Reality Labs has been shrinking for five months. The press has been reading the cuts as bad news for spatial computing. The press has it backwards, and the man at the center of the confusion is the only person who hasn't figured it out yet.
Everyone in immersive tech is supposed to be sad about Meta this year.
The picture gets painted in installments. In mid-January, Meta laid off roughly 1,500 people inside Reality Labs, about 10% of the division's headcount, and closed three of its first-party VR studios: Armature, Twisted Pixel, and Sanzaru. Camouflaj, the studio behind Marvel's Iron Man VR and Batman: Arkham Shadow, was disbanded later in the year. In late March, a smaller round took several hundred more across Reality Labs, Facebook, global operations, recruiting, and sales. On May 20, Meta announced an 8,000-person company-wide layoff plus 6,000 cancelled job requisitions, with the bulk of those cuts hitting non-AI product roles and corporate functions, and Reality Labs absorbing some collateral on top of what it had already lost. The metaverse unit's budget is being cut by up to 30%. Cumulative Reality Labs losses since the end of 2020 have officially crossed $83.5 billion. Q1 2026 alone delivered a $4.03 billion operating loss on $402 million in RL revenue, which works out to losing about ten dollars for every dollar earned.
The expected reaction from anyone working in spatial computing is to mourn. I'm not going to.
The man who has personally driven Meta's spatial computing strategy for five years is finally being forced to step back from it, and the rest of the industry is about to find out what spatial computing looks like when Mark Zuckerberg isn't the one writing the checks. That's the best news the field has had in a decade. If you don't see it yet, you're standing too close to the press release.
What $83 Billion Actually Bought
Here is what the press almost never says clearly. Meta's $83.5 billion in Reality Labs spending was a distortion of the immersive market masquerading as a subsidy for it.
For most of the last six years, Meta was the single largest check-writer in immersive entertainment, and it spent that capital with the strategic patience of a man who needs to justify a corporate name change. Between 2019 and 2021, the company acquired at least nine VR studios in two years: Beat Games (Beat Saber), Sanzaru Games (Asgard's Wrath), Ready at Dawn (Lone Echo), Downpour Interactive (Onward), Unit 2 Games, BigBox VR (Population: One), and Within (Supernatural), among others. The reported price tag on Within alone was around $400 million. The total ran well into the billions before you factor in studio incentives, exclusivity carve-outs, creator-program payouts, and the various 8-figure checks that never made the press release.
When a single platform owner is writing the largest checks in a nascent market, the market gets the shape that justifies the platform owner's capex. Independent creators who would otherwise have built smaller, sharper, weirder things chased Meta's incentives instead. App store curation steered users toward Meta-financed first-party content. The pricing of indie VR work distorted upward to match the checks Meta was cutting next door. And when the market resisted Meta's preferred direction (see the next section), Meta kept writing checks anyway, because the alternative was admitting that the bet wasn't working.
Renting a market and growing one are different activities.
The bills started coming due in January. That round, the heaviest of the year for Reality Labs specifically, took 1,500 people and closed three first-party studios (Armature, Twisted Pixel, Sanzaru). The March round was a smaller, mixed-division cut. The May 20 announcement was mostly about AI restructuring elsewhere in the company, but it confirmed the direction of travel. The studios that remain are being asked to ship more with less. The CFO, Susan Li, said on the Q1 earnings call that VR investment "will decrease significantly" as the company shifts spending toward wearables. That sentence is the most honest thing anyone at Meta has said about Reality Labs since 2021.
Good. The distortion is ending.
The Horizon Worlds Receipt
If you need evidence that the market never wanted what Meta was building, the receipt is in two halves.
In February 2022, four months after Mark Zuckerberg renamed his company in honor of the metaverse and committed it to being "the next chapter for the internet," Horizon Worlds reportedly had around 300,000 monthly active users. By October 2022, The Wall Street Journal was reporting fewer than 200,000. Meta's internal goal had been 500,000 by the end of 2022; they revised it down to 280,000 and missed even that. The 2023 plan, written by Horizon VP Gabriel Aul, set a target of 500,000 MAU for the first half of 2023 and one million for the year. The leaked memo also noted that most visitors did not return to the app after the first month.
