What Happened to the Metaverse? An Honest Post-Mortem

I used to write about the metaverse constantly. I built a whole site around it.

Then it collapsed as a concept, and being publicly wrong about something is a good reason to write the post-mortem rather than quietly delete the archive.

What the Metaverse Was Supposed to Be

A persistent, shared 3D space where people would work, socialize, shop, and own digital property. Accessed through headsets. Interoperable across platforms.

Facebook renamed itself Meta over it. Consulting firms published market size projections in the trillions. Brands bought virtual land.

What Actually Happened

The headsets stayed uncomfortable. This is the boring physical reason underneath everything else. Nobody wore them long enough for a persistent world to be persistent.

Interoperability never happened. It was always the least plausible part. Competing platforms had no commercial reason to let users take assets elsewhere, and they did not.

There was no reason to go. This is the real answer. Virtual worlds were solutions looking for a daily habit. Meeting people, shopping, and working already had interfaces that were good enough and cost nothing.

The financial layer poisoned it. Virtual land speculation attached the concept to crypto at exactly the moment crypto collapsed. The word became embarrassing.

And then AI took the attention. Capital and executive interest moved decisively. Meta itself reallocated toward AI.

What I Got Wrong

I evaluated the technology on capability rather than on whether anyone had a reason to use it daily.

That is the same error I made about AI search from the opposite direction, which I wrote up in the real limits of AI search engines. In both cases I was analyzing the technology instead of the user's alternative.

I also underweighted physical comfort. It sounds trivial next to platform economics and it decided the outcome.

And I took the market projections seriously. Those numbers were produced by firms selling advisory services on the transition. I should have discounted them heavily and did not.

What Actually Survived

Gaming. The largest persistent virtual worlds are games and always were. Roblox and Fortnite have the engagement the metaverse pitch described, and they never needed the word.

Industrial simulation. Design review, training, and spatial planning kept going because they solved expensive problems. I covered this in VR in manufacturing.

Virtual production. Film and television use of real time 3D environments grew steadily and quietly.

Phone-based AR. Try-on and placement tools work and are used by millions of people who would never call it spatial computing.

The pattern is consistent. Everything that survived solved a specific problem for a specific group. Everything that died was a general platform waiting for a use case.

The Lesson I Actually Use Now

When evaluating an emerging technology, I ask what a normal person can do with it today that they could not do last year, without buying hardware or learning anything.

If there is no answer, the timeline is longer than the enthusiasm suggests, whatever the market projections say.

The corollary is that adoption is a comparison against the existing option, not an evaluation in isolation. The existing option is usually fine, and fine is extremely hard to displace.

For the parallel story in the adjacent hype cycle, web2 vs web3 for marketers covers the other half of the same period, and betting on early tech covers how I decide what to chase now.

For a reasonable outside account of the collapse, the Financial Times' coverage of Meta's Reality Labs losses tracked the money more honestly than the technology press did.

Will the Metaverse Ever Actually Happen?

Something like it probably will, under a different name, on a device that is not a headset, and on a timeline measured in decades rather than quarters. What will not happen is the specific version that was pitched in 2021: interoperable virtual worlds accessed through goggles, replacing offices and shopping.

The pieces that are genuinely advancing are real-time 3D rendering, spatial understanding on consumer devices, and persistent shared state. Those are the ingredients, and they keep improving regardless of the branding cycle. Gaming platforms already run the most convincing version of a persistent shared world and never needed the term.

What has to change for a broader version: the hardware problem, which is physics rather than software. Weight, heat, and battery density improve on decade timelines, and no amount of investment compresses that much.

The interoperability promise I would simply write off. Competing platforms have no commercial incentive to let users carry assets elsewhere, and nothing in the last five years suggests that changed.

My practical position: stop watching for the metaverse and watch for the device. If a spatial computing device ever reaches the price and comfort of a phone, the software follows quickly. Until then, the applications that matter are the narrow industrial ones that already work, and phone-based AR that millions of people use without ever hearing the word.

Writing the post-mortem is worth more than quietly deleting the archive. I was wrong in public, at length, and the specific shape of that error, evaluating capability instead of asking whether anyone had a daily reason to use it, is now the first thing I check. The filter it produced is in betting on early tech.

Everything that survived solved a specific problem for a specific group. Everything that died was a general platform waiting for somebody to find a use case, which is a description of most failed technology.

What did you believe about the metaverse that turned out wrong? I believed most of it, publicly, and I would rather say that than pretend otherwise.