple’s about to have three distinct mobile platforms, apart from the Apple Watch:

• iPhones
• iPads
• Things that can be an iPhone-like thing or an iPad-like thing, depending on the task-by-task needs or the carefree whims of the user

I wonder if Apple’s tempted to unify all three operating systems under a single name? iPhone and iPad apps are substantially the same, aren't they?

Beyond the iPholde, if Apple isn’t investigating adding a desktop mode to the iPhone Pro — connect it to an external USB display for an iPad-like multiwindowing desktop — they damned-well ought to be. Because that would be downright awesome…and another reason to unify everything under a single mobile OS.

Liquid Glass

Overall, the reaction to the new OS design language was…aerobic. Apple chose the correct response to that. They didn’t need to Retreat or Retract…just Revise and Refine.

I’m looking forward to further Refinement. In whateverOS 26, Liquid Glass doesn’t provoke a reaction from me, one way or another. But I sense that it’s on the cusp of becoming something I actually like.

And of course, I can’t help but scrutinize every tiny adjustment and interpret some of them as a damning sign that Apple’s expanding the design language to accommodate as-yet-unannounced hardware.

Safari

The desktop edition desperately needs modernization. Chrome and Firefox are truly modern browsers with truly modern conveniences such as split views and vertical tabs. Safari is still stuck in an era of dancing hamsters and punchable monkeys.

Apple Creator Studio

I never expected the former iWork and iLife apps to get much love, year over year; I always assumed that Apple had lost interest in them.

But now! All of those apps (along with the Pro Apps and a couple of newly-acquired treats) have become part of Services revenue!

I think there’s a good chance that Apple will want to start adding value to the suite on a regular basis.

Flex

Apple has a lot of things to boast about. Will they yield to the temptation?

• The MacBook Neo is a huge hit. And it’s the best kind of hit: it makes many (though not all) similarly-priced Windows notebooks look dopey.
• One of Apple Silicon’s advantages is that it lets Macs and other Apple devices do more with less. Including less RAM. Might they put out the message that “only 8 gigs of system memory” is a far lesser problem for Apple hardware during this chip crisis than it is for Windows and Android hardware?

The Timxit

WWDC is the last major Apple event of the Tim Cook era.

Will the keynote directly address it? Maybe not, but it’s certainly going to be on everybody’s minds.

The Pre-Tape Thing

Finally, I wonder if this is the year I start warming up to the idea of assembling a live, on-campus audience of thousands of developers, media, and other observers to watch a 90-minute video.

I totally understand Apple’s rationale. They were forced to jettison the live event during COVID, and that’s when they discovered how much less stressful it made everything. All across the campus, people seemed to laugh more. People bit into their usual burritos and it was as if they were tasting chorizo for the first time all over again. Those inclined to play keepy-uppy discovered that it was suddenly as if the hacky sack had its own will not to land on the ground. Et cetera.

But a video can never have the same vibe or energy as a live performance. Despite the fact that the live keynotes were plotted and calculated and rehearsed to such an extreme that even God whatever-self would be intimidated into holding off on the earthquake until later…they were still inherently more interesting.