Apple's Skating in the Right Direction, But the Puck is Still Ahead

With the 2026 Worldwide Developers Conference (WWDC), I’m cautiously optimistic Apple is finally seeing what some of us have been saying for years. Having fumbled its big unveilings since at least 2023, while AI whizzed right past it, a glimmer of hope is emerging. Is it well founded?

The company’s last decade provides the model for why I am optimistic and also why I am cautiously so.

The Puzzle Came Together Before

Consider: Apple started this decade at one of its most focused points ever. WWDC 2020 announced Apple Silicon, the chipmaking project impressive enough that even those now six-year-old first releases remain competitive for 2026 workloads. Impressive in isolation, it is even more so when you consider where the Mac had been mired in the years leading up to that unveiling.

Under the talented, albeit at times myopic, Sir Jony Ive, Apple had forgotten its founder’s guiding sensibility: design is how it works. Ive’s too often design-is-how-it-looks direction produced Macs with too few ports, eschewing useful features like MagSafe charging and HDMI output and good keyboards for the sake of ever thinner, but more frustrating designs. Like a person who begins a quest for a healthful slim down and ends up with an eating disorder, Ive without Steve Jobs’s veto to guide him didn’t seem to recognize when he’d gone too far.

The most infamous aspect of this was the butterfly keyboard introduced initially for a novel ultralight MacBook, but then deployed inexplicably to every Mac. A laptop keyboard ought to feel nice to type on and type reliably. Apple had checked both of those points off for decades until that keyboard. No piece captured the mess that was late-2010’s Apple keyboards better than Joanna Stern’s iconic 2019 Wall Street Journal column that intentionally retained all the keyboard-caused typos to visualize the brokenness. Apple relented with an old design christened with the new name “Magic Keyboard.”

The older scissor switch mechanism had nothing genuinely magical about it and yes it had to travel further to work (that’s a feature, not a bug, to anyone who actually types), but it felt magical. Why? Because after a half-decade of slop, this one worked.

Ever since, Apple has resumed its place as producer of some of the best laptop keyboards.

Just seven months before Apple’s 2020 announcement of what would become the M1 processor line, Apple released a new MacBook Pro that was a beautiful mea culpa. The abandoned SD card slot and HDMI port came along with that beautifully magically un-magical keyboard. The company had made good on promises from April 2017 when it had admitted some missteps on the Mac, promises one might note for later reference, came in part via Apple’s Vice President of Hardware Engineering, John Ternus.

By the time 2020’s announcement bore its full 2021 fruit in Apple’s professional line, even the iconic MagSafe charging port — protector of countless laptops from unfortunate falls — had returned in the 2021 MacBook Pros. I’m typing on one of them. Is it chunkier than its predecessors? Yes. But who could complain? It was blazingly fast, functions well and still looked great to boot. Apple was locked in: great industrial design paired with best-in-the-industry chip engineering.

Apple has proven it can do a course correction that doesn’t just return it to its previous form, but exceeded it by any reasonable measure. So optimism right now isn’t mere pie in the sky fanboyism on my part.

The Elusive Search for Vision

But, my caution is also well founded. If Apple had just stayed focused, none of the columns I or others have penned in recent years critiquing the company would have even been warranted. But, in what I can only assume was CEO Tim Cook’s effort to prove, finally, that he too could be a visionary product leader like his predecessor, Apple sidetracked its victorious streak and announced the Vision Pro. As I wrote at the time,

Yes, it looks cool to see windows being manipulated in a “real world” space like Apple demonstrated, turning the whole room into one’s computer “desktop,” but does it help us in a meaningful way? I can’t shake the sense it is the cursed progeny of Sun’s two decade old Project Looking Glass and Microsoft Bob.

