As the buildout of AI and digital infrastructure continues to expand across the re/insurance industry, driving insurance-linked securities (ILS) into this emerging sector relies heavily on key structural advancements, including the development of standardised parametric triggers and portfolio aggregation mechanisms, according to Michael Moloney of Oliver Wyman.
“The buildout of AI and digital infrastructure represents one of the largest coordinated capital formation exercises in modern economic history. JP Morgan estimates US$5 trillion in global capital will be required between 2025 and 2030 to fund data center and AI infrastructure growth, with broader market forecasts extending to US$7 trillion across the same horizon,” Moloney said.
He continued: “The five largest hyperscalers alone are projected to deploy capital in excess of US$800 billion in 2027. Individual campuses now routinely exceed US$10 billion-30 billion in total insured value, drawing power equivalent to mid-sized cities and financed through capital stacks combining equity, senior secured debt, and project bonds placed with pension funds, sovereign wealth funds, and infrastructure allocators.”
However, the executive outlines that while on the surface this looks like the kind of environment in which ILS and sidecar structures can find natural application, the risks are particularly large.
Given this, Moloney notes that there are a number of barriers that could potentially halt the ILS market’s transition into the space.
The executive argues that modellability is the hardest barrier, highlighting how the ILS market is built on exceedance probability curves derived from decades of physical event data, validated by third-party catastrophe models.
“Digital infrastructure risk—power supply interruption, Power Purchase Agreement (PPA) counterparty default, construction delay cascades, grid interconnection failure—has no equivalent actuarial foundation. The risk is also not stationary: the correlation structure between power supply disruption and data center revenue loss evolves as grid topology, contractual arrangements, and operational dependencies change. The primary insurance market has not yet accumulated the loss experience or developed the analytical frameworks that modeling at scale would require,” Moloney explains.
Additionally, the executive emphasises that sizeability is also structurally complicated.
“Individual campus risks are large enough for bond tranching, but the portfolio diversification assumption underpinning cat bond structures—geographic and peril independence across a pool of risks—does not hold cleanly for digital infrastructure.”
Moloney also highlights remuneration, and states that the yields available in a market with genuine supply demand tension should prove attractive to ILS investors, but cautions that pricing requires modellability, which in turn leads back to the first constraint.
The executive also notes that the build out of the digital infrastructure market will take time, given the sheer amounts of development and modeling infrastructure needed.
“The cat bond market itself did not emerge fully formed after Hurricane Andrew. It required a decade of primary market development, loss experience accumulation, and modeling infrastructure before securitization became viable at scale. Digital infrastructure risk is likely to follow a similar trajectory, compressed by the scale of capital pressure and the commercial incentives to solve the problem,” Moloney said.
He continued: “Three developments would materially accelerate ILS entry into this space. The emergence of standardized primary insurance products with defined parametric triggers—particularly around power supply interruption and PPA performance—would begin generating the loss data and pricing benchmarks that modeling requires.
“Portfolio aggregation across multiple campuses and counterparties could create the diversification structures that make tranching viable even without geographic independence. And power supply agreements themselves— which in their most sophisticated current forms function as long-dated financial instruments with defined credit exposure—are structurally closer to the risk that ILS investors understand than the operational risks of the campus itself.”
Furthermore, Moloney acknowledges that while the US$123 billion currently held in alternative reinsurance capital represents two decades of gradual institutionalisation of a risk class that already existed, the digital infrastructure risk pool is different in kind.
The executive importantly reminds that the digital infrastructure segment is being created from scratch, with financing structures that treat risk transfer as a prerequisite for deployment rather than an optional enhancement.
“Institutional lenders and bond investors in the project finance capital stack require that the exposure profile of any asset they finance be defined, transferred and contractually managed before commitment. Risk transfer is not incidental to financeability in this market—it is part of the definition,” he added.
“The demand dynamic that would drive ILS growth in this space is therefore structurally different from anything the market has previously experienced.”
Moloney contrasts this emerging sector with the traditional natural catastrophe ILS market, which expanded as capital markets identified an established, well-modeled risk class and incrementally assumed market share from traditional balance sheets.
Conversely, he notes that digital infrastructure is unfolding in reverse; the risk class is being generated at a velocity that far outstrips the absorption capacity of traditional insurance balance sheets, even as institutional capital stands ready to deploy.
“The constraint is not investor appetite. It is the absence of the primary market infrastructure—the actuarial frameworks, the standardized products, the structured triggers— that would allow that capital to be deployed.
“How quickly those conditions are met will depend less on the supply of capital, which is present, and more on the pace at which the insurance market develops the product architecture this new asset class requires. The sequencing matters. What is being built now in primary insurance markets is the foundation on which the next chapter of capital convergence will be written,” Moloney concludes.
