Spatial Computing Doesn't Have a Hardware Problem. It Has a UX Problem
Source: Unsplash The Vision Pro is the best hardware nobody actually wants to use for more than thirty minutes. That gap is not about weig...
Source: Unsplash
The Vision Pro is the best hardware nobody actually wants to use for more than thirty minutes. That gap is not about weight, chips, or pixel density. It is an interaction design failure, and Apple is finally admitting it.
Apple's spatial computing bet is stalling, and the reason has very little to do with the front loaded weight or the $3,499 sticker shock. Apple has sold roughly 1 million Vision Pro units total since launch, most of them to enterprise customers, with only 45,000 units moving in Q4 2025. Apple cut Vision Pro digital ad spend by more than 95% across key markets in 2025. Mike Rockwell, the engineer who built the thing, was reassigned to Siri and is reportedly considering leaving Apple as soon as next year. After eight years designing enterprise SaaS and AI native interfaces for Fortune 500 companies, I can tell you what most of the coverage is missing. Spatial computing is not failing because the hardware is bad. It is failing because nobody has invented a real interaction language for it yet. Pinch and gaze are the resistive touch screen of 2026. They work, technically. They just do not feel like home.
"The cost, form factor and the lack of VisionOS native apps are the reasons why the Vision Pro never sold broadly."
Erik Woodring, Morgan Stanley analyst, January 2026
The Pinch and Gaze Problem Nobody Wants to Name
I have built workflows where users live inside a single SaaS interface for six hours a day. When you design for that kind of duration, you learn one thing fast. Every primary input has to be cheap. Cognitively cheap, physically cheap, mentally repeatable until it disappears. Mouse click is cheap. Tap is cheap. Keyboard is cheap. Pinch and gaze, on the Vision Pro, is not cheap.
Eye tracking sounds magical in a demo. Eight minutes in, your eyes are working harder than they do when you read a novel, because they are doing two jobs at once. They are looking for meaning, the way eyes normally do, and they are simultaneously the cursor. Your brain has to constantly decide whether the next eye movement is a glance or a click. That is a tax. It compounds every minute you wear the device.
Then there is the pinch. A pinch is a gesture designed for one shot intent. It is great for triggering. It is terrible for sustaining. Anyone who has tried to write a long email in Vision Pro knows what I mean. You can do it. You can do it the same way you can write a novel on a typewriter with one finger. The fact that it is technically possible is not the point.
This is the part that frustrates me as a practitioner. We spent twenty years figuring out the cognitive economics of touch. Why a 44 pixel target works and a 30 pixel one does not. Why scroll inertia matters. Why edge swipes feel safer than corner taps. Apple shipped Vision Pro with almost none of that homework done for spatial input. The visionOS gestures feel like a beta of an idea, not a refined system.
Spatial Computing Has No Equivalent of Swipe
Every successful computing platform earned a defining primitive that became invisible. Mouse won because of drag and click. Touch won because of swipe and pinch to zoom. Voice assistants stalled for a decade until wake words became reliable. The platform does not become a platform until the primitive disappears into your muscle memory.
Spatial computing has not earned its primitive yet. Three years after the Vision Pro reveal, there is no gesture that feels native to the medium the way swipe did to glass. The industry has not converged on what the equivalent of "pull to refresh" looks like in 3D. There is no consensus on how you switch contexts, how you commit a destructive action, how you express undo, how you handle multi step input.
I wrote a piece on Medium last month called "How to Build AI-Native Experiences: 14 Mindset Shifts for Product Teams", and one of the shifts is that interaction primitives are not optional decoration. They are the platform. If you skip that work, every app on the platform feels like a port. That is exactly what visionOS apps feel like today. iPad apps in a floating window. Web pages projected on glass. Nothing born of the medium.
The Numbers Apple Would Rather You Not Stare At
Let me lay out the actual scoreboard, because the broader spatial computing market is telling two completely different stories at the same time. Apple's bet is in trouble. The category is not. That distinction matters for any product team trying to figure out where to invest.
- 45,000 units in Q4 2025: Apple's holiday quarter Vision Pro sales, according to production reports leaked in early 2026. For context, Apple sells more than that many AirPods in a few hours.
- 95%+ ad spend cut: Apple slashed Vision Pro digital marketing across key markets through 2025, a quiet signal that the company has stopped trying to convince consumers.
- ~1 million total devices: Apple has reportedly sold around one million Vision Pros since launch, and most of them landed in enterprise hands. SAP deployed about 100 units. CAE uses them for pilot training. Porsche uses them in product design review.
- $221.56 billion projected spatial computing market in 2026: Forecast to climb to roughly $1.066 trillion by 2034. The category is growing fast. Apple is not capturing it.
- Meta Quest owns roughly 80% of the VR market: The category leader is shipping a $299 device, not a $3,499 one. The lesson is not subtle.
- Mike Rockwell reassigned to Siri (April 22, 2026): The architect of the Vision Pro is now running an AI division. The signal Apple is sending its own employees is louder than any keynote.
Read those numbers together and a pattern jumps out. The consumer side of spatial computing has collapsed for Apple specifically, while the underlying category is doing fine. That is not a hardware failure. That is a positioning and product fit failure. Apple built the most capable spatial computer ever made, and consumers responded the way they always respond to overbuilt tools. They went elsewhere.
What Enterprise Knows That Consumers Do Not
Here is the part of the story I find most interesting as someone who designs for enterprise SaaS. The Vision Pro is succeeding in pockets, and those pockets share a structural pattern. They are discrete, time bound, and high context.
CAE flight training is forty five minute scenario sessions. Porsche product design review is a one hour walkaround. Surgical training in Vision Pro is a thirty minute simulation with explicit start and stop. Visage Imaging in radiology is a few minutes of focused review per case. Notice anything? None of these are general purpose computing. None of them require typing a single sentence. None of them ask the user to live inside the headset for a full work day.
