The AI infrastructure buildout is colliding with a decades-old problem: the electrical grid's aging backbone is already worn out, and the surge in data center demand is accelerating failures across transformers, switchgear, and other critical equipment that utilities have neglected to maintain. According to industry experts, some transformers are already three-quarters through their operational lifespan, meaning the massive influx of power demand from AI facilities is straining infrastructure that was never designed to handle this load.

Why Is Aging Electrical Equipment Becoming a Crisis Now?

For years, utilities and industrial operators have underinvested in the maintenance of electrical assets like transformers, generators, and switchgear. That deferred maintenance has left critical infrastructure vulnerable at precisely the moment when demand is spiking. John Zuleger, president and CEO of Integrated Power Services, a company specializing in electrical equipment maintenance and lifecycle services, explained the cascading consequences of this neglect.

"We can see that demand is spilling over into steel mills and copper mines. I was visiting six copper mines, and they're all lifting their output against a base of mining equipment that has been under-maintained," said Zuleger.

John Zuleger, President and CEO of Integrated Power Services

The problem extends beyond just worn-out equipment. Hyperscalers like Amazon and Google are booking power capacity years in advance, which has created severe supply chain bottlenecks for electrical equipment manufacturers. Some original equipment manufacturers (OEMs) are so backlogged that they're delivering switchgear at only 70% completion, forcing companies like Integrated Power Services to perform last-mile manufacturing work that was designed to happen in controlled factory environments.

How Are Supply Chain Constraints Creating Maintenance Headaches?

The strain on equipment suppliers is creating a domino effect across industries. When major OEMs can't fulfill orders because they're prioritizing data center builds, other customers are forced to scramble for alternatives. Zuleger described a meeting with a copper mining company that attempted to place an $800 million switchgear order with their longtime supplier, only to be told the OEM couldn't accept it because of the data center buildout.

This supply crunch is pushing companies toward creative but risky solutions. Instead of replacing aging equipment, some are extending the life of existing infrastructure through targeted maintenance strategies. However, this approach creates a new problem: mixed fleets of equipment from different manufacturers, which complicates long-term maintenance planning and increases the risk of cascading failures.

Steps to Manage the Electrical Infrastructure Crisis

• Predictive Maintenance Programs: Using infrared energy analysis and other diagnostic tools to identify components reaching end-of-life inside electrical cabinets before they fail, allowing operators to plan replacements strategically rather than react to emergencies.
• Equipment Life Extension Strategies: Implementing comprehensive lifecycle services for transformers, generators, and switchgear to extend operational life when new equipment is unavailable, reducing the need for costly replacements during supply shortages.
• Emergency Response Agreements: Establishing service-level agreements with maintenance providers that guarantee rapid response times, critical for data centers that face penalties exceeding $20 million per day for downtime.

Data centers face particularly acute challenges because they must maintain 99.999% uptime, a standard known as "five nines" reliability. This requirement means data centers have complete redundancy, with dual circuits and backup systems. When an electrical failure occurs, technicians must navigate complex interdependencies to avoid triggering cascading failures across UPS (uninterruptible power supply) systems and backup generators.

"Our direct business with aftermarket emergency response for data centers is up 300% this year. We have agreements with hyperscalers where we have to acknowledge their call on a failure code within four hours, and then deploy our team within four to 24 hours," said Zuleger.

John Zuleger, President and CEO of Integrated Power Services

The financial stakes are enormous. While predictive maintenance costs a fraction of what a failure costs, the cost of downtime in a data center can reach $20 million or more per day. Yet despite these incentives, many industrial customers outside the hyperscaler ecosystem are unable to get support for preventive maintenance because they can't afford downtime while their products are in high demand.

What Are the Long-Term Risks of This Maintenance Backlog?

Zuleger raised a sobering concern about the future implications of the current crisis. Many of these new AI data centers have not been operating long enough to reveal whether current maintenance plans are adequate for handling failures. As AI becomes increasingly critical to life-saving applications like healthcare and emergency response, the consequences of data center downtime could shift from being purely financial to having human costs.

The situation is further complicated by the natural gas supply chain, which is also straining under AI infrastructure demands. According to energy analysts, natural gas consumption for data center power generation is expected to increase by 15 billion cubic feet per day over the next decade, meaning US data centers will consume more natural gas than most countries currently use. This surge in demand is expected to push natural gas prices higher, with Henry Hub, the benchmark for US natural gas pricing, forecast to approach $5 per million British thermal units by 2035, up from the $2 to $4 range that prevailed for most of the past decade.

Meanwhile, cooling infrastructure is also evolving to handle the density of modern AI workloads. Schneider Electric recently launched a new hybrid cooling unit designed to support data centers with up to 3.5 megawatts of cooling capacity per unit, with the ability to scale up to 20 units for data halls requiring 30 to 40 megawatts of cooling. These advanced cooling systems are critical because they help improve power usage effectiveness (PUE), a measure of how efficiently a data center converts electrical power into computing power, but they add another layer of complexity to the infrastructure stack that must be maintained.

The convergence of aging electrical infrastructure, supply chain bottlenecks, and surging demand creates a perfect storm for the power sector. Utilities and data center operators are caught between the need to deploy infrastructure quickly and the reality that the equipment they need is either unavailable or being installed in suboptimal conditions. The maintenance crisis unfolding behind the scenes of the AI boom may ultimately determine whether the infrastructure can support the computing demands of the next decade.