Microsoft HoloLens
Microsoft HoloLens is a self-contained holographic headset that projects see-through holograms into the wearer's physical environment. The first commercially available untethered mixed reality headset, it shipped to developers in March 2016 and established the design language and interaction model that defined enterprise AR for the decade that followed.
HoloLens 1 (2016)
HoloLens 1 ran on a custom Intel Atom processor and included a purpose-built Holographic Processing Unit (HPU) — a dedicated coprocessor designed to process spatial data from the device's depth, inertial, and camera sensors in real time without burdening the main CPU.4 The HPU fused accelerometer, gyroscope, magnetometer, and depth sensor data into a continuous 6DoF head pose, enabling the holographic display to remain anchored to the world as the user moved.
The display used waveguide optics to project light into the user's field of view, achieving a field of view of approximately 30 degrees diagonal — a limitation that made holograms feel like floating screens in the central vision rather than pervasive additions to the environment. The device was entirely wireless and self-powered, running Windows 10 Holographic with a CPU/GPU co-designed for the 2.8 Wh battery.
Gaze, gesture, and voice (GGV) was HoloLens 1's interaction model: head gaze for targeting, a small vocabulary of hand gestures (air tap, bloom) for selection, and Windows Speech APIs for voice commands. The gesture set was minimal — the camera-based hand tracking of the era could not support the full 26-DOF hand skeleton that would come with HoloLens 2.1
HoloLens 2 (2019)
HoloLens 2 represented a ground-up redesign of the optical, tracking, and interaction systems:2
Display: A new waveguide design increased the field of view to approximately 52 degrees diagonal — nearly triple the solid angle of HoloLens 1 — while improving brightness and colour uniformity. The holographic density rose to 47 pixels per radian.
Hand tracking: Combined time-of-flight depth sensing and neural network processing enabled full 26-DOF skeletal hand tracking — both hands simultaneously, with fingers individually tracked. This eliminated the need for a controller entirely and enabled direct manipulation: users could reach out and physically interact with holographic objects using natural hand poses.3
Eye tracking: Four NIR cameras inside the visor tracked both eyes simultaneously, enabling gaze-directed targeting, foveated rendering, and eye-gaze interaction.
Fit and comfort: HoloLens 2 moved from a forehead-heavy balance to a rear counterweight design, improving wearability for extended industrial use.
HPU 2.0: The second-generation HPU added hand tracking inference and eye tracking data fusion to its workload, offloading both from the Snapdragon 850 application processor.
Interaction Model
HoloLens 2 introduced what became the standard spatial interaction vocabulary: the hand ray (a virtual ray projecting from the palm for far-field pointing), direct touch (fingertip contact with near-field surfaces), and grab (closing the hand to pick up objects). These three primitives were codified in MRTK and influenced every subsequent XR interaction framework. The "select" voice command remained — a word that activated whatever the user was gazing at — maintaining voice as a first-class input modality.
Enterprise Focus
Microsoft positioned both HoloLens generations exclusively as enterprise and developer devices, with no consumer offering. Target verticals included manufacturing (guided assembly, inspection), healthcare (surgical planning, medical training), defence and field operations, and architecture and construction. Boeing, Lockheed Martin, Airbus, Volkswagen, and the UK's National Health Service were among the early large-scale enterprise deployments.5
See also: Hardware · Hand Tracking · Eye & Gaze Tracking · SLAM · MRTK · Microsoft Kinect · Spatial UI Design