The retention numbers are only half the receipt. The other half showed up in March 2026, when Meta drafted a plan to drop VR support for Horizon Worlds and push the product onto mobile, where the install base was actually growing. CTO Andrew Bosworth walked the decision back later that month after the creators who had built on the platform pushed back. The worlds stayed accessible on Quest. The lesson the creators absorbed from watching the plan get drafted in the first place did not get walked back with them.
You can build a thing on a Meta platform, and Meta can decide in a single afternoon that the thing you built is no longer worth supporting on the hardware it originally shipped on. The March reversal was a concession. The strategic intent had already been ratified inside the company before the backlash forced the walk-back: Horizon Worlds in VR was disposable, mobile was where the growth was, and creators were going to find out about it from a press release. The people who got their access back in March are also the people who learned, in that same month, how easily Meta could change its mind about supporting their work.
That kind of platform trust takes years to build and a single internal memo to break.
So when the dominant first-party can't retain users on its flagship social product after a five-year, billion-dollar push, and can't be trusted by its remaining creators not to flip the switch on a Tuesday afternoon, "spatial social computing" is the wrong target to blame. Meta's version of it is the broken thing here. The market has been trying to tell Meta as much since approximately 2017.
The press is writing "Meta layoffs hurt VR." The press is wrong about which noun is hurting.
The Atari Parallel
Here is the historical pattern this most resembles.
On October 1, 1979, four programmers walked out of Atari and founded Activision. David Crane, Alan Miller, Bob Whitehead, and Larry Kaplan, known internally at Atari as the "Gang of Four" because their games had accounted for roughly 60% of the company's sales, were tired of being treated as anonymous labor by management. They wanted credit, royalties, and creative control. Atari refused. They left, set up shop in a garage, and built the first independent third-party console publisher in history on the Atari 2600 hardware they had just left.
Activision didn't just succeed. It cracked the model open. Within three years, Imagic, Coleco, Mattel, Parker Brothers, US Games, and at least a dozen smaller publishers had followed the same pattern. The 1983 video game crash hit Atari hardest precisely because the independent publishers had eaten the market Atari thought it owned. When Atari finally collapsed in earnest, the talent dispersed further, and the next generation of companies (Electronic Arts, Lucasfilm Games, eventually the entire 90s indie wave) inherited the ground.
You can run the same pattern through almost every dominant-company collapse in tech history. Sega exits hardware in 2001, and within a decade the talent diaspora seeded Crystal Dynamics, Bizarre Creations, and the foundations of several Sony and Microsoft first-party teams. Compaq's decline funneled a generation of PC hardware engineers into the broader Texas tech scene. Bell Labs spent a decade dying and produced, on its way out, an absurd fraction of the people who built the cloud computing era.
The pattern is consistent enough to make a rule out of: a dominant first-party hoards talent and dictates direction until the model fails. The talent walks. The market finally gets to vote.
Meta isn't collapsing. What Meta is doing is releasing several thousand of the most experienced XR engineers and designers in the world from the obligation to build whatever Mark Zuckerberg has decided this quarter is the future of human connection. The shape of the immersive entertainment market over the next thirty-six months is going to look approximately Atari-shaped, scaled to 2026.
The People Who Were Right All Along
Before we talk about where the talent is going, let's talk about what most of the press coverage will not.
According to several people I've spoken with who have worked inside Reality Labs or close to it, the picture inside the organization for the last four or five years looks almost nothing like the picture the press has painted. The engineers and designers there knew exactly what their userbase was asking for: better content tools, lighter hardware, more open ecosystems, and a social product meaningfully different from whatever Horizon Worlds was turning into that quarter. They could tell you in detail why Horizon wasn't retaining users, which content categories were underserved on the platform, and what the deal structures and tooling investments would need to look like to unlock the indie creators who could ship the experiences people actually wanted.
And then they got told no.
