Can satellites connect directly to data centres? NTT and Cailabs are building the missing optical layer

Satellite communications traditionally divide neatly into two worlds. Spacecraft transmit data toward a ground station, after which terrestrial telecommunications networks carry the information onward to processing centres, customers or storage infrastructure.

NTT, Inc. and French photonics company Cailabs are examining whether those layers can be connected much more directly. Under a memorandum of understanding signed on September 17, the companies plan to explore integrating NTT’s digital-coherent and network-control technologies with Cailabs optical ground stations and linking selected stations to nearby NTT data centres through the All-Photonics Network developed under NTT’s IOWN initiative.

The idea is still a development programme rather than a commercial network. The companies plan a phased technical approach beginning with demonstrations and potentially moving toward higher-capacity coherent optical communication and orchestration across satellites, ground stations and terrestrial network resources.

Nevertheless, the architecture points toward a larger shift in the satellite economy. As Earth-observation constellations, communications satellites and future orbital computing systems generate more data, the bottleneck increasingly becomes not merely collecting information in space but moving enormous datasets quickly into the computing infrastructure that can use them.

Why are satellite operators interested in laser communications instead of relying only on radio?

Traditional satellite communications rely heavily on radio-frequency spectrum. Radio is mature and reliable, but rapidly increasing data volumes can pressure available bandwidth and require larger antennas or greater transmitter power to support higher rates.

Optical communications use infrared laser links instead. NASA says laser systems can deliver much higher data rates while using flight terminals that may be smaller, lighter and less power intensive than comparable radio systems. Its Laser Communications Relay Demonstration describes potential bandwidth improvements of roughly 10 to 100 times over radio-frequency systems for appropriate applications.

The practical benefit is not that light travels faster than radio waves because both move at the speed of light. The advantage is that optical links can encode and transmit substantially more information within a communication session.

NASA’s TBIRD demonstration has already shown a 200-gigabit-per-second space-to-ground optical link, illustrating that the technology can support data rates more commonly associated with terrestrial high-capacity networking than traditional satellite communications.

That capacity becomes increasingly valuable as satellites carry higher-resolution cameras, synthetic-aperture radar, scientific instruments and other sensors capable of generating datasets far larger than legacy telemetry systems were designed to handle.

Why does connecting optical ground stations directly to data centres matter?

A high-speed satellite downlink solves only part of the problem if data then enters a conventional terrestrial route with additional network hops, bandwidth limitations or operational handoffs before reaching computing infrastructure.

NTT and Cailabs are exploring a more integrated architecture in which selected ground stations connect through NTT’s All-Photonics Network to nearby data centres. The objective is to create an end-to-end path spanning the satellite, optical ground terminal, terrestrial optical network and computing environment.

That could be particularly important for workloads where the value of satellite data depends on how quickly it can be processed. Earth-observation imagery, disaster monitoring, defence applications and other time-sensitive services may benefit when information moves from orbit into cloud or artificial-intelligence infrastructure with fewer intermediate bottlenecks.

The architecture could also change the commercial role of a ground station. Instead of functioning mainly as a specialised antenna site, the station becomes an edge gateway feeding high-capacity terrestrial compute networks.

This creates potential convergence between telecommunications companies, data-centre operators and space-infrastructure providers that historically occupied relatively separate parts of the technology stack.

What prevents optical ground stations from replacing radio networks completely?

Weather is the most obvious limitation. Laser beams are narrow and can be disrupted by clouds, mist and atmospheric turbulence, whereas radio links can operate through many conditions that block optical communications.

NASA addresses this problem partly through geographic diversity. Its optical systems use ground stations in locations selected for clear weather, and the agency has emphasised that multiple stations allow a spacecraft to redirect communication when cloud cover prevents one site from receiving the signal.

Pointing accuracy is another challenge. A spacecraft must direct a narrow beam toward a relatively small receiver across enormous distances, making acquisition, tracking and atmospheric compensation technically demanding.

