Our corporate site Contact us

Article

The market has appetite. The question is how much capacity is there?

Understanding the structural shift in a data center’s power requirements and what that means for insurability is now one of the most consequential questions in the market.

A different kind of asset

Five years ago, a $500 million data center was a headline-grabbing investment. Today, it barely registers on the market radar. A facility that would have cost $10 billion to build at scale in 2020 now carries a price tag closer to $14 or $15 billion before you account for the AI-optimized infrastructure that hyperscalers are racing to deploy.

This is not inflation in the traditional sense. There has been a structural shift in a data center’s power requirements, the generation assets it now encompasses, and the insurance programs required to protect it. The industry has not simply grown; it has been fundamentally reconstituted. Understanding what that means for insurability is now one of the most consequential questions in the market. . There has been a structural shift in a data center’s power requirements, the generation assets it now encompasses, and the insurance programs required to protect it. The industry has not simply grown; it has been fundamentally reconstituted. Understanding what that means for insurability is now one of the most consequential questions in the market.

Data centers are no longer just buildings with cooling capabilities

The most visible facet of this transformation is the emergence of the mega-campus. Where a large data center built five years ago might have occupied a handful of acres and slowly grew to 50 megawatts, today's hyperscale campuses routinely span hundreds of acres, target power capacity in the hundreds of megawatts (or even gigawatts), and carry construction timelines extending five to seven years from ground break to full operation. Some of the largest developments currently underway encompass over 1,000 acres and price tags exceeding $10 billion. These single projects built by single companies are not outliers but the new norm.

If cost and scale define the physical transformation of these assets, the more strategic change lies in data centers’ power profile. A typical enterprise facility a decade ago drew modest electricity from the local grid, largely unnoticed by the utilities serving it. Today, a single hyperscale campus can draw enough to rival the consumption of an entire city, and the aggregate demand across the sector is reshaping how national grids are planned and financed. Data centers are projected to drive 60% of total US electrical load growth through the end of the decade.

The consequence of this rapid increase has been acute: grid interconnection queues in many US markets now stretch five or more years, and power, rather than capital or land, has become the primary bottleneck in the sector's expansion.

The demand for power has driven a structural pivot toward behind-the-meter power, with dedicated generation assets built specifically for data center campuses that want to entirely bypass the public grid. Prevailing power sources range from natural gas turbines and solar-plus-battery storage to recommissioned nuclear plants and small modular reactors.

This shift is transforming the risk profile of a modern campus. A facility that was already among the largest and most complex construction projects in the economy is now also a significant power-generation asset. New risks include fuel handling exposure, environmental liability, and generation equipment failure, alongside the more traditional construction delay, property, and technology business interruption risks. The insurable value of these campuses has grown not only because these assets are bigger, but also because their risk profile has changed. Energy generation risk is now layered on top of significant construction and operational exposures, creating a combination of challenges that the insurance market has not previously had to evaluate at this scale and complexity. There is no established playbook. Instead, the risk is being understood in real time, as campuses are being built.

The capacity question: A measured yes

There is one frequent question that is posed as project sizes grow: Is there enough insurance capacity to support what is being built?

The answer is “yes”, but nuance matters. Marsh’s digital infrastructure specialists are currently observing increasing evidence that clients benefit from advanced engineered risk analytics and industry consulting expertise throughout the project lifecycle, from site selection through operations. This insight helps align financing capacity in the marketplace with the needs of clients and their equity and financing partners.

Without a comprehensive engineered risk analysis, data center owners and developers may not have a clear view of their risks. This can lead to inefficient risk decisions, whether through excessive retention, unnecessary transfer costs, or both, that can put additional pressure on their balance sheets and potentially slow their ability to scale.

The occasional difficulty in placing very large, complex programs is rarely due to capacity. The majority of insurers have an appetite for data center risks, but seek additional information and an analytical framework to understand what is in front of them. A behind-the-meter campus spans multiple underwriting disciplines — property, energy, construction, and technology — that have historically operated in entirely separate markets and were governed by different playbooks. When a carrier cannot fully evaluate the risk, they cannot fully commit to it.

Lack of information is the real constraint. To overcome this challenge, data owners and developers need to provide the necessary information that allows underwriters to engage with confidence rather than uncertainty, helping to unlock the needed capacity.

A prime example of coverage that is still emerging in the marketplace is insurance for graphics processing units Graphic Processing Units (GPUs) within the data centers. While GPUs are arguably the single most critical asset in a data center, we do not yet have a mature market geared toward insuring these critical pieces of infrastructure. Marsh specialists are working alongside industry experts to better understand residual values, insurability, and coverage terms to build and support a mature and accessible market for GPU cover.

More broadly, our teams are applying this approach across all lines of business within the digital infrastructure ecosystem. We are working with insurers and reinsurers to build the understanding and frameworks needed to keep pace with the sector's growth. That includes bringing insurers together around emerging risk types, helping underwriters assess the specific exposures created by behind-the-meter and mega-campus developments, and structuring programs in a way that makes complex, multi-line risks easier for markets to engage with. The result of these efforts is reflected in the market signals we are seeing: major carriers have expanded single-risk limits substantially, and alternative capital through Insurance-Linked Securities (ILS) and private markets is beginning to play a meaningful role in the sector for the first time.

Securing the needed capacity requires expertise, relationships, and a level of risk articulation that goes well beyond a standard placement, underscoring the importance of selecting a broker familiar with the industry.

From buying limits to designing capacity architecture

The next phase of digital infrastructure risk management requires a move from buying limits to designing capacity architecture, a deliberate structure across retained risk, commercial insurance, reinsurance, captives, contract transfer, and alternative capital that reflects each project's financing, construction schedule, and operational risk profile. The goal is not to transfer every dollar of exposure, but to make retained and transferred risk explicit, intentional, and aligned with the economics of what is being built.

Executing this kind of strategy requires capabilities that extend well beyond insurance placement — and it is here that the breadth of the Marsh platform becomes genuinely relevant. Marsh Risk structures and executes the market-facing placement and capacity strategy, with the relationships and cross-sector expertise to access markets that others may not be able to access. Guy Carpenter brings advanced catastrophe modeling, reinsurance strategy, and alternative capital expertise to bear on the portfolio-level risk questions that large campuses inevitably raise. Oliver Wyman provides the strategic perspective to help clients understand where the sector is heading and how to position themselves ahead of risks that are still emerging.

Contact us

For a more detailed analysis on how you can manage insurability and capacity challenges, and structure insurance solutions for complex digital infrastructure risks, contact your Marsh representative.

Data flow and network connections in digital space

Podcast

The market has appetite. The question is how much capacity is there?

Key takeaways

Innovation

Acceleration

Transformation