The "no" was rarely a single voice from a single corner office. The shape of it was institutional, and it had been baked into Meta's spatial computing org since the 2014 Oculus acquisition. The Oculus team that came over with the deal was VR-native, with game-design DNA, field experience, and the kind of spatial intuition you only develop after hundreds of hours in headset. The Facebook and Instagram product organization that absorbed them brought a different reflex. Defer to behavioral data. Treat the historical FB/IG numbers as infallible. Design for the lowest common denominator of non-gamers, the audience the data said would never tolerate the game-shaped content the VR-native team knew worked.
The tension got worse the further Meta scaled. A rapidly growing product staff included a meaningful number of people who could not comfortably wear a headset for more than twenty or thirty minutes without vertigo, which meant they were shaping spatial UX without ever spending the hundreds of hours needed to develop the underlying instinct. Proposals from the VR-native side got softened or restructured to fit what the FB/IG data said users would tolerate. Some of them just got buried. Senior leadership tried, more than once, to flatten the org and reduce the number of approval gates a project had to pass. Until 2024, tactical decisions still ran through Zuckerberg and Bosworth, which meant anything that landed crosswise with the public-facing metaverse narrative had a hard time surviving. After 2024, the tactical gating delegated to the VPs, but by then the cultural pattern was already set.
This is the bottleneck that explains why the platform underperformed despite the capex. The problem at Reality Labs was structural, not a talent shortage. Permission for the right product moves wasn't waiting in any single office; the people most qualified to make spatial calls kept getting outvoted by the people most familiar with the prior generation's data.
The Reality Labs employees who shipped what they were allowed to ship — the Quest line, the hand-tracking, the inside-out tracking that made standalone VR commercially viable, the foveated rendering, the runtime architecture the rest of the industry has been benchmarking against for half a decade — were the only reason the platform got as far as it did. They spent five years trying to drag the company toward what the market was telling them it wanted, against an executive layer that was actively resisting.
The correct posture toward those people this week is gratitude. They held the line under conditions most product orgs wouldn't survive. The next wave of spatial computing exists because they kept showing up. They were doing exceptional work inside an organization that wouldn't let them finish the job, and the press writing this as a Reality Labs failure is dishonoring them.
What Comes Out of This Diaspora
Now they finally get to.
The talent leaving Reality Labs is the highest-leverage cohort of XR builders on the planet. These are the people who shipped the Quest, the hand-tracking, the optics stacks, the Asynchronous Spacewarp algorithms, and who spent the last several years writing internal memos arguing for what they're now finally going to be free to actually build. The Horizon OS social VR team has spent five years working on one of the hardest problems in software with their hands tied behind their backs. They understand the problem space better than almost anyone alive.
A meaningful slice of them will start companies. The early indications are already there: former senior RL engineers have founded XR startups in the months since the January layoffs, and the trickle is on track to be a wave by Q3. Others will join the spatial computing startups that have been quietly assembling teams for the last eighteen months, waiting for exactly this moment. A wave will go to Apple, Samsung, Snap, and ByteDance. The rest will end up consulting and contracting on work that quietly upgrades the entire field.
What all of them have in common is that they'll be building for the market rather than for whatever justifies a $145 billion AI capex line. That's the entire industry's win. It's the win the engineers who used to work on Quest spent years setting up from the inside.
The Headset Was Not the Problem
Let me head off the most common misreading of this argument before it lands.
The Quest line is one of the most successful consumer hardware platforms of the last decade. Meta has sold tens of millions of headsets. The hardware is good, the tracking is excellent, the price floor is low enough to make the platform genuinely accessible, and the developer SDK has matured into something serious people build serious products on. The Quest is a success by every reasonable measure of a consumer hardware platform.
The piece $83 billion in Meta capex never managed to solve is the content. The hardware got built. The content market that needed to ride on top of it stalled, because Meta's content strategy depended on Meta picking winners with checkbook diplomacy rather than letting the market discover them. You can sell tens of millions of headsets and still leave the average owner with not enough reason to put one on more than twice a month. What you're looking at there is a creator-economy gap, and the way you close a creator-economy gap is by getting a few thousand experienced XR builders out of corporate orgs and into the wild.
Meta is broadening its hardware portfolio at the same time: Phoenix, the ultralight tethered headset; Ray-Ban Display glasses; the Neural Band; the Spatial SDK being repositioned for mobile-first XR. That broadening is the correct call. Different use cases want different form factors, and a company that ships only one form factor loses to a company that ships several. The Quest line stays the flagship. It still has more installed base than any standalone XR product on the market. What it has been waiting for is third-party content depth that makes the hardware indispensable rather than optional.
