Lasers in space: optical links between satellites and Earth
Updated October 8, 2026
Lasers move far more data than radio and largely sit outside spectrum licensing, but they only work if you can hit a target metres wide from hundreds of kilometres away — and, pointing at the ground, only when the sky is clear.
Why a laser carries more — and why it is so hard to aim
Both radio and laser links obey the same diffraction limit: the bigger the aperture and the shorter the wavelength, the tighter the beam. Light has a wavelength roughly ten thousand times shorter than Ka-band radio, so the same size aperture produces a dramatically narrower beam. That single fact explains the capacity, the pointing difficulty, the interception resistance and the absence of interference coordination.
| Wavelength | Aperture | Beam spread | Spot at 1,000 km | Pointing | |
|---|---|---|---|---|---|
| Ka-band radio | ~1 cm (30 GHz) | 40 cm dish | ~61 mrad (~3.5°) | ~61 km across | Milliradians — a steerable dish is enough |
| Optical laser | ~1 µm (infrared) | 10 cm telescope | ~24 µrad | ~24 m across | Microradians — needs fine steering and active tracking |
- Capacity: the laser concentrates its energy into a spot thousands of times smaller than the radio footprint, so the receiver captures a far larger fraction of the transmitted power — still a tiny fraction, but enough to support far higher data rates.
- Pointing: a 24 m spot at 1,000 km means aiming to within microradians while both ends move at roughly 7.5 km/s. That is why laser terminals combine a motorised mount with a fast fine-steering mirror and an active tracking sensor.
- Interception: to eavesdrop you would need to place a receiver inside a beam metres wide. A radio beam tens of kilometres wide can be picked up by any suitably equipped receiver within it, which is one reason radio links rely heavily on encryption.
- Coordination: beams this narrow rarely overlap, so optical links need far less frequency planning between operators than radio does.
Sources: Free-space optical links for space communication networks (arXiv)
Two kinds of laser link
Satellite ↔ satellite — Optical inter-satellite link (OISL)
- Medium
- Vacuum — no atmosphere, no weather, no absorption
- The hard part
- Both ends are moving at roughly 7.5 km/s, often in different orbital planes, so the geometry never stops changing. Links are made and broken continuously and the network topology reshapes itself constantly.
- Status
- Operating at scale today
Satellite ↔ ground — Optical feeder or downlink
- Medium
- Atmosphere — cloud, turbulence and absorption all apply
- The hard part
- One end finally sits still, but the beam has to cross the whole atmosphere. Cloud is an absolute blocker, not a fade: no amount of extra power gets an infrared beam through thick cloud. The fix is geography, not engineering — several ground stations far enough apart that their weather is uncorrelated.
- Status
- Demonstrated (for example NASA TBIRD); not yet routine
The cloud problem, and why optical ground stations come in networks
A single optical ground station is unavailable whenever cloud sits over it. A published study modelling an eight-station network across Australia and New Zealand found an average site availability of 69% — but, because the sites’ weather is only weakly correlated, the network as a whole could reach up to 99.97% reliability to a geostationary satellite, defined as the probability that at least one station is clear. Site diversity is the central design principle of the ground segment.
How much faster? Radio and laser, side by side
Kevin Coggins, program manager of NASA’s Space Communications and Navigation programme, puts it plainly: optical links can transfer 10 to 100 times more data than radio. The real systems below show the scale — plotted on a logarithmic axis, because a linear one would make the radio bar invisible.
| System | Link | Speed |
|---|---|---|
| NASA Roman telescope | Ka-band radio downlink | 500 Mbps |
| ESA/Airbus EDRS | laser relay, LEO → GEO | 1.8 Gbps |
| Starlink laser mesh | per inter-satellite link | 100 Gbps |
| NASA TBIRD | laser downlink from a 6U CubeSat | 200 Gbps |
Can you bundle lasers to go faster? Yes — two ways
More colours through one beam
Wavelength-division multiplexing puts many independent channels, each on its own infrared colour, through the same telescope — exactly how a terrestrial fibre reaches terabits. DLR and ADVA showed 13.16 Tbps this way in 2018 over a 10.45 km ground link using 53 wavelengths. In orbit it has been shown on a small scale — NASA’s TBIRD combined two wavelengths — and fibre-scale versions are planned, for example in ESA’s HydRON.
