Apple Vision Pro began shipping on February 2nd and reviews are in. It's technically impressive but also expensive, heavy, and currently lacks compelling software. Enterprise applications seem more promising than consumer ones.
The Vision Pro hardware is the best spatial computing device shipped so far. Its displays are impressive, hand and eye tracking work well, and the EyeSight feature shows your eyes to people around you. The battery lasts about two hours and is separate from the device. It weighs between 600-650 grams, making it impractical to wear for eight hours.
Consumer applications like spatial video, movies, and gaming are challenging to sell at $3,499. Enterprise uses, however, are more interesting. Spatial computing is useful for overlaying information on a physical environment, such as maintenance technicians reading repair instructions while working on a machine.
In automotive manufacturing, we've seen pilot projects where technicians use AR to inspect engine components. A specific example is BMW's use of a Microsoft HoloLens 2 variant for quality control, where 97% of defects were caught in real time. Vision Pro could replicate this with better resolution but faces trade-offs: its 2-hour battery life limits continuous use in 8-hour shifts. Pairing it with a tethered industrial power bank adds $1,200 per unit, which many manufacturers won't absorb.
Examples of enterprise applications include maintenance technicians reading repair instructions while looking at the machine they are repairing, architects reviewing spatial models in the actual space they are designing, and medical professionals reviewing 3D imaging without a flat-screen intermediary.
At launch, Vision Pro ran iPad apps in a floating window, which isn't spatial computing. Genuinely spatial apps built for visionOS were sparse. The developer ecosystem is still early, and building for a device most developers don't own is difficult.
Simulators don't fully replicate the experience of looking at a 3D object in a real space. The software maturity curve will take 18-24 months. Until then, the device's capabilities will be limited.
A concrete challenge we've seen in production is depth sensing accuracy in low-light environments. Vision Pro's LiDAR and cameras struggle below 50 lux, which is common in industrial facilities. This forces developers to add redundant lighting systems, adding $300-$500 per deployment in retrofit costs. For example, a Siemens plant in Germany had to install 120 additional LED panels to meet Vision Pro's operational requirements.
The key thing to watch is not consumer adoption but enterprise line-of-business application development. If healthcare, manufacturing, or field services organisations invest in visionOS development for specific operational workflows, Vision Pro has a path to meaningful utilisation.
If Vision Pro stays a general-purpose computing device competing with laptops and iPads on price and portability, it will struggle. Its success depends on finding a niche in enterprise applications where its unique capabilities add significant value.
The current price point and limited software availability are significant hurdles. However, if developers create applications that take advantage of the device's spatial computing capabilities, it could find a place in specific industries.
For now, the device's high price and limited software make it a niche product. Its future success will depend on how well it can adapt to the needs of specific industries and organisations.