What was the metaverse?

Late last year, Meta confirmed it would effectively be abandoning the metaverse, a nebulously defined project that spurred the company’s 2021 rebrand and has cost it over $70 billion since. At a strategy meeting at Mark Zuckerberg’s Hawaii compound, Reality Labs, the division responsible for the metaverse, was told to cut its budget by 30%, versus only 10% across the rest of the company. Reality Labs’ fate was arguably a long time coming: The division has never turned a profit, with cumulative losses these past five years totalling $73 billion. Wall Street reacted positively to the news, adding $69 billion to its market capitalization. 

You remember the metaverse, don’t you? The next stage of the internet’s evolution: a virtual reality full of legless avatars, sprawling, lifeless, digital malls, and nausea-inducing headsets. Upon the inception of the metaverse, its enthusiasts looked at vast swaths of the economy—gaming, online retail, digital advertising, compulsory Zoom meetings—and said: Imagine we did more of this, but on virtual reality platforms, mediated by micro-transactions and facilitated by cryptocurrency-backed assets. Relabeling the digital economy as the “metaverse” was a simple, elegant move—as well as a deeply cynical effort to rebrand already existing digital markets as the next internet—that allowed forecasts to assume an air of inevitability. Until it wasn’t. Perhaps more urgently now, the metaverse should also be understood as a dress rehearsal for today’s AI boom: The former was to succeed the mobile internet, while the latter now promises to be “more profound” than electricity or fire. Perpetually inflating definitions. A single-minded focus on profit that identifies but fails to address egregious harms. Manufactured narratives about inevitability and technological progress. Burning eyewatering sums on infrastructure for a product nobody wants.

Any of this sound familiar?

Talking it into existence

At the heart of the metaverse derangement was the persistent inflation of its definition. McKinsey & Company proclaimed in June 2022 that the metaverse could generate “up to $5 trillion in impact by 2030—equivalent to the size of the world’s third-largest economy today, Japan.” McKinsey also estimated that e-commerce would comprise $2 trillion to $2.6 trillion of that share. Of the 3,400 consumers and executives McKinsey surveyed, 95% of “business leaders” expected positive impact from the metaverse in five to ten years, while 61% expected moderate changes to their industry’s operations.

Incredibly, McKinsey’s was among the more conservative estimates. A few months before, Citigroup predicted the metaverse would become “the next iteration of the internet, or Web3.” While it would be “community-owned” and governed and guarantee “privacy by design,” it would also have a total addressable market (TAM) between $8 trillion and $13 trillion by 2030, with some 5 billion users to boot, the bank estimated. And one month before that, Morgan Stanley sent an investor note anticipating that the metaverse represented an $8-trillion market opportunity in China alone.

In an essay analyzing Web3 and the metaverse, tech critic Evgeny Morozov observes that a great deal of what was going on at this time was a performance aimed at conjuring new realities into being through language that, itself, spun up visions unmoored from reality. “The advocates of Web3 are quite explicit about this, we’ve got this beautiful map on our hands – all that’s missing is the territory it is supposed to refer to … if there’s no reality, we’ll create one by talking it into existence.”

A mass hallucination

Why was this mass hallucination indulged for so long? Part of it was because profit-driven surveillance and enclosure were core ambitions of the metaverse pivot. When it came to labor, the best-case scenario resembled employers’ platonic ideal: bypassing labor laws through remote work, misclassifying full-time workers due protections and benefits as contractors, paying arbitrage wages across borders, all while subjecting workers to cold, algorithmic overseers.

As for consumers, they would be enlisted into digital sharecropping. Take Axie Infinity, the “play-to-earn” game once hailed as a crown jewel of Web3 and the metaverse. “Managers” in wealthy countries bought expensive NFTs, or non-fungible tokens (remember those?), then rented them to “scholars” in the Global South, who grinded for hours and hours in the game for a few pennies an hour, all in hopes of earning enough to one day become a manager with their own scholars. Was this a new economy? The future of the internet? Or the same old bitter taste?

At the same time, a land grab for virtual real estate broke out. Speculators poured millions into Decentraland, The Sandbox, and other virtual worlds where “land” should, theoretically, be limitless and abundant. Yet the speculators imposed artificially limits, in hopes of inflating valuations of the digital real estate. This would allow investors to realize eyewatering returns on fundamentally worthless assets, like a slice of land in an abandoned virtual world. It would be akin to “buying property in Manhattan, but in a world where anyone could feasibly create an infinite amount of alternative Manhattans that are just as easy to get to. Which means the only reason for users to buy into this Manhattan is if it offers a better service than the others,” as Wired put it. 

The humbling 

Still, the emptiness did not deter Facebook, which rebranded as Meta on October 28, 2021 during Connect 2021, its annual developer conference. 

During the October announcement of Facebook’s pivot to the metaverse, Zuckerberg offered that “the last few years have been humbling for me and our company in a lot of ways.” That’s one word for it. That year, whistleblower Frances Hauge testified that Facebook products had harmed children, torched our democracy, while reaffirming its complicity in genocide in Myanmar and in “literally fanning ethnic violence” elsewhere.

On another front, Apple changed iPhone privacy settings so that users could opt out from being tracked for personalized ads — Meta told investors the changes would cost it $10 billion of revenue in 2022. The impact may have been so steep that the firm is currently accused of “deliberately bypass[ing] privacy rules on Apple iPhones in a bid to boost revenues.”

Amid all this came the metaverse Hail Mary: A transparent, desperate rebrand to sell the promise of “presence” in a virtual world. The pivot was about building a “total service environment” — a closed garden where consumers spend all day exclusively using one firm’s goods and services, a new world where Facebook was not seen as a parasite, but understood to be the landlord — the benevolent god watching over everything. “We should all be concerned about how Facebook could and will use the data collected within the metaverse,” warned Bree McEwan, a VR researcher.  

