Jack Kenneally, class underwriter, HIVE Underwriters exposures the risks that will come with a drive for the siting of data centres in space.

Since Google announced Project Suncatcher in November, the idea of building AI data centres in space has moved rapidly from science fiction to serious industry discussion. As demand for artificial intelligence accelerates, so too does the need for vast computational capacity, driving an unprecedented expansion of data centre infrastructure.

These facilities are already major consumers of energy, and global electricity demand from data centres is expected to double by the end of the decade. Power availability is fast becoming a bottleneck for the AI boom and, if addressed through increased reliance on fossil fuels, risks undermining global efforts to reduce carbon emissions.

A continuous supply of energy?

It is this tension that has renewed interest in space-based solutions. Solar arrays deployed in orbit can benefit from near-continuous exposure to the sun, unimpeded by weather or day-night cycles. In theory, this makes space an attractive environment for powering energy-hungry AI workloads.

Google is not alone in exploring this concept. SpaceX, Blue Origin and others have all signalled interest, while Starcloud, a start-up focused exclusively on space-based data centres, launched a demonstrator satellite in November equipped with an Nvidia GPU. That satellite has reportedly been running a Google large language model in orbit since launch. While large-scale deployment remains a distant prospect and significant technical and economic hurdles remain, momentum behind the concept is building rapidly. The more immediate question for the insurance market is how these risks would be assessed and underwritten.

The radiation conundrum

One of the most fundamental challenges lies in the space radiation environment. Satellites are exposed to levels of radiation far beyond anything encountered on Earth, with energetic particles capable of disrupting or damaging sensitive electronics. For decades, this has driven the use of radiation-hardened components that lag well behind the cutting edge of terrestrial computing, having undergone years of qualification and testing.

Space-based data centres, however, will be driven by the commercial imperative to use the most advanced processors available. This creates a familiar tension for insurers involved in “new space” risks, but at a vastly increased scale. Underwriters may be asked to accept technology that has not been tested to traditional space standards, while simultaneously carrying sums insured far higher than those typically associated with experimental missions.

Thermal management presents an equally significant challenge. Although space is often described as cold, with no atmosphere to carry heat away, satellites must rely entirely on radiative cooling, which is inherently inefficient.

Satellite design is evolving

The heat generated by high-performance AI chips will therefore need to be transported to large external radiators, adding mass and complexity to spacecraft design. Crucially, this will require mechanically pumped fluid cooling loops operating continuously for years without the possibility of maintenance or repair.

Historically, satellite designers have avoided moving parts wherever possible due to the risk of wearout. In this context, they become unavoidable. From an insurance perspective, these permanently operating mechanical systems represent a clear and unfamiliar source of attritional and catastrophic risk.

Business Interruption from space?

Beyond physical damage, the operational nature of data centres introduces new considerations around service availability. Terrestrial data centre operators are accustomed to operating under strict service level agreements, with financial penalties triggered by downtime or performance degradation. It is reasonable to expect similar expectations to apply in space, particularly if orbiting facilities are integrated into critical AI supply chains. This raises the prospect of demand for business interruption cover alongside traditional space insurance. The space insurance market has historically focused on physical asset loss, and pricing business interruption exposure in an environment with little historical data will require a significant shift in underwriting approach and risk modelling.

Scale further complicates the picture. To replace or meaningfully supplement terrestrial data centres, orbiting facilities would need to be enormous. Some concepts envisage individual spacecraft several kilometres in size, representing an unprecedented aggregation of value in a single asset. Even today, communications satellites worth three or four hundred million dollars can be challenging to insure fully. Space-based data centres could far exceed those values. Alternative architectures, such as constellations of tens or hundreds of satellites flying in close formation, may distribute assets but do little to reduce aggregation risk in practice. Satellites separated by only hundreds of metres remain highly correlated exposures, particularly if they are launched together on a single dedicated rocket to achieve the launch costs required for commercial viability.

From concept to launch

By the time these concepts reach operational maturity, the insurance market will undoubtedly look different from today. However, the limits required to support such ventures may still stretch beyond what traditional space insurance capacity can provide, forcing the market to consider alternative capital structures and risk transfer mechanisms.

What is clear is that space-based data centres are unlikely to be an isolated anomaly. Rather, they represent the leading edge of a broader and rapidly evolving shift in how space is used to support critical terrestrial infrastructure. Keeping pace will require close attention, technical fluency and a willingness to rethink long-standing assumptions – this is a new wave of emerging risks to cross the underwriting event horizon, and there will be many more to come.