A VCSEL (pronounced “vixel”) is a type of semiconductor laser that emits light straight up from its surface rather than out the edge. The name stands for vertical-cavity surface-emitting laser. This design makes VCSELs cheap to manufacture, easy to test on the wafer before packaging, and simple to arrange into dense arrays of hundreds or thousands of emitters. You encounter VCSEL technology every time you unlock a modern smartphone with your face, and it quietly powers much of the short-range fiber optic networking inside data centers.
How a VCSEL Works
A conventional semiconductor laser sends light out the side of a chip. A VCSEL flips the geometry: it bounces light between two mirrors stacked above and below a thin active layer, then releases the beam vertically through the top surface. Those mirrors are called distributed Bragg reflectors (DBRs), and they’re built from alternating thin films of materials with different optical properties. A typical bottom mirror might consist of 12 pairs of layers, each only about 40 to 70 nanometers thick. Together, these pairs reflect more than 99% of the light back into the cavity, which is necessary because the active region the light passes through is extremely thin.
The active region itself is where the laser light is actually generated. It contains a stack of quantum wells, semiconductor layers just a few nanometers thick where electrons and holes recombine to produce photons. Because the light bounces vertically through this short cavity rather than traveling a long horizontal path, VCSELs need those ultra-reflective mirrors to build up enough optical gain to lase. The entire resonator can be as short as a few micrometers from top to bottom. That compact structure is a big part of what makes VCSELs inexpensive: manufacturers can fabricate thousands of them on a single wafer and test each one before cutting the wafer apart, something edge-emitting lasers don’t allow.
Wavelengths and Materials
The semiconductor material determines what wavelength of light a VCSEL emits. Gallium arsenide (GaAs) is the most mature platform and covers the 850 to 940 nanometer range in the near-infrared. This is the wavelength window used in most consumer and data communication VCSELs. For longer wavelengths, indium phosphide (InP) based designs cover 1,300 to 2,000 nanometers, which is useful for telecom fiber and gas sensing applications. Gallium nitride (GaN) VCSELs target shorter, visible wavelengths, though they remain harder to manufacture because growing high-quality mirrors on nitride materials is still a significant fabrication challenge.
Facial Recognition and 3D Sensing
The application that brought VCSELs into the mainstream is smartphone facial recognition. Systems like Apple’s Face ID project a pattern of 10,000 to 30,000 infrared dots onto your face within a roughly 60° by 45° field of view, then read the distortion of that pattern with a camera to build a 3D depth map. The infrared flood illuminator and dot projector in these modules are VCSEL arrays.
These arrays operate at 940 nanometers. Early designs used 830 to 850 nanometers because camera sensors are more sensitive there, but moisture in the atmosphere absorbs sunlight around 940 nanometers, which reduces background noise and improves detection accuracy in outdoor light. The shift to 940 nanometers also removed aluminum from the laser’s active layer, boosting long-term reliability.
A typical array packs several hundred individual VCSEL emitters into a chip footprint of 0.5 to 1 square millimeter, with emitters spaced about 20 to 30 micrometers apart. Each emitter has an optical aperture around 10 micrometers in diameter. While a single VCSEL for data communications produces less than 10 milliwatts, a facial recognition array delivers several watts of peak power in short pulses.
Data Center Interconnects
Inside data centers, VCSELs are the dominant light source for short-reach fiber optic links up to about 300 meters. They’re paired with multimode fiber and offer a combination of high speed, low power consumption, and low cost that edge-emitting lasers can’t match at these distances. Researchers have demonstrated directly modulated VCSELs at 850 and 910 nanometers reaching gross data rates of 224 gigabits per second over multimode fiber, which points toward the next generation of data center interconnects. Rates around 180 gigabits per second are achievable with standard error correction, and pushing to 224 Gbps requires more advanced forward error correction techniques.
The energy efficiency matters enormously at data center scale. A single hyperscale facility can contain millions of optical links, so even small per-link power savings translate into significant reductions in electricity consumption and cooling costs.
Automotive LiDAR
LiDAR systems for autonomous vehicles need high-power infrared laser sources that can illuminate objects at long range. VCSELs are increasingly competitive here because of a technique called multijunction stacking, where multiple light-generating layers are built on top of each other within a single device. Leading manufacturers are now mass-producing VCSELs with five to seven junctions, and larger numbers are in development.
Stacking junctions multiplies the optical output without increasing the chip’s footprint. A six-junction design with an antireflective coating can triple its brightness and cut its beam spread in half compared to a conventional single-junction VCSEL. A single 7 micrometer emitter in this architecture produces 28.4 milliwatts in a clean single-mode beam. Researchers project that an optimized 18-junction array on a 100 micrometer square chip could reach power densities of 15,000 watts per square millimeter.
The multijunction approach also saves energy in an indirect way. For a given optical output, stacking junctions reduces the current flowing through the driver circuit and raises the voltage instead. Lower current means less waste heat from electrical resistance in the wiring and driver electronics, improving overall system efficiency. This tradeoff works well for vehicles, which have plenty of voltage headroom from their batteries, though it’s more constrained in battery-limited consumer devices.
Cost is another advantage. Compared to photonic crystal surface-emitting lasers (PCSELs), which require expensive lithography steps, VCSELs use 10 to 100 times less semiconductor area to generate the same optical power, even before accounting for the more complex fabrication PCSELs demand.
Why VCSELs Over Other Lasers
Edge-emitting lasers produce light from the side of a chip, which makes them harder to test before packaging and difficult to arrange into two-dimensional arrays. VCSELs emit from the top surface, so hundreds or thousands can be fabricated, tested, and operated as a single array on one chip. The circular beam profile of a VCSEL also couples more efficiently into optical fibers without expensive beam-shaping optics.
The multijunction advantage is structurally unique to VCSELs. In an edge-emitting laser, adding junctions doesn’t increase power density because the junctions must be spaced micrometers apart to avoid absorbing each other’s light. They end up behaving like separate emitters stacked vertically. In a VCSEL, the standing wave pattern inside the vertical cavity creates natural positions for quantum wells and tunnel junctions to sit at intensity peaks and nulls, respectively, allowing them to be packed tightly and contribute to a single, brighter output beam.
Market Growth
The global VCSEL market generated an estimated $1.3 billion in revenue in 2024 and is projected to reach $2.1 billion by 2030, growing at a compound annual rate of 8.8%. That growth is driven by expanding use in 3D sensing across smartphones and industrial automation, rising demand for high-speed data center optics, and the emerging automotive LiDAR segment. As multijunction fabrication matures and new wavelength platforms become reliable, VCSELs are likely to displace other laser types in applications that previously required more expensive or power-hungry sources.