Jony Ive had tried with impressive engineering to solve a keyboard problem we didn’t have eight years prior and we all suffered for it. With the Vision Pro, Apple again was turning its considerable engineering prowess to the wrong problem to the detriment of its users. This time the misdirection was even worse, because the Cupertino leadership preoccupied itself with a, well, vision of what was next few want at the exact moment anyone with vision could see the real “next” was emerging rapidly:

How is the company that started the trend of building specialized AI processors into its hardware not focusing on casting vision about machine learning? Charitably, perhaps Apple is doing its classic “wait until we have something better” approach. Uncharitably, Apple has had its head stuck too much in its Vision Pro to see where things are really going.

A better keyboard auto correction using machine learning? Great. The ability to find photos of a specific pet? Nice. But, while Microsoft, Google and the like press ahead with “generative” AI that can summarize information, create artwork and the like, Apple is a no show. We need privacy focused, locally run, top-tier AI to counter the centralized options others are putting forward that could easily invade our lives and privacy at a disturbing scale.

If we’re looking for where Apple got off kilter on Mac hardware design, Steve Jobs’s passing is a sensible marker. Ive is brilliant, but it was Ive’s genius rarified by Jobs own that produced the iconic iPhone, iPod, iMacs and — relevant to the present discussion — what is essentially the form factor of modern professional laptop.

You can tell if a competing laptop was released before or after the aluminum PowerBook G4, because the whole industry was changed by the Jobs and Ive duo’s brilliant choices. Unmoored, however, Ive kept iterating slimmer and more streamlined systems long after those returned were diminished or even outright negative.

At an overall leadership level, it appears perhaps Cook was doing the same. Cook is an undeniable virtuoso when it comes to supply chain management. He oversaw an incredible recalibration of how mass production takes place while growing Apple from an underdog to often the most valuable company in the world. Most organizations would rightly covet having either Cook or Ive in leadership. But just as Ive-without-Jobs struggled to know when to stop streamlining products, Cook-without-Jobs struggled without the founder’s “skate to where the puck is going” sense.

Steve Jobs would not have missed the early 2020’s signs of an AI seismic shift. After all, features such as Apple’s facial recognition tools in photos were early leading edge AI decades before ChatGPT. Jobs would have seen that using Apple’s abilities to, say, make a computer be able to answer “when is Mom’s flight coming in and where can we go to dinner afterwards” was far more worthy of betting its braintrust on than one to “simulate the real world while one’s head is inside a helmet.”

Cook, obviously, did not.

Parlor Tricks Make Not a Strategy

While Cook’s Apple — to the company’s credit — did course correct in 2024 and spend its annual keynote on AI, that Mom and her flight could be tracked in 2024 is little consolation when, in truth, the demo was an illusion with no more real world consequence than a magician’s hacking in half of his assistant. The bad decisions of the past meant Apple could only visualize what it should have been releasing, it had no ability to actually deliver.

Would things have been different if Apple hadn’t put all of its research and development that went into butterfly keyboards, Apple Car and Vision Pro? Consider the rapid iteration of Elon Musk’s xAI, which was founded just three months before the Vision Pro debut and is now perhaps the driving part of parent SpaceX’s record setting, nearly two trillion dollar IPO. xAI is far better than it is given credit for and is only three years old; Apple frittered away its early AI advantages over about a decade of ill-advised projects.

WWDC 2023 could have — should have — been when the company hit a genuinely crowning achievement for the Cook era: the iPhone equivalent of the AI era. Instead, it made a way to strap screens on our head for $3,500.

Yes, the Vision Pro was and is technically impressive. But Apple’s genius has never been in merely being technically impressive. What made the iPhone (and iPad and iPod and the original Mac) great was that even within each era’s technical limitations, the product as it was delighted us all by how well it lived up to the hype. We can’t say that of either WWDC 2023’s Vision Pro nor 2024’s first take of Apple Intelligence.

Even now we don’t know if the company thought the Apple Intelligence it was showing in 2024 was ready for prime-time (and it was just mistaken) or if it just made up in whole cloth segments of what it was presenting, betting it could narrow the gulf between hype and reality thereafter. Whatever the behind-the-scenes intrigue, the result has been undeniable: there is no moment in the time since Jobs return in 1997 that Apple offered a clearer demonstration of pure vaporware.