Sources: NASA TBIRD paper (SPIE 2022) · ESA HydRON
More terminals in parallel
A satellite can carry several laser terminals, each linked to a different neighbour or ground station at the same time, and their capacities add up across the network. The roughly 42 petabytes a day reported for Starlink’s mesh in 2024 moved across thousands of links at once, not one beam. Parallel terminals also allow a new link to be set up before an old one is released.
"Fibre in space": why vacuum beats glass on long routes
Light travels at full speed in vacuum but is slowed by glass. Standard single-mode fibre has a group index of about 1.468 at 1550 nm, so light in fibre moves at roughly 204,000 km/s — about 68% of its speed in vacuum. A laser path through space is therefore about 47% faster over the same distance.
The catch is the trip up and down: a satellite at 550 km adds roughly 1,100 km of climb and descent. In the simplest model, space wins once the route is longer than about 2,350 km; allowing for the longer path at orbital altitude pushes that to roughly 2,900 km, before counting switching delay in each satellite. Real fibre rarely follows the straight great-circle path, which favours space; zig-zagging between satellites works the other way.
Sources: Corning SMF-28e+ datasheet (group index 1.4682 @ 1550 nm)
A provider network in orbit: LEO for access, MEO for the backbone
Terrestrial carrier networks are layered: an access layer close to customers, an aggregation layer, and a high-capacity core. Some proposed and planned systems copy that structure in space — many low satellites close to users, a smaller layer of medium-orbit satellites above them acting as a backbone, and laser links tying the layers to each other and down to ground gateways and fibre.
Core / backbone
MEO (thousands to tens of thousands of km up). Fewer, larger satellites. Each sees a large part of the Earth, so links are longer and change less often. Carries traffic between regions, and can also connect straight down to the ground — by laser to an optical ground terminal, as TeraWave’s FCC filing describes, or by radio, as SES’s operating O3b mPOWER MEO system does today.
Access
LEO (a few hundred km up). Many satellites close to users, giving low latency. Each one passes overhead quickly, so links hand over constantly. Users connect by radio; LEO satellites can also send bulk data down by laser to optical ground stations, as NASA’s TBIRD demonstrated.
Ground
Users and gateways. User terminals reach the access layer by radio; gateways connect the constellation to terrestrial fibre, clouds and the internet.
Blue Origin TeraWave Proposed — FCC application pending
Blue Origin’s FCC application narrative, filed in January 2026, describes 5,280 LEO satellites at 520–540 km and 128 MEO satellites at 8,000–24,200 km, with optical links from the MEO layer to ground optical terminals and optical inter-satellite links in both layers. Blue Origin’s announcement adds that customers can each reach up to 144 Gbps over Q/V-band radio from the LEO layer, and attaches "up to 6 Tbps" to optical links from the MEO layer without stating whether that is per link or aggregate; neither figure appears in the application narrative. The application was accepted for filing in February 2026 — an administrative step, not an approval — and Blue Origin says deployment is to begin in Q4 2027.
Sources: Application narrative (FCC filing, copy via GeekWire) · FCC accepted-for-filing notice (Feb 2026) · Blue Origin announcement
ESA HydRON Planned — European programme
ESA’s High-thRoughput Optical Network aims to demonstrate the world’s first all-optical, multi-orbit transport network at terabit-per-second capacity, extending terrestrial fibre networks into space. Its published design splits each payload into a network part and an application part — essentially a backbone and an access layer — with wavelength-multiplexed laser terminals and optical routing on board.
Sources: ESA HydRON · HydRON design paper (SPIE 2019)
Why split the layers? A LEO-only mesh keeps latency low but every satellite moves fast, so links make and break constantly and long-distance traffic needs many hops. A MEO layer has far fewer, slower-moving nodes, so long-haul paths are steadier with fewer hops — at the cost of distance: a satellite 8,000 km up is roughly 27 ms away in one direction from the ground below it, compared with about 2 ms for one at 550 km (our arithmetic, straight up, at the speed of light).
As far as we could find, no multi-orbit optical provider network of this kind is yet operating; the examples above are proposed or planned. Single-layer LEO laser meshes, such as Starlink’s, are operating today.
Laser links in operation and in test
SpaceX Starlink Operating
The largest optical mesh publicly described in operation
In a January 2026 FCC application, SpaceX described its network as a "high capacity (petabit) and high reliability laser mesh". A SpaceX conference paper presented in January 2024 describes a fleet of 100G laser inter-satellite links with roughly 99% link uptime; its abstract refers to more than 5,000 terminals. Press coverage of that presentation reported about 42 petabytes carried per day and links of up to about 5,400 km.
