Meta’s Metaverse: How an $80 billion vision collapsed in  5years 

On October 28, 2021, Mark Zuckerberg walked onto a virtual stage and announced that Facebook was dead. In its place: Meta. The vision was audacious. “Our hope is that within the next decade, 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.” For a moment, the world believed him. Tech analysts called it the next internet. Institutional investors opened wallets. And ordinary people, armed with $500 Quest headsets, started buying virtual land. The metaverse was not just a product. It was a whole new address. A frontier. And unlike the actual frontier of space (no $20 million SpaceX ticket required), you could get there for the price of a decent television set.

What followed was the largest speculative land rush since the dot-com boom, except this one happened on servers. Between 2021 and 2022, the combined virtual land market across the four main platforms, namely Decentraland, The Sandbox, Somnium Space, and Cryptovoxels, topped $500 million in annual sales. The average parcel price climbed from $100 in January 2021 to $18,000 by January 2022. People were not buying pixels. They were buying proximity to a promised civilization.

The psychology was identical to real estate on Earth: location, location, location. Snoop Dogg built a digital recreation of his Diamond Bar mansion in The Sandbox, complete with a music venue and members-only parties. He held 122 plots. His neighbour? An anonymous buyer who paid $450,000 for the plot next door. Not to live there. Not even to visit daily. But to be adjacent to celebrity, which in the metaverse, as in Bel-Air, carries a market premium.

The Weeknd, Will Smith, Nas, Paris Hilton, and Gene Simmons were all either ambassadors or landholders. The logic was sound on paper. Wherever the audience gathered, brands and buyers would follow. HSBC bought a plot in The Sandbox. JP Morgan opened the Onyx Lounge in Decentraland, complete with a photo of CEO Jamie Dimon, making it one of the stranger moments in banking history. Samsung launched a full digital recreation of its New York flagship store. Gucci built the Gucci Vault. PwC spent $10,000 on a virtual parcel. Even Barbados purchased an embassy.

MTN, Africa’s largest telecoms group, made a landmark move on February 28, 2022, acquiring 144 plots of digital land in the Africarare Ubuntuland metaverse for an undisclosed sum, a statement of continental ambition that signalled the metaverse conversation had moved well beyond Silicon Valley. Republic Realm set the record for the most expensive single purchase: $4.3 million for 800 parcels in The Sandbox. Metaverse Group paid $2.4 million for 116 plots in Decentraland’s Fashion Street district, planning a virtual luxury shopping district. Tokens.com, their parent company, paid another $2.4 million for separate parcels. Atari landed $4.3 million in Sandbox real estate. Adidas, Warner Music, Miller Lite, Sotheby’s, and Prager Metis International all staked positions. The virtual world had, on paper, a functioning economy.

The collapse was swift and almost surgical. Google search interest in “metaverse” peaked in January 2022, reflecting a 106% year-over-year surge. Then the macroeconomic knife fell. Central banks raised interest rates to kill inflation. The era of cheap money, which had turbocharged NFTs, crypto, and digital land alike, ended. Capital fled speculative assets. The numbers read like a medical chart for a patient who did not survive.

The Sandbox floor price fell 95% from 2.86 ETH in 2021 to 0.13 ETH by 2024. Decentraland dropped 89%. The Metaverse Group’s $2.4 million Fashion Street estate, which once represented the peak of the market, is now valued at approximately $8,900. Republic Realm’s $4.3 million Sandbox city is worth just over $65,000. By 2026, land prices across all major platforms are down 99% from peak, with premium parcels trading between $500 and $5,000 and most ghost-platform assets functionally worthless.

Inside Meta itself, the wound was deeper. Reality Labs, the division carrying the entire metaverse vision, posted losses of $13.7 billion in 2022, $16.1 billion in 2023, $17.7 billion in 2024, and $19.1 billion in 2025. The cumulative operating loss since 2020 has now crossed $83.6 billion. Horizon Worlds, the flagship social platform, peaked at 300,000 monthly active users in early 2022, well below an internal target that had already been revised down from 500,000. By 2024, independent monitoring found as few as 900 daily active users. Roblox, a children’s building game, reports over 100 million daily. The gap is not rounding error. It is civilisational.

The retreat was not announced. It was executed in stages. In late 2022, Meta cut 21,000 employees as its stock collapsed toward $90, down from peaks near $380. In 2023, Zuckerberg declared a “year of efficiency,” and the company stopped discussing the metaverse on earnings calls by Q3 2024. In January 2026, Meta cut 1,500 Reality Labs employees, roughly 10% of the division, shuttered three VR studios including the in-house Ouro Interactive, and retired its VR Workrooms collaboration app. On March 18, 2026, Meta confirmed it would pull Horizon Worlds from the Quest store entirely, with the VR experience going dark on June 15. The platform that was supposed to be the next internet will survive only as a mobile app targeting users who never owned a headset and are unlikely to notice the difference from a dozen other platforms already competing for their attention.

The tools were simply not ready. The hardware asked too much: headsets were heavy, isolating, and expensive. Comfort declined after 20 minutes. The social dynamics never achieved critical mass because the product never became culturally mandatory. Nobody needed to go to the metaverse the way they needed to check Instagram. And without necessity, virtual worlds become what they have always been: niche spaces for the dedicated few, not the platform infrastructure of a generation.

To dismiss the metaverse entirely is to misread the story. The bet was not wasted in totality. Meta’s Ray-Ban smart glasses, powered by AI, have tripled in sales year over year. The company’s Orion AR prototype, unveiled in September 2024, represents a decade of spatial computing research that would not exist without the metaverse investment. The AI infrastructure Meta has built to power its platforms, including the open-source Llama model family and the AI assistant embedded across WhatsApp, Instagram, and Facebook, draws directly on capabilities developed during the VR era. The $83 billion was not entirely incinerated. Some of it became the foundation of what came next.

The question investors are now asking in 2026 is one they should have asked in 2021: is this category real, or is it speculative infrastructure for a market that does not yet exist? Meta alone has committed $115 to $135 billion in capital expenditure for 2026, the majority directed at AI data centers and chips. Microsoft, Google, and Amazon are in the same arms race. OpenAI’s Stargate initiative targets $500 billion in infrastructure over four years.

The environmental dimension compounds the risk. Data centers are projected to consume 1,050 terawatt-hours of electricity by 2026, which would place the sector between Japan and Russia on the global energy consumption table. In Ireland, a European tech hub, AI could drive data center usage to 35% of the country’s total electricity by year end. In Virginia, where data centers already consume 26% of state power, Dominion Energy projects residential electricity bills will more than double by 2039, primarily due to this build-out. Emissions are rising. Amazon reported its carbon footprint grew from 64 million to 68 million metric tons between 2023 and 2024, its first increase since 2021, driven largely by data center expansion. Google’s 2023 emissions represented a 48% increase from 2019.

The parallel to the metaverse is real and uncomfortable. The metaverse also promised civilisational transformation. It also attracted billions in institutional capital. It also had early adopters who built real things inside it. And it also ran on infrastructure costs that never translated into proportionate user adoption. The critical difference, if there is one, is that AI is not asking users to change their behaviour. The AI assistant lives in the apps people already use. The metaverse asked people to put on a headset and go somewhere new. That distinction may make all the difference. Or it may not be enough.