This is the lesson Apple keeps refusing to learn in public. Spatial computing is not a replacement for the laptop. It is a specialty instrument. The marketing pitched it as the next Mac. The actual product market fit is closer to the next surgical microscope. Excellent at one thing, awkward at almost everything else.
The AI Layer Just Changed the Math Completely
This is where the timing gets ugly for Apple. Vision Pro launched into a world where AI native interfaces were already eating the desktop metaphor. By the time the headset shipped, voice plus intent plus ambient agents had quietly redefined what "rich personal computing" even means.
I wrote about this for reloadux in March, in a piece called "AI's Impact on Customer Expectations: SaaS Founder's Guide". The core argument applies directly here. Users no longer want more interface. They want less interface that does more. A $300 wearable with a microphone and a context graph can already do half of what a Vision Pro promises, without putting two pounds of glass on your face.
That is exactly why Apple is reportedly reallocating resources from the Vision Air toward Project Atlas, a smart glasses initiative aimed at the Ray Ban Meta market. Smart glasses are an AI delivery vehicle. Vision Pro was sold as a computing platform. Those are very different product categories, and Apple picked the wrong one to anchor on.
Meta saw this earlier. Their bet on Ray Ban Meta plus the Quest line is essentially two different product strategies running in parallel. One is for ambient AI augmentation. The other is for immersive experiences. Apple tried to do both with one device. The compromise satisfied neither audience.
What Designers and Product Teams Should Take From This
If you are building anything in the spatial, ambient, or AI native space right now, the Vision Pro story is not a cautionary tale about hardware. It is a cautionary tale about three specific things, and they apply whether you ship glass, screens, or APIs.
First, the interaction primitive matters more than the resolution. Apple poured engineering into 23 million pixels and almost none into figuring out how typing should work in 3D. The result is a device that looks like the future and feels like a kiosk. If you are designing AI native interfaces, the primitive is the conversation. If you skip the work of making your conversational layer feel inevitable, the rest of the product cannot save it.
Second, enterprise pilots reveal the truth that consumer launches obscure. The signal in spatial computing was always coming from CAE and SAP, not from the influencer Vision Pro try ons. Watch where the dollars get spent in long horizon use cases. That is where your product market fit hypothesis lives.
Third, AI is rewriting what device class a user actually needs, faster than hardware companies can respond. Every team building a hardware first product right now should ask one question. If a smart pair of glasses with great voice could do 60% of what we do, do we still have a moat? For Vision Pro, the answer is uncomfortable. For a lot of enterprise SaaS tools, the answer is the same.
The single biggest design mistake of the Vision Pro era is mistaking spatial output for spatial input. Apple solved how to render the future. Nobody, including Apple, has solved how to navigate it.
Where This Goes Next
Vision Pro 2 is reportedly arriving late 2026 at $2,499, with a lighter design, the M4 chip, and no external EyeSight display. That is a respectable iteration, but the strategic question is unchanged. Without a new interaction primitive, a cheaper headset just means a cheaper version of the same friction. The price was always the second problem.
The honest read on the next twelve months is that Apple is going to quietly transition Vision Pro from "the future of computing" into "a premium enterprise tool." That is not a failure. SAP, CAE, Porsche, and the surgical training market are perfectly happy paying enterprise prices for tools that work for narrow tasks. It is just a far smaller story than the one Apple wanted to tell.
Meanwhile, the real spatial computing future is probably going to ship from a much less obvious direction. AI native wearables. Ambient agents that live in your earbuds and your glasses and your watch, sharing one context graph. The headset becomes a peripheral, not a platform. That is the version of the future that actually fits into a human's day.
Apple knows this. The Mike Rockwell to Siri reassignment is the loudest possible internal signal that the next generation of personal computing is being built on AI primitives, not glass. The company that taught the world to swipe is now trying to figure out what the next universal gesture looks like, and it almost certainly is not pinch.
What do you think? Are you using Vision Pro for actual work, or did it become a $3,499 paperweight after the first month? Drop your honest take in the comments. I want to hear how this is playing out in your own product or design practice.
Sources:
1. Hypebeast, "Apple Vision Pro Faces Cuts as Spatial Bet Stalls" (Jan 2026): https://hypebeast.com/2026/1/apple-vision-pro-faces-cuts-as-spatial-bet-stalls
2. MacRumors, "Vision Pro Creator Mike Rockwell Has Considered Leaving Apple" (Apr 22, 2026): https://www.macrumors.com/2026/04/22/vision-pro-creator-considered-leaving-apple/
3. Fintool News, "Apple Vision Pro Production Halted as Sales Collapse" (2026): https://fintool.com/news/apple-vision-pro-production-halt
4. Winbuzzer, "Apple Halts Vision Pro Production as Sales Collapse, Pivots to Smart Glasses" (Jan 2026): https://winbuzzer.com/2026/01/03/apple-halts-vision-pro-production-as-sales-collapse-pivots-to-smart-glasses-xcxwbn/
5. Web And IT News, "Valve Brings Steam To Apple Vision Pro" (Apr 13, 2026): https://www.webanditnews.com/2026/04/13/valve-brings-steam-to-apple-vision-pro-and-the-implications-for-spatial-computing-are-enormous/
6. Precedence Research, Spatial Computing Market Forecast 2026 to 2034: https://www.precedenceresearch.com/spatial-computing-market
7. Ahmad Ullah, "How to Build AI-Native Experiences: 14 Mindset Shifts for Product Teams", Medium / Bootcamp
8. Ahmad Ullah, "AI's Impact on Customer Expectations: SaaS Founder's Guide", reloadux.com/blog
