Powering the AI era: Building resilience beyond the data center
Feature Article

Powering the AI era: Building resilience beyond the data center

Publish Date 22 September 2026


As AI drives unprecedented demand for compute, data centers are becoming larger, denser and more complex. Long gone are the days when 20 megawatts represented the energy ceiling for a supercomputer. Today, some projects are approaching gigawatt scale, reshaping decisions about where data centers can be built and how they are designed and operated.

At the second annual AI Compute Summit, held by Economist Enterprise in Amsterdam on May 19, 2026, industry experts explored what this transformation means for the future of data center development and resilience in a panel discussion sponsored by FM.

One message emerged clearly from the conversation - energy access has overtaken land availability and fibre connectivity as a primary constraint on development. Increasingly, data center operators have to go where the power is or find new ways to generate and manage it themselves. That makes energy resilience an imperative from the earliest stages of development.

“Energy resilience has become the make-or-break parameter for success in the AI compute ecosystem,” said Sandor Becz, senior vice president of research at FM.

A changing power equation

Data centers are developing a new, two-way relationship with the electricity grid. They can adjust consumption, generate power and help balance supply and demand. Developers are also considering hybrid energy strategies that combine grid connections with behind-the-meter generation, renewable energy and battery storage.

At gigawatt scale, that relationship brings new responsibilities. Data centers place significant pressure on existing grids, requiring closer coordination with utilities and government to support grid stability.

Power disruption presents the greatest resilience challenge. Conventional backup strategies may no longer be sufficient at this scale.

“The business model needs to support the idea that you’re not going to have 100% backup all the time,” said Adrian Oxley, principal engineer at FM.

Developers and operators need to understand what happens when part of the system becomes unavailable. Can failures be contained? Which operations take priority? How quickly can full operations be restored?

Finding those answers should start long before a facility comes online.

Designing resilience into the ecosystem

FM helps clients address these questions early with FM Intellium, combining insurance capacity, engineering expertise and scientific research in one solution to manage risk across the data ecosystem.

Developed specifically for data centers and their interconnected infrastructure, FM Intellium draws on more than 25 years of data center insurance and engineering experience. Today, FM engineers help protect more than 1,500 data centers valued at approximately US$620 billion.

Its whole-system approach starts with three fundamentals: place, protection and people. Where a facility is built, how its interconnected systems are protected and how prepared its operators are can all determine its ability to withstand disruption.

FM Intellium supports clients across the data center lifecycle, beginning with site selection and construction, when there is the greatest opportunity to mitigate risk. FM engineers work closely with clients to understand each facility's goals, constraints and operating environment.

By examining dependencies across data, power and renewable energy infrastructure, they can identify vulnerabilities, anticipate how disruption could spread and develop practical measures to strengthen resilience. This collaborative approach brings together the client's knowledge of its business and FM's engineering expertise to help shape resilience strategies tailored to each project.

That expertise is informed by scientific research into technologies reshaping data center operations, including liquid cooling, uninterruptible power supply systems, lithium-ion battery safety and automation.

People are equally critical. Human factors contribute to as many as half of data center outages. FM Intellium helps clients evaluate operational risks alongside physical infrastructure, including the people and processes responsible for keeping these advanced systems running.

From energy consumer to energy resource

Considering community needs and opportunities early in development can help operators identify ways to deliver tangible benefits and build stronger relationships with the people who live and work there.

With careful planning, the considerable heat generated by data centers can become one such resource. Siting facilities near homes and businesses creates opportunities to share that excess heat with the surrounding community.

The same site-specific thinking applies across data center resilience. Grid reliability, natural-hazard exposure, access to resources and regulation all vary by location, making these factors critical considerations from site selection onward.

During the AI Compute Summit discussion, Becz suggested that policymakers avoid blanket regulation, preserving the flexibility for data centers to address site-specific risks and respond to local needs.

FM Intellium draws on FM’s global experience across these disciplines to help clients assess the specific combination of natural hazards, power and water availability, infrastructure dependencies and operational challenges each site presents.

Building resilience for what comes next

Today’s data center operates within a complex, interconnected ecosystem of data and compute, power generation, utilities, renewable energy, cooling systems, supply chains and communities.

As the industry races to build the infrastructure for what comes next, resilience will help determine how far and how sustainably it can grow.

FM Intellium works with developers and operators across the lifecycle of that ecosystem, helping them identify risks early, strengthen resilience and prepare for the demands ahead.

You can learn more about FM Intellium here.