Simon Glairy is a veteran of the risk management sector, having spent years navigating the complexities of insurance-linked securities (ILS) and the evolving landscape of high-value infrastructure. As the digital economy pivots toward artificial intelligence, the physical backbone—data centers—presents a unique set of challenges and opportunities for the capital markets. In this conversation, we explore the emergence of specialized investment vehicles, the staggering concentration of risk in modern tech hubs, and how institutional capital is filling the capacity gap left by traditional insurers. The discussion covers the $1 billion sidecar structure for data centers, the environmental threats like tornadoes and water stress facing these assets, and the historical parallels to post-disaster insurance booms that informed today’s market strategies.
The strategy involving a $1 billion sidecar to achieve returns exceeding 15% through quota shares represents a significant shift into specialized risk. How does this model effectively distribute risk between the primary insurer and your investors, and what specific underwriting rigor is required to maintain these high-yield targets without compromising long-term stability?
The sidecar structure is a sophisticated way to create a partnership where investors take a proportional quota share of an insurer’s specific risks. By targeting capital in the $1 billion range, we can provide the necessary capacity that traditional markets might be hesitant to hold entirely on their own balance sheets. We are looking at returns north of 15%, which reflects the specialized nature of these policies and the “historic opportunity” currently present in the insurance-linked market. To keep these targets sustainable, our underwriting must go beyond traditional actuarial tables to understand the specific mechanical and technological vulnerabilities of AI-driven infrastructure. We focus on teaming up with insurers who have deep technical expertise, ensuring that every dollar of risk we take on is backed by a granular understanding of the asset’s operational environment.
With individual data center valuations now hitting the $30 billion mark, they are dwarfing traditional assets like major bridges. When several of these high-value facilities cluster within a tight 20-mile radius in regions like Texas or Virginia, how do you mathematically account for that concentration risk and prepare for the possibility of simultaneous massive claims?
The scale of these assets is truly unprecedented, with a single location now reaching values of up to $30 billion, which is three times the $10 billion valuation we see for some of the world’s largest bridges. This concentration risk is most acute in places like Northern Virginia or parts of Texas, where multiple facilities are packed into a 20-mile radius, creating a single point of failure for a massive amount of capital. When we model these clusters, we have to account for the “insurance gap” that emerges when the total insurable value exceeds what the traditional reinsurance market can comfortably absorb. We use advanced spatial modeling to understand how a single event, like a localized storm or a power grid failure, could trigger a series of simultaneous claims across that entire radius. Managing this requires us to leverage third-party capital to provide a buffer that prevents a single catastrophic event from destabilizing the broader portfolio.
Data reveals that over 40% of U.S. data center capacity is located in significant tornado zones, compounded by threats from extreme heat and water scarcity. In your view, how do these environmental pressures impact the physical integrity of these assets, and what specific metrics do you look for to determine if a facility is resilient enough for institutional coverage?
The overlap between data center growth and natural hazard maps is a serious concern, particularly with more than 40% of U.S. capacity sitting directly in the path of potential tornadoes. Beyond wind damage, we are seeing rising temperatures threaten the resilience of more than half of the 100 largest data-center hubs, which drives up cooling costs and puts immense strain on energy requirements. Many of these hubs are also located in high water-stressed areas, making the cooling process—which is vital for AI hardware—both expensive and environmentally risky. When we evaluate a facility, we look at the structural integrity of the buildings and their ability to withstand acute climate events that could disrupt operations. We specifically measure the efficiency of their cooling systems under extreme heat and their backup water supplies to ensure they can remain functional even when local resources are under pressure.
There are fascinating parallels between the current AI boom and the period following the Deepwater Horizon disaster, where bespoke vehicles provided critical capacity for oil rig insurance. Based on your experience with those structures, how does the current demand for data center capacity compare, and how do you decide when the market has stabilized enough to return capital to investors?
The current surge in data center construction feels very similar to the spike in demand we saw after the 2010 Deepwater Horizon event, where traditional providers couldn’t meet the sudden need for high-value oil rig coverage. Back then, we managed a vehicle for about 200 high-value rigs that delivered returns between 18% and 24% over a two-year period by stepping in when the market was most constrained. Data centers represent a similar “high-conviction” niche where the specialized risk allows for double-digit returns because the technical requirements are so high. We monitor the market closely to see when traditional capacity begins to catch up and the “insurance gap” starts to close. Once that extra capacity is no longer commanding a premium and the niche becomes commoditized, we simply return the capital to our investors and exit the position, just as we did with the oil rig vehicle.
The ILS market has traditionally been defined by residential catastrophe bonds, but there is a clear pivot toward specialized industrial risks. How will the influx of third-party capital reshape the broader insurance landscape, and what steps must reinsurers take to successfully offload these multi-billion dollar portfolios to global investors?
The shift away from a pure focus on residential property is a natural evolution, especially as the cat bond market grew by 24% last year and investors are looking for more diverse ways to deploy capital. By bringing third-party investors into the data center space, we are effectively creating a new asset class that allows reinsurers to write more business without over-leveraging their own books. For this to work, reinsurers need to be incredibly transparent with their data, allowing the capital markets to see the exact risk profile of these multi-billion dollar portfolios. We are seeing more clients talk about leveraging this capital to write larger data center policies, which indicates a level of market maturity we haven’t seen in several years. The ultimate goal is to create a seamless pipeline where industrial risk can be packaged into securities that provide the liquidity and scale required by the world’s largest institutional investors.
What is your forecast for data center insurance-linked securities?
I expect to see the very first catastrophe bond specifically dedicated to data center risk issued within the next 12 months, marking a major milestone for the industry. As the demand for AI infrastructure continues to skyrocket, the reliance on third-party capital will only grow, transforming how these facilities are financed and protected. We are entering an era where the capital markets will become the primary backstop for the digital world’s physical foundations, moving beyond natural disasters to encompass the complex technological risks of the future. This expansion of capacity will be essential for the continued build-out of global tech hubs, and I anticipate that data center ILS will become a permanent and vital fixture of the alternative investment landscape.
