EG, AJG: Data Center Liability Insurance Leaves Admitted Market

Data center liability insurance is moving out of the standard admitted market and toward the handful of carriers that have built the expertise to price it. On earnings calls held between February and July 2026, Everest Group (EG), Chubb (CB), RenaissanceRe (RNR) and Arthur J. Gallagher (AJG) all described insurance and reinsurance demand created by data center construction. Everest has moved its US wholesale liability book off the open-market lead layer and into higher attachment points on data center risks.

---

Where a liability policy sits in the tower sets its loss frequency and its price

Data center liability cover is written in layers, and the layer a policy occupies determines how often it pays and how much it can charge. Liability insurance covers harm the project causes to third parties, which is underwritten separately from the property cover on the buildings and equipment themselves. A large project's liability exposure is split into layers: losses hit the lowest layer first, and only losses above a set amount reach the excess layers above it. The lead layer attaches low and almost any carrier can write it, so pricing there has been competed flat. An excess layer is reached far less often, but it requires the carrier to judge how large the knock-on losses would be if that specific data center went down or caught fire.

Everest has repositioned its US wholesale liability book over the past two years. It used to write mostly commoditized lead umbrella business in the open market, and it has now shifted almost entirely into new data center risks, writing a $5 million umbrella limit where a $10 million limit used to sit and attaching farther up the tower [1]. Whether competitors can follow depends on having a team that can price the risk, not on how much capacity they can put up.

---

What three quarters of disclosure show about limits, pricing and book size

Everest has described the shift in more detail each quarter, moving from a loss-ratio answer to the size of the book. On the April 30, 2026 call, an analyst pressed the CEO on why the attritional loss ratio in Global Wholesale & Specialty — the insurance-side wholesale and specialty segment — had improved so much, and the answer turned on mix: the company now writes data center risks with a dedicated product and team, attaches higher, and gets "dramatically better" pricing [1]. Earlier, on February 5, 2026, management said in its reinsurance segment remarks that the global build-out of data centers, supporting energy capacity and other infrastructure has helped propel and diversify a specialty book now carrying roughly $2 billion in premium at an attritional loss ratio — the loss ratio excluding catastrophes — in the mid-80s [2]. These are two different segments, and the company has never broken out how much of either comes from data centers.

Gallagher, which places risk rather than carries it, points the same way. On July 30, 2026 it said that AI-related infrastructure including data centers, difficult liability risks and other emerging specialty exposures often do not fit neatly in admitted markets, which creates a multi-year opportunity for its wholesale teams [3]. On April 30, 2026 it said the fastest-growing part of the excess and surplus lines market comes from emerging specialty risks such as data centers and AI infrastructure [4]. One side is repositioning what it underwrites; the other is watching the same risk leave the standard channel.

---

Carriers that can price the layer choose it, and brokers pick up placement work

Pricing power over this risk sits with the ability to underwrite it rather than with the ability to supply capacity. Chubb said on February 4, 2026 that it has been writing data centers and that this is a global opportunity, describing itself as part of "a rare group" on capability, though the capabilities it listed sit on the property side, in builder's risk and operational property [5]. RenaissanceRe said the same day that it already reinsures data centers and that the new development is the number of mega projects requiring reinsurance or third-party capacity, which it called early stage [6]. Carriers that can underwrite the exposure can therefore pick their layer and their price, while brokers pick up placement work because the standard market cannot absorb these projects.

The boundary comes from Gallagher itself: on April 30, 2026 its CEO said that as a percentage of the overall market, this is not earth-shattering [4]. What would confirm the mechanism is whether Everest's Global Wholesale & Specialty attritional loss ratio keeps falling now that attachment points are higher, and whether the company starts disclosing the size of its data center book separately.

---

Companies exposed to this change

• Marsh McLennan (MMC): One of the largest insurance brokers globally, it places client risk with carriers. The 2,000 to 3,000 data centers under construction over the next couple of years that an analyst cited on Chubb's call was Marsh's figure [5]. Projects looking for underwriting capacity reach a broker first.
• Aon (AON): Also a global insurance broker, arranging liability cover and reinsurance for large infrastructure projects. The more risk the standard market cannot absorb, the more projects need a broker to assemble cover layer by layer, though Aon has made no corresponding disclosure about data center liability.

---

Sources

[1] Drillr · Everest Group (EG) · 2026-04-30 · Q1 2026 earnings call

"We're farther up in the tower. Pricing is dramatically better."

[2] Drillr · Everest Group (EG) · 2026-02-05 · Q4 2025 earnings call

[3] Drillr · Arthur J. Gallagher (AJG) · 2026-07-30 · Q2 2026 earnings call

[4] Drillr · Arthur J. Gallagher (AJG) · 2026-04-30 · Q1 2026 earnings call

[5] Drillr · Chubb (CB) · 2026-02-04 · Q4 2025 earnings call

[6] Drillr · RenaissanceRe (RNR) · 2026-02-04 · Q4 2025 earnings call

---

This is only meant to surface industry changes and companies you may have overlooked - it is not a stock recommendation.