For these reasons, optical communication is more likely to complement radio than eliminate it in the foreseeable future. Radio can provide robust command, telemetry and fallback connectivity, while optical links handle extremely high data volumes when atmospheric and geometric conditions permit.

A commercially useful network therefore requires orchestration as much as photonics. Systems need to know which ground station has visibility, which site has suitable weather, where terrestrial capacity is available and how traffic should be routed when conditions change.

Why could a global network of optical ground stations become strategically valuable?

A single optical terminal has limited availability because satellites move rapidly and weather can interrupt service. A geographically distributed network increases the probability that at least one station can establish a link when data needs to come down.

This creates infrastructure economics resembling other network businesses. Each additional station can make the wider system more useful, particularly when sites connect to high-capacity terrestrial fibre and cloud resources rather than operating as isolated terminals.

NASA has already demonstrated the importance of multiple ground stations in its laser-communications programmes, while the European Space Agency has developed optical links within its European Data Relay System to increase access to satellite data.

NTT’s advantage is potentially its terrestrial footprint. If optical ground stations can connect directly into a photonic network reaching data centres, the company may be able to combine a space-communications service with infrastructure it already operates on Earth.

Cailabs contributes expertise at the atmospheric boundary, where sophisticated optics are needed to capture and stabilise laser signals affected by turbulence.

Could optical ground stations become part of the cloud-computing ecosystem?

The commercial opportunity becomes clearer when satellite data is treated as another cloud workload rather than as a specialised aerospace product. Earth-observation companies increasingly sell analytics rather than raw imagery, meaning data must move rapidly from sensors into storage, artificial-intelligence models and customer applications.

A ground station connected directly into a data-centre network can shorten that chain. Satellite imagery could potentially be downlinked, routed and processed in an integrated infrastructure environment rather than transferred through separate systems operated by multiple providers.

Future orbital computing could make the relationship even more complex. Some processing may occur in space before data reaches Earth, while other workloads could be distributed dynamically between satellites, edge facilities and terrestrial data centres.

NTT has said the Cailabs collaboration could eventually support a broader network of optical ground stations connected to NTT data centres under its NTT C89 space-business initiative. That remains a future ambition rather than a committed deployment, but it indicates the company is thinking beyond individual terminals.

What would prove that space-to-cloud optical networking is becoming commercially real?

The first milestone is successful technical integration. NTT’s coherent optical systems need to work reliably with Cailabs ground terminals under real atmospheric conditions rather than only in controlled demonstrations.

The second is network orchestration. Commercial customers will need dependable mechanisms for switching among ground stations, allocating terrestrial capacity and maintaining service when weather blocks individual optical paths.

The third is utilisation. A network becomes economically attractive only if enough satellite operators generate traffic to justify investment in ground terminals, high-capacity fibre and associated data-centre infrastructure.

Pricing will matter as much as technical performance. Optical networking can provide higher bandwidth, but operators will compare the cost per delivered unit of data against improving radio systems and alternative ground-station networks.

What does NTT’s stock performance suggest about investor sentiment?

NTT shares closed September 18 at ¥177.3, down 1.5% for the session after touching ¥180.1, which was also the stock’s year-to-date high. The shares had closed at ¥172.2 on September 9, meaning NTT remained roughly 3% higher over that period despite the September 18 pullback.

It would be inappropriate to attribute that move specifically to the Cailabs memorandum, given NTT’s enormous telecommunications business and the early-stage nature of the collaboration. The market value of the optical-ground-station programme is currently tiny relative to the group.

The strategic significance lies instead in how NTT is attempting to extend its terrestrial photonics technology into another infrastructure layer. If satellite networks generate substantially more data over the next decade, the companies controlling the gateways between orbit and high-performance computing may become increasingly important.

The satellite economy is often imagined as rockets and constellations, but information ultimately has to reach somewhere useful on Earth. Connecting a laser beam from orbit directly into the optical fabric of a data centre could make that boundary considerably less visible.