A Single Fallen Power Line Revealed How Data Centers Can Destabilize the Grid – Here's the Fix

A Fallen Line, a 3.1 GW Disconnect, and Flickering Lights from Virginia to Chicago

On July 23, a power line went down outside Washington, D.C. Normally, the grid would self-correct within seconds. Instead, voltage across the PJM Interconnection – the largest U.S. grid operator, covering 67 million customers from New Jersey to Illinois – spiked as more than 3 gigawatts of data center load vanished in about 30 seconds. Lights flickered in multiple states, but no blackout occurred.

The trigger was data centers in Northern Virginia, home to the world’s highest concentration of such facilities. When they sensed a voltage dip caused by the line failure, their backup systems kicked in, disconnecting the sites from the grid. That rapid, near-simultaneous removal of 3.1 GW of demand – roughly 3% of PJM’s total load – left the grid with a surplus of around 3.49 GW. It took 11 minutes for operators to stabilize the system.

This was not an isolated event. A similar mass disconnection occurred in 2024, when 60 data centers simultaneously dropped 1.5 GW of load. Back then, data centers accounted for about 6% of PJM’s load. Projections show that share rising to 24% by 2040, making the grid’s vulnerability to such cascading trips an urgent infrastructure challenge.

Why Simultaneous Data Center Tripping Turned a Minor Fault into a Regional Voltage Surge

The Physics of a Mass Trip

Data centers protect sensitive IT equipment by switching to backup power when they detect voltage sags. The problem is that when hundreds of facilities – often clustered in the same region – make that decision independently and in less than a second, the result is a huge, instantaneous drop in demand. Because the grid must maintain near-perfect balance between supply and demand at all times, the sudden loss of load sends voltage soaring, which can trip more equipment or cause protective relays to activate.

“We need to figure a way for these loads that are located next to each other to sequentially either disconnect or reconnect,” said Ali Zain Banatwala, senior market models specialist at the Independent Electricity System Operator. Without coordination, each facility’s legitimate self-protection becomes a collective threat to grid stability.

ON. Energy’s Battery-Backed Ride-Through Solution

One technology pathway is to design data centers that do not automatically abandon the grid when a fluctuation occurs. Startup ON. Energy has developed a campus-wide uninterruptible power supply (UPS) that sits between the grid and the entire facility – servers, cooling, and auxiliary equipment. A large battery bank and power conversion system make the data center appear to the grid as a single, steady load that can absorb voltage deviations rather than react to them.

During a grid surge, the system uses excess power to charge its batteries; during a dip, it can dispatch stored energy to servers within milliseconds. This allows the data center to “ride through” events like the PJM line failure without contributing to the imbalance. ON. Energy is currently installing a total of 3 GW of its systems across four different data center campuses, signaling early commercial traction at a scale that matches the problem.

Regulatory Pressure is Accelerating

Grid operators are beginning to mandate ride-through capability. ERCOT, the Texas grid operator, is set to require large loads like data centers to remain connected and assist the grid during disturbances. The PJM incident, which was twice the size of the 2024 event, is likely to accelerate similar rule-making in other regions. Data center developers who fail to plan for these requirements now may face costly retrofits or delayed grid connections.

What Grid Operators and Data Center Developers Must Do Next

• Data center operators in PJM territory should immediately audit their facility’s ride-through settings. The ability to delay disconnection for short periods or to absorb power fluctuations – as ON. Energy’s system does – can prevent a facility from contributing to grid instability and may soon become a regulatory requirement.
• Grid operators should model worst-case simultaneous disconnection scenarios for data center clusters and set clear interconnection standards. ERCOT’s upcoming mandate provides a template; PJM and other ISOs can use the July 2026 event data to calibrate their own requirements.
• Developers of new data center campuses should evaluate campus-wide UPS architectures during design, not as a retrofit. With ON. Energy already deploying at 3 GW scale, the technology is moving from pilot to commercial viability, and early adopters may gain a competitive edge in securing grid capacity.
• Energy storage providers and investors should track the ride-through mandate trend. As regulations tighten, the addressable market for grid-supportive UPS systems at large loads will expand rapidly, creating opportunities beyond the four campuses ON. Energy is currently supplying.