By
Stuart Kerr, Technology Correspondent,
LiveAIWire
Meta announced in late 2025 that it
was considering cuts of up to 30 percent to its Reality Labs budget in 2026,
its metaverse division, according to Bloomberg
reporting citing people familiar with the talks. The unit that Mark
Zuckerberg bet his company’s future on when he rebranded from Facebook to
Meta in 2021, committing tens of billions of dollars to what he called the
next frontier of computing, is in retreat. Daily active users in
crypto-metaverse platforms like Decentraland and The Sandbox dwindled into
the hundreds at their nadir. Consumer metaverse platforms became ghost towns.
And yet, at exactly this moment of apparent collapse, the underlying
technology that the metaverse was always meant to deploy is maturing,
rebranding, finding enterprise applications, and positioning itself for a
growth trajectory that has nothing to do with virtual real estate
speculation. Understanding what actually happened to the metaverse, and what
is growing in its place, requires separating the concept from the hype cycle
that distorted it.
This guide covers what the metaverse was
and what it has become, what spatial computing is and how it differs from the
original vision, where genuine traction is being found, and what the
technology will realistically look like for businesses and consumers over the
next three years.
What the Metaverse Actually
Was
The metaverse, as Zuckerberg and other proponents
articulated it in 2021, was a persistent, interoperable, three-dimensional
internet where users would spend significant portions of their working and
social lives. The vision combined virtual reality hardware for full
immersion, blockchain technology for digital ownership of virtual assets,
avatars as persistent digital identities, and vast virtual worlds that would
host the social and commercial interactions that currently happen on
conventional screens. Each element had legitimate technical precedent and
genuine research investment. The combination was extrapolated far ahead of
what any of the components could deliver in the timeframe
implied.
The problems that produced the collapse were
fundamental rather than cosmetic. VR hardware remained expensive, heavy, and
isolating in ways that prevented the casual social use the vision required.
Most people do not want to spend hours immersed in a headset, and those who
tried reported fatigue, discomfort, and the specific loneliness of being
physically isolated from the room they were physically sitting in.
Interoperability between platforms never materialised, because each major
player had stronger incentives to build a closed ecosystem than to contribute
to a shared open infrastructure. Digital assets backed by blockchain
technology lost most of their value as speculative interest cooled. And AI,
which offered immediate, practical productivity improvements without
requiring new hardware or new behaviour patterns, absorbed the attention and
investment that the metaverse had been counting
on.
Spatial Computing: The Rebrand That Actually Means
Something
The term that has replaced “metaverse”
in serious technology discourse is spatial computing. Where the metaverse
implied escape from physical reality into a digital one, spatial computing
implies integration of digital information into physical reality. Where the
metaverse required headsets, spatial computing works across a spectrum of
interfaces from standard screens through augmented reality overlays to full
immersion. The shift in framing reflects a real shift in what the technology
is actually good at in 2026: augmenting physical work and physical spaces
with digital information, rather than replacing them with virtual
ones.
Apple’s Vision Pro, released in 2024, exemplifies the
spatial computing approach. It is positioned as a productivity and
professional tool rather than a consumer entertainment device, and its most
compelling use cases involve superimposing digital content onto physical
workspaces rather than immersing users in virtual ones. Enterprise uptake,
while modest in absolute numbers, has been more sustained than consumer
adoption, because businesses have specific, bounded problems where the
ability to overlay digital information onto physical environments, during
manufacturing inspection, surgical planning, architectural review, or remote
assistance, produces measurable value.MIT
Technology Review’s computing coverage has tracked this enterprise
pivot consistently, noting that the platforms with genuine traction in 2026
are those solving specific workflow problems rather than offering
general-purpose digital social experiences.
What Actually
Has Traction in 2026
JigSpace, an enterprise spatial
computing platform, reported 50,000 Apple Vision Pro installations and 80,000
hours of usage by mid-2025, with the platform reducing sales cycles for
immersive product demonstrations from six months to six weeks. That kind of
concrete, measurable ROI is precisely what consumer metaverse platforms could
never demonstrate and what enterprise spatial computing is beginning to
accumulate.
Nvidia Omniverse, which provides industrial
simulation and digital twin capabilities for manufacturing, engineering, and
robotics development, has become one of the most practically significant
spatial computing platforms without appearing in most metaverse coverage.
Siemens, Schneider Electric, and ABB have all built significant industrial
metaverse applications on top of digital twin technology that have concrete
engineering and operational applications rather than consumer social ones.
The common thread across all the cases with genuine traction is specificity:
they address defined problems in defined industries with measurable outputs,
rather than seeking to recreate human social interaction in digital
form.
Generative AI has added a new dimension to spatial
computing adoption. Users can now describe a virtual scene or object and have
AI automatically generate the 3D assets for it, removing the technical
barrier of 3D modelling expertise that previously limited who could create
content for spatial computing environments. This lowers the cost of spatial
content creation significantly and opens up the technology to organisations
that could not previously justify the specialist investment required.