That is the gap the diaspora closes. The hardware was always going to need a creator economy that Meta's center could not build by itself. We're about to get one.
Mark Zuckerberg, Briefly
In twenty years of running Meta, Mark Zuckerberg has been right about exactly one thing: that an attention economy can be monetized through advertising. Everything else he has personally driven has gone the other way. Facebook Home, the phone shell, lasted months in 2013. Free Basics got banned by the Indian government in 2016 after rolling out to scornful reception across the subcontinent. Beacon, the 2007 cross-site tracking program, settled a class action for $9.5 million. The 2021 metaverse pivot itself, two and a half years later, quietly became "AI was the plan all along."
He told us in October 2021 that the metaverse "will reach a billion people, host hundreds of billions of dollars of digital commerce, and support jobs for millions of creators and developers" within the decade. Five years in, he has missed his own internal Horizon MAU targets by more than 60%, posted $83 billion in cumulative losses, and laid off three rounds of the people who were building the thing he committed to. Within the decade is now within four and a half years. The math is doing its own work.
There is a certain kind of founder who confuses being early with being right. Mark Zuckerberg's metaverse bet was early, and it was also wrong in the most expensive way a tech bet can be wrong: it told the market what it wanted instead of asking. The friendlier read of his record is that he's a generational distributor of an attention economy and a poor diviner of the next one. The less friendly read is that he hasn't said a sincere thing in twenty years of public statements about user experience, and the smart move is to assume he's selling something every time he opens his mouth.
Either read points to the same conclusion. Get him out of the way and the rest of the industry can finally get back to work.
What Comes Next
The next wave of spatial computing won't be built at Meta. It'll be built by people who used to work at Meta, at Apple, at Samsung's recently-staffed XR group, at Snap, at ByteDance, and at the dozens of smaller companies that have been quietly assembling teams for the last eighteen months waiting for exactly this moment. The founders of the next Activision-shaped indie publisher diaspora are the people walking out of their RL exit interviews this June, sketching companies on the train home, and shipping something in 2027 that nobody at Meta would have funded.
This is the news.
The press will continue, for the next several weeks, to write that Meta's spatial computing retreat is a setback for the industry. That's the wrong framing. What it is is a setback for Meta. The industry has been waiting for it since at least 2017. Most of the people I respect in this field, the ones building the actual products rather than running the press releases, have been quietly assuming for years that the moment Meta finally stopped paying for the wrong direction would be the moment the right one came into view.
That moment is now. Mark Zuckerberg has been telling us what the metaverse is for five years. He is finally being quiet about it. Listen. You can hear the rest of us thinking.
Sources: Meta Q1 2026 10-Q filing (SEC.gov); Meta Q1 2026 earnings call (Susan Li remarks on VR investment); TechCrunch, "Meta to reportedly lay off 10% of Reality Labs staff," January 14, 2026; GeekWire, "Meta laying off 331 workers in Washington state as part of broader cuts to Reality Labs division," March 2026; CNBC, "Meta cutting several hundred jobs across Reality Labs, Facebook and other departments," March 25, 2026; CNBC, "Zuckerberg's Meta layoffs memo," May 20, 2026; Bloomberg / The Wall Street Journal reporting on Horizon Worlds MAU goals (2022, 2023); Road to VR coverage of Meta VR studio acquisitions and closures, 2019–2026; Activision corporate history (Wikipedia; Game Informer, "Activisionaries," 2013); Mark Zuckerberg, Connect 2021 keynote (Tech at Meta transcript); TechRepublic, "Facebook Data Privacy Scandal: A Cheat Sheet"; Wharton, "How Facebook Lost Face in India"; Webpronews coverage of Andrew Bosworth's January 9, 2026 Reality Labs all-hands. Author has no current commercial relationship with Meta or any of its competitors named here.
Follow the Front Row
Immersive concerts, behind-the-scenes, and spatial video — don't miss what's next.
No thanks, maybe later