Stratus exchange routes alternative capital into data-centre risk

The property market's next heavyweight risk is a building full of servers, and Marsh is betting that capital-markets money can price it. The broker has launched Stratus, a $10 billion property insurance exchange that routes alternative and traditional capacity into the operational risks of digital infrastructure, according to Artemis. Stratus offers capacity on a single-placement basis for the global exposures of US-domiciled companies, with 30 capital providers lined up to evaluate each risk individually.

Data centres are the demand story. Allianz Commercial has projected the global data centre insurance market will grow from roughly $11 billion today to more than $24 billion by 2030, as Artemis reported. The scale is so large that numerous senior industry leaders have been commenting that conventional property capacity will not be enough to support all the facilities coming online, which is the gap Stratus is designed to fill. Marsh's first pitch, carried by Artemis, is that this more structured trading environment enables providers to better assess, quantify, and diversify risk while addressing the aggregation concerns that arise during the operational phase of these projects.

Stratus is a placement venue, not a securitised vehicle. Capacity is written through the global property risk transfer market, with each provider deciding how much of a particular risk to take, a structure that keeps the capital flexible in a way a catastrophe bond is not. Marsh contributes its data and analytics and advisory services to match each placement to the providers' appetites.

The risk itself is operational rather than meteorological. During the operational phase of a digital-infrastructure project, the exposures include multi-billion-dollar campuses, self-generating power systems, and the emerging technologies built into processing units and critical equipment. As Marsh put it, 'This more structured trading environment enables providers to better assess, quantify, and diversify risk and address aggregation concerns.' Aggregation is the telling word. In a catastrophe bond portfolio, aggregation is a model output; on a data-centre campus, it is likely physical, with power grids, cooling supply chains, and the same generation of processors connecting one placement to the next.

Stratus's answer is to let each provider see a risk before committing and to use Marsh's analytics to size the accumulation across a portfolio. That may be enough to move the market. The carriers are willing; the correlation, until now, has been the obstacle. An exchange that surfaces the aggregation in real time is the obvious fix, and it explains why Marsh is preparing 30 providers rather than negotiating with a handful of incumbents.

The growth projection is what makes Stratus more than an experiment. Allianz Commercial's $11 billion-to-$24 billion trajectory, as carried by Artemis, implies a market that roughly doubles in four years. That kind of curve is rare in commercial property insurance, and it is precisely the sort of growth that pulls third-party capital in. It also hands the ILS market a new option, just as the search for product expansion beyond natural catastrophe was finding new targets. As this publication has argued, ILS managers have been moving into cyber, transactional risk, and sovereign parametric; data-centre property is a natural next stop, because it remains property risk but with an exposure profile that behaves nothing like wind or quake.

Two days before Stratus was unveiled, this publication argued that cat bonds and sidecars are prolonging the soft market, as record issuance and casualty sidecar demand keep feeding the property market's price decline. Stratus introduces a new pool of capital, pointed at a risk class that has not yet been softened by years of rate-cutting competition. That is exactly the right time to bring alternative capital in, before the market learns the loss history and starts pricing the class the way it priced cat bonds after a benign season.

The harder question is whether the capital that crosses Stratus will hold its discipline. The soft market's lesson, from this publication's own coverage, is that new money tends to follow returns into whatever class is cheapest to write. Data-centre property is not cheap to write; it demands engineering, power-market, and supply-chain modelling. But it is growing, and a manager under pressure to deploy cat-bond capital may find a data-centre placement easier to justify than another year of waiting for the next windstorm. That temptation is worth watching.

The deeper risk is concentration on the same AI theme that general-account investors have already been absorbing on the investment-grade side. As this publication has argued, the AI power build-out has become an investment-grade concentration problem for insurers, and a data-centre insurance book is the underwriting equivalent of that same trade. A portfolio of Stratus placements is, in effect, a portfolio of AI-power exposure. The insurance market learned to diversify hurricanes by region and earthquake risk by fault line; the equivalent diversification for data centres runs across grid regions, processor vendors, and cooling technologies, and a single grid failure can touch a dozen placements at once.

The Artemis account does not identify which of the 30 participants are ILS funds, nor does it disclose whether a catastrophe-bond or collateralised-reinsurance vehicle is already committed to the exchange. That detail will matter, because the ILS market's pricing discipline has been one of the few constraints on a softening property market. If Stratus draws in the same capital that anchors cat-bond towers, the exchange may get the most demanding underwriters in the room. If it draws in only traditional property carriers, the 'alternative capital' framing will prove to be more marketing than market structure.

Stratus is a test, and the test has two stages. The first stage is placement: can the exchange actually close a data-centre placement with its 30 providers selecting risk by risk? The second stage is loss: when a covered campus suffers a power outage, a cooling failure, or a processor defect, the claim will show whether the capital that priced the risk priced it correctly. Until that loss arrives, Stratus is a structural idea worth watching — a $10 billion promise that the capital markets can underwrite a building full of servers as well as they underwrite a windfield.