Helix DC and Axiom Power to deploy 5 GW of AI UPS™ at multiple hyperscale campuses.

Helix DC, the AI factory company, and Axiom Power today announced a partnership to power the next generation of AI factories, committing 5 GW of AI UPS™ medium-voltage capacity across multiple hyperscale campuses.

The deployment places Axiom Power modules between the utility and the hall rather than downstream of it, so disturbance is absorbed in both directions: grid-safe on the supply side, waveform-clean where the compute actually draws. Waveform is corrected in 380 ms at P95, with no transfer the load can feel.

Why the power layer moved upstream

Hyperscale GPU campuses draw in steep, concentrated steps. Conventional low-voltage protection sits too far from the interconnection to catch a disturbance before it reaches the rack, and the utility sees a load profile it cannot model. Moving protection to medium voltage resolves both sides of that problem at once.

“They committed to a number and then published how it was measured. That is the new standard for compute in America.”
‍
Elena Marsh · VP, Infrastructure, Helix DC

Across the operating fleet, Axiom Power reports 99.982% uptime on a trailing twelve-month basis, weighted by capacity and read at the point of interconnection. Hardware is built on Tier-1 components with a twenty-year design life, manufactured in the United States and FEOC compliant end to end.

Schedule, not a queue

Owned manufacturing scales output to 10 GW per year, which makes supply a schedule rather than a position in a queue. Commissioning across the first two halls is complete, with no derating events recorded.