The physical world was becoming increasingly hostile to Meta’s relentless profit-seeking. Before Zuckerberg preached democratization, Meta spent the past few years busy at work on patents aimed at optimizing ad delivery through eye-tracking, gait analysis (to identify users by how they  walk), and haptic feedback suits monitoring heart rate and emotional arousal. Parents and children were raising concerns about its impact on mental health and social relations. European regulators and American competitors were implementing changes that thwarted data extraction. 

Rise and fall

Yet within a year of the rebrand, there was already trouble in (digital) paradise. By October 2022, Meta’s flagship virtual-reality game, Horizon Worlds, proved so buggy and unpopular that it was placed on “quality lockdown.” There was a time when Horizon Worlds claimed to have 200,000 monthly users (walking back claims of 300,000) and hoped to hit 500,000 by the end of the year. But by August 2023 it wasn’t even clear if there were more than 1,000 daily active users. Other virtual worlds like Decentraland and Sandbox appeared to fare even worse. 

Some may insist that we can’t learn too much from the rise and fall of the metaverse — that it and Meta, more generally, are rogue mutations, aberrations from normal technological development or even from capitalism itself. But Meta is, actually, a more straightforwardly boring company than some critics might have you think. 

Facebook enthusiastically became Meta, and patented surveillance tools were adopted as a means to an end: making more and more of the rhythms of human life legible to markets. This is old wine in new bottles. From its earliest days, surveillance has helped minimize capitalist dysfunction by regimenting labor, stimulating consumer demand, satiating Wall Street’s hunger for reliable returns, and indulging the security state’s demand for total information awareness. 

Meta has been on a vision quest for business ventures that might rival (or bolster) its core advertising business ($51.2 billion Q3 ‘25, up 26 percent year-over-year). It tried and failed to take on the global financial system with a cryptocurrency called Libra, before stripping it down and selling what remained. It tried and failed to enter hardware with Building 8, which became Portal, which became nothing. Lacking his own currency or device, Zuckerberg made a bet that he could graft a virtual interface onto the digital and physical world (while pocketing a few more advertiser bucks along the way).

Aberration vs. symptom

If you are reading this in the year 2,025 A.D., you may have noticed there are many similarities between our former inevitable future (the metaverse) and our current inevitable future (generative artificial intelligence).

While the word “metaverse” was not uttered once on Meta’s most recent earnings call, executives gushed about generative AI and anticipated “notably larger” growth in capital expenditures in 2026 than 2025 — driven primarily by the AI infrastructure overbuild. The company expects to lose $72 billion on artificial intelligence through 2025. Reality Labs is expected to reallocate some metaverse funding to Meta’s Ray-Ban smart glasses — pitched as a new AI hardware product—that have seen huge growth in sales, even as the public galvanizes against the return of glassholes. 

There is the matter of narrative. The metaverse was hailed as “the successor to the mobile internet,” whereas artificial intelligence is “the next general-purpose technology” that will revolutionize human civilization. Just as the metaverse’s future was so obviously entwined with surveillance and enclosure, so too is the project of remaking the digital world for AI agents — regardless of whether they will ever exist, let alone work.

There is the tiny problem of the numbers. The metaverse got multi-trillion dollar TAMs by reclassifying all digital activity; artificial intelligence gets multi-trillion dollar GDP contribution estimates by assuming unprecedented productivity improvements and sidestepping questions about the $2 trillion in revenue it needs by 2030 to justify capital expenditures on AI infrastructure.

There is also the burning question of demand. In the metaverse era, startups offered crypto-tokens and complicated (Ponzi) schemes to artificially inflate demand. Today, tech firms are “investing” billions in AI startups but requiring those dollars be spent on the investor’s own cloud compute. Subsidizing your own revenue growth to impress Wall Street and create the illusion of organic demand is a tale as old as our tech sector’s origins. How will it go this time?

And then there is the question of the fate of our physical world. Intel estimated the metaverse might have required a thousandfold increase in computing capacity, powered by data centers whose energy and environmental costs would be excluded from glossy demos and deks. The metaverse also prominently featured cryptocurrency, which itself demanded substantial amounts of energy. One White House report notes that “from 2018 to 2022, annualized electricity from global crypto-assets grew rapidly, with estimates of electricity usages doubling to quadrupling” landing somewhere between 120 and 240 billion kilowatt-hours per year—on the lower end, that’s more than the entirety of Argentina, but on the higher end this would rival Australia’s electricity usage. 

Had Meta succeeded, we would’ve built out much more energy-intensive computational infrastructure with a growing ecological cost. But we also would’ve fleshed out supply chains to extract and deliver critical minerals—meaning we would likely intensify child and slave labor that already prominent figures into these enterprises. A good thing, then, that Meta abandoned this path. 

Ironically enough, Nvidia offers a bridge between the two worlds: the fusion of the dead metaverse and the living generative AI hype in the “Omniverse.” In The New Yorker‘s 2023 profile of Nvidia chief executive Jensen Huang, he shows off “Diane,” a hyper-realistic avatar with blackheads on her nose and an “uncanny shimmer” in her eyes. “We’re working on that,” the specialist shares with the reporter. The goal is to speak “whole universes into existence.” The writer “felt dizzy” and shared that “I thought of science fiction; I thought of the Book of Genesis.” If that reaction is any guide, Nvidia may well succeed with its proselytizing where Meta failed. 

It would be a mistake to simply celebrate the death of the metaverse. Instead, we should understand why such a delusional fervor took hold so that we can inoculate ourselves as the next one spreads. 

source https://www.fastcompany.com/91467599/metaverse-zuckerberg-facebook-ai