Apple Intelligence was exactly what the company should have offered and felt like where the best parts of the company’s vision had been pointing for years. Using AI not to have a virtual buddy, but to get the stuff done we need done each day is precisely the right (and obvious) focus. I didn’t need insider leaks or prophetic gifts for my 2023 critique of the Vision Pro, cited above, and its week later follow up to essentially describe much of what the following year’s unveiling attempted. Yet all the wasted time on the wrong visions meant the company was stuck doing the same thing I was: describing, not executing, on what should have been its AI triumph.

The mea culpa that was woven in the subtext of last week’s presentation (sadly not as explicit as those of the hardware reboot of 2017-2019) ought to have been the one from a year ago. But the second prong of focus drift still had teeth to bite us.

Liquid Glass’s Much Needed Windex

Apple did need to rework its platforms to be ready for the era of AI power, and this year it seems like they are doing a lot of that cleanup. But that should have been last year’s story. As I mused on a “Snow Sequoia” release last spring,

The company’s struggle to release its most important new features in years may be more than tangentially related to everything I’ve bemoaned in this column. Reports suggest Siri is actually divided into two different systems — the old, core, limited Siri and a newer one for the latest features — because they haven’t been able to pull off integrating them.

You can put beautiful new windows on your house when the wood is solid; when it is rotten, you need to replace the rotted-out structure first. Snow Leopard’s clean-up paved the way for years of solid, reliable upgrades to MacOS, including many of the flashy features we now take for granted.

Instead, Apple opted for the new windows on the rotten wood last year. And they weren’t even beautiful ones.

I was an early optimist on the new Liquid Glass user interface that debuted in the current operating system releases. I’ve missed the more vividly alive user interface design of early 2000’s Mac OS X’s Aqua. The flattened design language of virtually everyone in the late 2010’s — on and off the computer — was always a bit depressing, after all.

But, what made those iconic Apple products great was twofold: great functionality and great design. Liquid Glass could have captured both, but it hasn’t so far. The structural problems remained, they just got covered up in gloss — the exact opposite of Mac OS X’s Aqua.

Aqua was the beautiful and user-friendly front to a new, powerful UNIX foundation key to modernizing the Mac. Liquid Glass, on the other hand, is a sloppy, user-antagonistic bandage on top of a platform still in need of cleanup. Almost everyone who uses it — user interface snob or not — gets frustrated at key content or buttons that get obscured or covered up. Margins and white space are good things, but Liquid Glass both added needless margins and did them in noisy ways (like sidebars on Mac OS) that only served to make it harder to see everything on screen. Everything got busier, nothing got easier to find.

Perhaps they rushed Liquid Glass just to have something other than AI failures to talk about. Last year should have been that rebuild-the-foundations year many of us were begging for, but no doubt Apple wanted something — anything — that could change the subject from its AI misses.

Error compounding error is not a good strategy, but it is often an irresistible one. A rushed, buggy, visually questionable Liquid Glass only gave a deeper sense that Apple was in a new butterfly keyboard era, not the hope Cook and Co. wanted.

Still, there is reason for hope a year later as we look at the newly unveiled MacOS 27 which even Apple is willing to admit is essentially that new “Snow” release. Not since 2009’s Snow Leopard has Apple spent so much time talking up performance enhancements and bug fixes. That’s a very reassuring thing.

While Liquid Glass still has annoyances, reversions to past designs where it was outright bad are very encouraging — for example, those aforementioned margins. Maybe it can actually be a decent interface for the coming years. If all of last week’s claims are borne out, MacOS “Golden Gate” and its siblings on other Apple hardware may be the software equivalent of 2019’s MacBook Pro: not quite at the full realization of a course correction yet, but butterfly keyboard free enough to celebrate.

And, if the OSes 27 prove to be the revised MacBook Pro, perhaps the second stab at Apple Intelligence — which largely was the same demo as 2024, albeit with real demos this time — may be that software’s 2020 Apple Silicon that takes a reset and turns it into a genuine victory.

