Technologies
Virtual and augmented reality
The strongest numbers in virtual reality are the losses and the clinical trials; the shipment forecasts and adoption claims come from vendor-funded trackers.
Virtual reality has better evidence behind its medical uses than behind its consumer market. Two decent clinical bodies exist: a Cochrane review of 72 stroke rehabilitation trials, and a randomised trial of automated VR therapy in psychosis that found real but small effects. The consumer picture is the opposite. No independent body counts headsets. Apple discloses nothing about Vision Pro, Meta stopped publishing Quest unit numbers, and the trackers that fill the gap are funded by the industry they measure. The clearest consumer number is an accounting one: Meta's Reality Labs division reported a $19.2 billion operating loss on $2.2 billion of revenue in 2025, three years after McKinsey projected up to $5 trillion of metaverse value by 2030. For a practical example of software used to turn activity into operational metrics, see how Microsoft Teams tracks activity.
Nobody independent counts headsets
The only free, regularly published hardware series anyone can inspect is Valve's opt-in survey of Steam users. It measures a narrow population — people who play games on a PC and agreed to be surveyed — but it is transparent about what it is.
1.71% of Steam users had a VR headset connected, with Quest 2 at 26.66% and Quest 3 at 26.25% of those headsets.
Direction: No detectable effect. Strength of evidence: Mixed.
Caveat PC gamers only, and it counts headsets attached to a PC, so standalone-only Quest use is invisible and the population is not representative of consumers.
Everything else is modelled. Shipment trackers are sold to the manufacturers whose shipments they estimate, and their forward numbers are projections rather than counts.
IDC forecasts about 3.2 million mixed-reality headsets shipped in 2026 rising to 10.4 million by 2030, while display-less smart glasses grow to about 13.6 million units in 2026.
Direction: Increase. Strength of evidence: Weak.
Caveat Commercially interested projection from a vendor-funded tracker; shipment figures are modelled rather than audited, and the growth is in camera glasses rather than VR headsets.
The detail worth carrying away from that forecast is that the unit growth sits in glasses with no display at all — cameras and speakers on a frame. That is a different product from the headset the metaverse was sold on. The same pattern of vendor estimates standing in for measurement runs through the counting of connected devices.
The losses are audited and the revenue is not growing into them
Company results are the one place in this subject where the numbers are audited, filed and comparable year to year.
Meta's Reality Labs division reported full-year 2025 revenue of $2.207 billion against an operating loss of $19.193 billion.
Direction: Decrease. Strength of evidence: Strong.
Caveat Segment reporting mixes VR, AR glasses and software research, so it is not a pure measure of headset economics.
Three years of that pattern give a sense of the scale involved, though the total below is arithmetic across two company releases rather than a figure Meta itself publishes.
Reality Labs operating losses across 2023 to 2025 total roughly $53.0 billion against cumulative revenue of about $6.2 billion.
Direction: Decrease. Strength of evidence: Strong.
Caveat Summed by this site from two audited company releases; it excludes losses before 2023 and is not a total Meta reports.
Roughly eight and a half dollars of loss for every dollar of revenue, sustained for three years, is a statement about demand rather than about engineering.
The clinical evidence is stronger than the consumer evidence
Stroke rehabilitation is the best-studied use. The Cochrane finding is specific and it cuts both ways.
VR gave no significant benefit over conventional therapy alone for upper-limb function after stroke, but did help when added to usual care, at a standardised mean difference of 0.49.
Direction: Cuts both ways. Strength of evidence: Strong.
Caveat The authors graded most evidence low quality, many trials were small, and the benefit may reflect extra therapy time rather than anything specific to VR.
That last point is the whole question. If VR works because it adds hours of practice a therapist could not supervise, the technology is a delivery mechanism, not a treatment. Mental health has one large randomised trial, with a declared conflict of interest attached to it.
Automated VR therapy produced statistically significant but small effects at six weeks, with agoraphobic avoidance at d = -0.18 and distress at d = -0.26.
Direction: Decrease. Strength of evidence: Strong.
Caveat Single-blind design, and the lead author founded and was a director of the company commercialising the intervention.
Effects that size are real and modest — the sort of difference that shows up in a trial of hundreds of people and would be hard to notice in one patient. Regulators have accepted the category in a narrow form: the US FDA granted De Novo authorisation to EaseVRx on 16 November 2021, creating a device class for VR behavioural therapy in chronic pain (FDA, 2021). That is an assurance of safety and effectiveness for one indication, not proof of durable benefit.
The largest published number was a projection
The figure that shaped a decade of coverage was not a measurement of anything.
McKinsey projected that the metaverse could generate up to $5 trillion in value by 2030.
Direction: Increase. Strength of evidence: Weak.
Caveat A commercially interested consultancy projection published at the peak of metaverse marketing, with no published methodology and no subsequent public reconciliation against outcomes.
A 2030 scenario is not falsified by 2025 accounts. But the direction of travel is evidence about the projection's plausibility, and the exercise of setting the two side by side is the skill this section is about. How to read a technology forecast sets out the general case.
The short version
- Meta's Reality Labs lost $19.193 billion on $2.207 billion of revenue in 2025, the clearest audited number in consumer VR (Meta Platforms, 2026).
- Valve's July 2026 opt-in survey found 1.71% of Steam users with a headset connected, and it is the only free hardware series open to inspection.
- Cochrane's 2017 review of 72 trials found VR added to usual care helped upper-limb recovery after stroke, but VR replacing conventional therapy did not.
- McKinsey's June 2022 projection of up to $5 trillion of metaverse value by 2030 was a consultancy scenario with no published method, not a finding.
- Nobody has published a longitudinal measurement of headset abandonment, so claims about how quickly people give up on VR have no evidence base at all.