The traffic and distance figures come from press reports of the January 2024 presentation, not from a SpaceX publication, and we found no newer figures. Jonathan McDowell’s independent catalogue listed about 10,550 Starlink satellites of the v1.5 and later generations in orbit on 2 October 2026 — the generations widely reported to carry laser links. We found no SpaceX statement of how many terminals each satellite carries.
Sources: FCC Public Notice DA 26-113 (Feb 2026) · SpaceX paper, SPIE 12877 (2024) · Press report of the presentation (Hackaday) · Starlink statistics (Jonathan McDowell)
NASA TBIRD Demonstrated (mission ended 2024)
200 Gbps down from a CubeSat — using ordinary telecom optics
In April 2023 TBIRD achieved a 200 Gbps optical downlink, moving up to 4.8 terabytes in a single pass from a 6U CubeSat to a 1 m telescope at JPL’s Table Mountain facility. The transceivers were commercial off-the-shelf 1.5 µm coherent fibre units of the kind used in terrestrial networks, with forward error correction and retransmission to ride out atmospheric fading.
This is strong public evidence that terrestrial coherent optics can work in orbit. NASA described it in 2023 as the fastest downlink then achieved from space. The mission ended in September 2024.
Sources: NASA: fastest space-to-ground laser comms link · NASA: TBIRD mission end (Oct 2024) · MIT Lincoln Laboratory / NASA technical paper SSC23-I-03
ESA / Airbus EDRS Operating
A commercial laser relay running since 2016
The SpaceDataHighway relays data from low-orbit satellites up to geostationary nodes at 36,000 km by laser, at up to 1.8 Gbit/s, then down to the ground by radio. Airbus reports more than 80,000 laser connections over eight years of operation, with 99.53% reliability.
These are Airbus’s own figures; the service has run since November 2016.
Sources: Airbus: SpaceDataHighway · Airbus: 50,000 laser connections (2021)
NASA DSOC (Psyche) Demonstrated (concluded Sept 2025)
Laser comms beyond the Moon
In December 2023 the Deep Space Optical Communications demonstration streamed ultra-high-definition video from the Psyche spacecraft at a maximum 267 Mbps from about 31 million km. Later passes sustained megabit rates from hundreds of millions of kilometres; by the time the demonstration concluded in September 2025 it had returned 13.6 terabits in total.
Rates fall steeply with distance: 267 Mbps at 31 million km became single-digit Mbps at 390 million km.
Sources: NASA JPL: first video from deep space via laser · NASA JPL: first phase complete · NASA JPL: DSOC mission
US Space Development Agency Delays reported (GAO, 2025); Tranche 1 launching
A cross-vendor military mesh, with delays reported by the GAO
The SDA designed its Proliferated Warfighter Space Architecture around optical terminals built to a common standard so different vendors’ satellites can link. Its identified rate is 2.5 Gbps, which the GAO describes as "relatively lower than some commercial technologies" at 100 Gbps; DOD officials told the GAO this was partly because DOD encryption requirements do not support higher rates.
In February 2025 the GAO reported that the SDA "has not yet fully demonstrated" laser communications in space, and that as of December 2024 most of its contractors had not yet achieved the planned capabilities. Tranche 1 transport satellites began launching in September 2025, so the picture has moved on since that report. Interoperability across vendors is the hard part, and it is the reason the standard exists.
Sources: GAO-25-106838 (Feb 2025) · Breaking Defense (Jul 2026)
What optical does not do
We are not aware of any single optical link to or from a spacecraft publicly demonstrated at terabit speed.
Terabit figures in circulation are either whole-network aggregates, future programme targets, or ground experiments. DLR’s well-known 13.16 Tbps record was set over a 10.45 km horizontal path between a ground station and a mountain, using 53 wavelength-multiplexed channels — not a space link, and not one beam.
Radio is not going away.
Radio passes through cloud, needs far looser pointing, and works while a spacecraft is tumbling or has lost attitude control. Commanding, telemetry and safe-mode recovery stay on radio precisely because it is forgiving. Optical is the capacity layer, not a replacement.
Optical links largely sit outside spectrum licensing.
In the United States, the FCC’s International Bureau stated in its 1999 Teledesic order that the Commission lacks jurisdiction over optical inter-satellite links, because they involve neither wire nor radio transmission, and applicants still cite that position. Optical links therefore avoid spectrum assignment and interference coordination. It also means regulatory filings tend to carry little detail on optical link performance. Other rules, such as laser-safety requirements, still apply.