Meta questions the metaverse — $70 billion later

Mark Zuckerberg built a $70 billion ghost town. Now, Meta is quietly killing the lights in parts of the metaverse, trimming Reality Labs spending, and talking up AI and the core apps that actually make money — instead of legless meetings in virtual conference rooms.

According to Bloomberg News, Meta is considering slashing its metaverse spending by as much as 30% in 2026 — although no plans are final yet — after meetings at Zuckerberg’s Hawaii compound laid out a slimmer future for its namesake ambitions. That includes Meta Horizon Worlds, the Quest virtual reality unit, and the broader VR group that accounts for the bulk of metaverse spending. Layoffs could hit as early as January. Shares rose about 4% on the news as investors, who have spent most of the decade begging Meta to stop treating virtual worlds like a blank check, cheered the shift as one more sign that Zuckerberg’s grand experiment is finally being resized to reality.

For three years, the company tried to brute-force another new reality into existence: new name, new stock ticker, new keynote vocabulary about “presence” and “embodied internet.” What Meta mostly produced was a spectacularly unbalanced P&L. Zuckerberg wanted the metaverse. It seems no one else did. Zuckerberg reportedly asked Meta teams to find about 10% in savings, but the metaverse group was asked to cut deeper because the company “has not seen the level of industry-wide competition over the technology that it once expected.”

Reality Labs — the division that makes Quest headsets, Horizon Worlds, and Ray-Ban Meta glasses — has stacked up more than $60 billion in operating losses since 2021, including annual losses of $10.2 billion in 2021, $13.7 billion in 2022, $16.1 billion in 2023, and $17.7 billion in 2024. In the most recent quarter, Reality Labs posted a $4.4 billion loss on about $470 million in revenue.

The metaverse kept minting red ink. The social apps kept minting cash. Facebook, Instagram, and WhatsApp continued to do what they’ve always done — siphoning attention, selling ads, and throwing off cash. In late 2021, when the company rebranded as Meta and announced that the metaverse would be its new “north star,” the social apps were already mature. The metaverse was supposed to be the sequel — the next computing platform, the thing that justified spending like a sovereign wealth fund on VR and AR. What it actually became was a very expensive way to prove that most people don’t want to work, shop, or socialize as floating torsos.

Crucially, Meta isn’t walking away from the hardware. Quest headsets still exist. Horizon Worlds is still there for anyone who wants to attend a stand-up meeting in Uncanny Valley. The Ray-Ban glasses, which continue to sell well, are sticking around — and increasingly framed as AI hardware first, AR fashion experiment second. And Meta recently poached Apple design exec Alan Dye to lead a creative studio in Reality Labs.

But the old narrative, where all of this added up to a new, all-encompassing, capital-M “Metaverse,” is fading out of the investor deck.

The narrative that’s replacing it is much more 2025: AI, AI, and more AI. Zuckerberg has spent the past year recasting Meta as an AI-first company, promising $70–$72 billion in 2025 capital expenditures on data centers, custom chips, and the Llama model family (which, yes, landed to a lukewarm reception). On earnings calls, the word “metaverse” is scarcer; the stars of the show now are “compute,” “capacity,” and “efficiency” — and, of course, "superintelligence." Meta launched a "Superintelligence Lab" and put $14.3 billion into Scale AI for a 49% stake, pitching the move as part of its long-term AI roadmap.

AI infrastructure has partners (Nvidia, AMD, cloud landlords), customers (advertisers, developers, enterprises), and revenue models that Wall Street can spreadsheet without smiling through frustration. The metaverse has a handful of games, some corporate pilot projects, dismal engagement metrics, and a per-unit economics problem, because it’s hard to convince people to live inside a headset when the device is bulky, the content is thin, and all the alternatives — phones, laptops, the actual outside — are right there. Zuckerberg bet that if everyone just wore plastic goggles on their face long enough, culture would catch up.

It never did. Most people didn’t want to hold meetings as caricatures of themselves. They didn’t want to buy virtual real estate from a company that can’t even keep its notifications straight. They wanted better feeds, cheaper ads, and maybe a chatbot that could summarize their email. The rest of the industry read the room and quietly rebranded the metaverse into less embarrassing pieces: “spatial computing” for Apple, “digital twins” for industrial players, “3D collaboration” for whoever’s still demoing virtual factories to bored executives.

Sure, some of the tech bundled into the metaverse era will probably outlive the branding that sold it. Digital twins and mixed reality are useful in industrial settings; smart glasses may yet become a mainstream accessory if they get lighter, cheaper, and better at being invisible. Zuckerberg has already pitched them as the ideal gateway to “personal superintelligence,” the place where Meta’s models are supposed to live once they escape the browser tab.

Meta helped push the hardware frontier, in the way a very rich company can when it decides physics is just another product problem. But Meta also tried to turn a niche into a destiny, and then Zuckerberg seemed surprised when users declined to rearrange their lives around its roadmap.

Did AI kill the metaverse? Yes, but also no. The metaverse struggled on its own. AI simply walked into the same boardroom with a cleaner pitch and a revenue model that doesn’t require everyone to move into Horizon Worlds. Now, the metaverse goes from raison d’être to R&D line item. The ghost town stays on the map. It just has a smaller budget for streetlights, while the servers powering your feed and your chatbots get all of the new construction  — not headsets begging you to abandon the real world for a future almost no one asked for.