Optical inter-satellite links
Optical inter-satellite links carry gigabit-class data between spacecraft using narrow infrared beams, reducing dependence on ground contact windows and cutting latency across multi-satellite architectures.
What a laser link actually does that a radio link cannot
A radio crosslink between two satellites spreads energy across a wide beam. Useful, but every watt that misses the receiver is wasted, and spectrum must be coordinated with the ITU. An optical inter-satellite link (OISL) concentrates the same energy into a beam measured in microradians. The result is orders-of-magnitude higher spectral efficiency, no ITU frequency coordination requirement for the optical band, and data rates that current operational systems demonstrate at 1.8 Gbps per link (Tesat LCT-135 terminals on the European Data Relay System, EDRS). SpaceX Starlink's second-generation shells carry laser crosslinks that the company states operate at roughly 100 Gbps aggregate across the mesh, though per-link figures are not publicly decomposed.
The physics matter for mission architects. Because the beam divergence is so small, pointing error of even a few microradians drops the link margin to zero. That constraint drives the entire mechanical and control design of the terminal, and it is the reason OISLs carry a pointing, acquisition and tracking (PAT) subsystem that can consume more mass and power than the optical bench itself.
Pointing, acquisition and tracking: the hard engineering in the middle
Acquisition is the unsolved-looking problem that is actually solved, just expensively. Two spacecraft in relative motion must find each other's beams before they can lock. Typical acquisition sequences use a coarse wide-angle beacon, often at a different wavelength, to bring the terminals into rough alignment, then hand off to a fine-pointing mirror or fast-steering mirror (FSM) loop running at kilohertz rates. EDRS uses a two-stage approach: coarse pointing via the terminal's gimbal, fine correction via an FSM with a position-sensitive detector in the feedback loop. Acquisition times on operational GEO-to-LEO links run to tens of seconds; LEO-to-LEO links between fast-moving spacecraft can be harder, with geometry changing quickly enough that the acquisition window is short.
Vibration isolation matters enormously. Reaction wheels, cryo-coolers and solar-array drive mechanisms all inject micro-vibrations into the bus at frequencies that overlap the PAT bandwidth. Terminals intended for small satellites, where structural stiffness is lower, must either include vibration isolation mounts or accept a pointing error budget that eats into link margin. This is not a detail; it is frequently the difference between a terminal that works on paper and one that closes a link in orbit.
Standards, interoperability and the SDA OCT framework
For most of its history, OISL was a bespoke bilateral agreement: two terminals from the same vendor, designed to talk to each other and nothing else. The US Space Development Agency changed that calculus with its Optical Communications Terminal (OCT) standard, which specifies wavelength (1550 nm), modulation (DPSK or OOK variants), and interface requirements so that terminals from different suppliers can interoperate across the Transport Layer constellation. The SDA awarded OCT contracts to CACI (formerly SA Photonics), Mynaric, and others, with the explicit intent of creating a competitive, multi-source supply chain.
For a sovereign programme buyer, the SDA OCT framework matters even if your constellation has no relationship with the US government. It has driven commercial vendors to build terminals against a published specification, which means procurement is less dependent on a single supplier's proprietary roadmap. Mynaric's CONDOR Mk3, for instance, is designed to the OCT specification and is available on the commercial market. That said, interoperability between non-US government constellations and SDA Transport Layer nodes is not guaranteed by the standard alone; link-layer protocol agreements still need to be negotiated.
When the mass and power budget is worth it, and when it is not
A mature OISL terminal in the small-satellite class, such as Mynaric's CONDOR Mk3, is quoted at roughly 3 kg and under 60 W average power. For a 100 kg LEO satellite, that is a significant fraction of both budgets. The question is whether the mission architecture actually needs what a laser link provides.
OISLs earn their mass when three conditions align: the constellation has multiple nodes that need to exchange data faster than ground contact windows allow; the coverage requirement means some satellites spend long periods out of sight of any ground station; and the data volumes are high enough that RF crosslinks would require large antennas or excessive spectrum coordination. Earth-observation constellations collecting synthetic aperture radar or hyperspectral data at high cadence are natural candidates. A three-satellite constellation in a sun-synchronous orbit with a well-placed ground network is almost certainly not, because the contact windows are frequent enough and the data volumes manageable enough that a radio downlink is simpler and cheaper. Honest programme architecture starts with that question before specifying the payload.
Limits the procurement documents rarely mention
Atmospheric turbulence is irrelevant for pure inter-satellite links (both terminals are in vacuum), but it becomes the dominant impairment the moment a link must close through the atmosphere to a ground optical terminal. For satellite-to-ground optical links, scintillation and cloud cover can block the link entirely. This is why EDRS uses Ka-band for its ground segment rather than optical: the reliability requirement for a commercial relay service is incompatible with cloud outages.
Thermal management is a persistent nuisance. The optical bench must maintain dimensional stability across eclipse-to-sunlight thermal cycles that can span 100°C or more. Coefficient of thermal expansion mismatches between the bench, the detector mount and the aperture housing introduce slow pointing drift that the PAT loop must correct continuously. On early operational systems this proved harder than pre-launch analysis predicted.
Lead times for qualified optical terminals are long. As of the mid-2020s, the commercial supply chain for space-qualified OISLs remains thin. Mynaric and Tesat are the most visible suppliers with flight heritage; a new entrant ordering terminals for a first constellation should expect 18 to 30 months from contract to delivery, and should not assume that a terminal tested in the lab will meet its link budget on orbit without a margin reserve of at least 3 dB beyond what the vendor's datasheet implies.
Fitting OISLs into a sovereign programme
A government building its first national constellation rarely needs OISLs in the first generation. The complexity, cost and supply-chain risk are real, and the benefit only materialises once the constellation is large enough and the ground network sparse enough to justify it. The more useful near-term decision is to reserve the mechanical and power interface on the bus so that a future generation can add a terminal without a redesign.
Where OISLs do belong in a first programme is in the requirements conversation. If the mission calls for persistent wide-area surveillance with low latency to a national command centre, and the geography means ground stations cannot provide adequate contact time, that is the moment to scope an OISL-capable architecture from the start rather than retrofit it later. Satellize's programme architecture work includes that conversation at the mission-concept stage, before the bus specification is locked.
Engineering parameters
| Typical terminal mass (small-sat class) | 2 to 5 kg (Mynaric CONDOR Mk3 quoted at ~3 kg) |
| Average power consumption | 40 to 80 W depending on link distance and modulation |
| Demonstrated per-link data rate | 1.8 Gbps (Tesat LCT-135, EDRS operational); up to 10 Gbps on newer terminals in development |
| Operating wavelength | 1550 nm (telecom C-band; SDA OCT standard wavelength) |
| Beam divergence | Typically 2 to 10 µrad half-angle; drives pointing accuracy requirement |
| Pointing accuracy required | Sub-microradian fine pointing after acquisition; coarse gimbal to ±30° or more |
| Acquisition time (LEO-LEO) | 10 to 60 seconds depending on ephemeris uncertainty and beacon design |
| Link range (operational) | Up to ~45,000 km (GEO-to-LEO, EDRS); LEO-to-LEO typically 1,000 to 5,000 km |
| Supply-chain lead time (qualified terminal) | 18 to 30 months from contract to delivery, current market |
| ITU coordination requirement | None for the optical beam; RF beacon or acquisition aid may require coordination |
One contract, one accountable engineer
Commissioned as one programme, not a stack of contracts: spacecraft, launch, ground segment, mission control, training and handover are priced together. Source-access terms and audit rights are agreed in writing before signature. Request a link-budget review.