Understanding how ambient
invisible computing is developing alongside spatial computing helps
frame why spatial computing is becoming a convergence point rather than a
standalone technology: as AI becomes embedded in physical environments, the
spatial layer for interacting with that embedded intelligence becomes more
rather than less important.
The Hardware
Reality
AR and VR hardware shipments declined
approximately 12 percent in 2025 due to delayed product launches from major
manufacturers, but analysts project approximately 87 percent growth in 2026
as new product cycles resume and accumulated demand is released. The
longer-term trajectory shows a 38.6 percent compound annual growth rate for
units shipped between 2025 and 2029. These numbers reflect a market that
contracted through a product cycle trough rather than one that collapsed due
to fundamental rejection of the technology.
AI-powered
smart glasses represent the most likely near-term mass consumer entry point
for spatial computing. Google’s partnership with Warby Parker for AI-powered
smart glasses, with $150 million allocated to co-development, and Meta’s
Ray-Ban AI glasses, which have sold significantly better than any of the company’s
VR hardware, both point toward wearables that add ambient AI capability to
physical reality without requiring users to adopt a new relationship with
their physical environment. The glasses form factor has none of the social
awkwardness and physical isolation problems that plagued VR headsets, which
makes its adoption curve look more like the smartphone than the
Segway.
What This Means for Businesses and
Users
For businesses evaluating spatial computing investment
in 2026, the relevant question is whether there is a specific workflow or
customer experience problem where overlaying digital information on physical
reality produces measurable value. If the answer is yes, the technology
exists to address it and the cost of experimentation has fallen
significantly. If the answer is “it seems like it might be useful
eventually,” the track record of the metaverse era is the appropriate
caution: eventual utility and current adoption are different things, and the
cost of maintaining a virtual presence that nobody visits is the specific
lesson of 2022-2024 that businesses should not need to learn
twice.
The Guardian’s
technology coverage of augmented and mixed reality has documented
the pivot from consumer enthusiasm to enterprise pragmatism across 2024 and
2025. The privacy implications of spatial computing, which requires sensing
physical environments to overlay digital content on them, are covered from a
rights and accountability perspective in the
digital resistance movement’s response to AI surveillance.
Understanding how
to design AI experiences that people actually keep using is
directly applicable to spatial computing UX, which faces the same challenge
as every previous new computing paradigm: the technology that works in a
demonstration needs a different design approach than the technology that gets
used every day. The metaverse failed the everyday test. Spatial computing is
being built to pass it.
The Three-Year Outlook: What Is
Actually Coming
The spatial computing landscape over the
next three years will be shaped by three convergences. First, AI-generated 3D
content will eliminate the specialist design bottleneck that has historically
made spatial computing expensive to populate with content. When non-technical
users can describe a product, environment, or training scenario and have AI
generate the spatial content automatically, the cost economics of spatial
experiences for mid-market businesses become viable in ways they have not
been. Second, form factor improvements in smart glasses will bring ambient
spatial computing to a mass consumer audience that has never adopted VR headsets.
The Ray-Ban Meta smart glasses model, which overlays audio AI assistance on
physical reality without requiring wearers to look different from non-wearers
in public settings, has demonstrated that the social barrier to augmented
reality adoption is more significant than the technical barrier. Third,
enterprise digital twin adoption is expanding from manufacturing into urban
planning, infrastructure management, and logistics, creating large
institutional deployments that will drive the platform and tool maturation
that consumer applications will subsequently benefit
from.
The metaverse narrative failed because it
extrapolated from what technology enthusiasts found exciting rather than from
what ordinary people and businesses would actually use. The spatial computing
narrative is more grounded, but it is not immune to the same failure mode.
The applications that will define the next decade of immersive technology are
more likely to look like digital overlays that make existing physical work
easier than like immersive virtual environments that replace physical
experience. That is less cinematically compelling but more commercially
durable. The companies that built product on the former premise while the
hype favoured the latter are the ones with the strongest foundations now that
the correction has clarified which vision was more accurately calibrated to
how humans actually live and work.
For individuals choosing
which AI tools to invest in now, comparing
the leading AI platforms includes understanding which are
developing the most capable spatial content generation features, since that
capability will determine which platform best supports spatial computing
workflows over the next two to three years. The brands that integrate spatial
content generation early will have the library advantage that determines
which platform enterprises build on when the deployment environment matures.
That library advantage compounds, which is why the choices being made in 2026
about which AI platforms to build spatial computing tools on will shape the
competitive landscape of the category for years after the hardware adoption
curve catches up with the software capability that already
exists.
About the Author
Stuart Kerr is
Technology Correspondent at LiveAIWire, covering artificial intelligence,
cybersecurity, and the social impact of emerging technology. He publishes
daily at LiveAIWire.com.