Learning on the AI Curve

Of course, we still need to see if Apple really gets what went wrong. AI is changing how so much is done; realigning to where the puck is today is better than skating the wrong direction, but no one gets a prize for facing the right direction. Apple still needs to show a willingness to revise some of its approaches. Even the “one a year big release” model that has been a hallmark of Tim Cook’s time is outmoded now.

The company’s newly announced features catch up with the technology AI leaders were showing off last year, albeit with Apple’s signature polish and integration. That’s fine just now, but AI innovation hasn’t halted, it’s accelerated. It will prove nearly impossible for Apple to ever catch up if bound to the cadence of eras past.

Apple would do well to imitate the speedy urgency of Musk’s xAI, which has done an admirable job of catching up as its competitors were racing forward themselves. That company’s progress — the Terafab venture with Intel and Tesla aimed towards space data centers, rapid iteration of the Grok Build agentic tools (sometimes multiple releases on a single day) and ruthless replacement of models with new ones every few months — come in such a blur it often feels like figments of Musk’s imagination. Except those figments are actually happening before our eyes.

If the iPhone maker wants to lead, it need only look to its scrappy early 2000’s self, when it was unafraid to move quickly and to even kill some of its cash cows (“hi, iPod mini”) for its future cash herds (“hi, iPod nano”), to find that same DNA within. It is the underdog even to xAI here and the time isn’t to rest on laurels like OpenAI (itself potentially a victim to its slow speed that has yielded the moment to a more focused Anthropic).

It also needs to apply the best of Tim Cook’s doctrine — owning its own crucial technologies from the chips to the software and everything in between so it can do amazing things unhindered by third parties — to AI. A telling part of the correction is that Apple is beholden to Google’s AI technology to pull it off. That’s a good part of a correction, but a terrible long-term strategy.

Apple’s been here, too, before. Depending on Internet Explorer and Microsoft Office in the early part of Jobs’s second stint at the company was risky but necessary. Switching to the Intel processors Windows systems used was, too. But they helped the company when its chosen paths proved dead ends. What it didn’t do is simply settle on the choices, though: Apple switched to Internet Explorer and soon started building Safari. Apple switched to Intel and almost immediately bought the company that was the foundation for Apple Silicon.

Staying with Google long term means Apple’s attempt to get ahead of the puck will be at the mercy of one of its primary rivals. We’ll truly know if Apple is getting ahead of the puck again if Apple’s own “Foundation Models” are able to stand on their own in a few years. Investments into them will be a critical first step for a once-VP-of-Hardware-Engineering-turned-incoming-CEO John Ternus.

But, a good first step is at least to be fixing past wrongs and start heading to the right end of the rink. The fresh snow of bug fixes and a coherent strategy on AI are that step. That won’t win the game, but it at least makes a victory possible. [*]

Full Disclosure: Tim has long held some Apple (AAPL) stock and, more recently, bought a bit of SpaceX (SPCX) stock as it went public.

Photocopier No More: The Reckoning with AI Creativity Has Arrived

Two esoteric programming events bubbled up this past week. If you’re not into computer science, they may appear irrelevant to you. They’re not. The arcane managed to bring to life our pressing questions about whether AI can create or is a regurgitation machine.

At Arms Over Anthropic

Remember when everyone on the Right was rightly upset at the government censoring opinions it found distasteful? Somehow that seems forgotten in the other war of this weekend.

DeltaForce 65: Das Wilde Keyboard

One of my all time favorite brands is Metadot’s Das Keyboard. They typically don’t include all the bells and whistles, but they are built so well. But what if a Das Keyboard had all the current extras?

Satechi SM3: Affordable Mechanical Meets Magic Keyboard

Satechi is a long-running name in stylish computer accessories, but one that comes to mind for, say, a Thunderbolt hub, not a keyboard. The SM3 is a strong argument the time has come for them to be a part of the keyboard discussion.