Building the intelligent power-modulation layer AI racks are missing.
XMight Smart Energy Corp. is focused on active power orchestration at the rack level for AI data centers. We give OCP open rack architectures — ORv3 / ORW today, Mt Diablo ±400 VDC / 800 VDC next — a rack-level coordination layer that delivers both grid compliance and compute headroom.
MISSION
We build the missing software layer between AI rack power and grid compliance — so the next 10 GW of AI capacity can deploy without breaking the grid.
THE PROBLEM
Two power crises inside every AI rack.
AI server workloads generate millisecond-scale power transients with instantaneous current peaks exceeding 2.5× steady-state load. To avoid tripping facility breakers, operators are forced into conservative power capping — stranding an estimated 15–20% of GPU compute capacity at any given time (XMight, The Active BBU: Dynamic Power Orchestration for Stable and Efficient ORv3 AI Racks, OCP Global Summit 2025).
The OCP ORv3 / ORW rack architecture distributes battery backup (BBU) across individual racks for energy efficiency, but provides no orchestration layer. On a grid sag, PSUs disconnect and BBUs operate independently; on grid restoration, PSU startup follows a randomized 0–5.5 s delay (ORv3 PSU specification) plus a cold-boot of several seconds. That is incompatible with the recovery windows North American grid codes now set — ERCOT requires at least 90% of load back within 2 seconds of voltage recovery (PFAPR).
Around 7:00 p.m. Eastern on July 10, 2024, a lightning arrestor failed on a 230 kV transmission line in the Eastern Interconnection, producing a permanent fault. Auto-reclose protection cycled six times within 82 seconds, each event depressing voltage to 0.25–0.40 p.u. in the local area. The cumulative disturbance caused approximately 1,500 MW of load to drop offline simultaneously — exclusively data center-type load, and disconnected entirely by customer-side protection schemes, not by utility equipment.
NERC published its formal incident review on January 8, 2025: Incident Review — Considering Simultaneous Voltage-Sensitive Load Reductions. The report's central finding: the grid has historically planned around large generator trips, but simultaneous load losses of this magnitude were never anticipated. As voltage-sensitive load (AI data centers in particular) continues to grow, this class of simultaneous disconnection now threatens system-wide frequency and voltage stability.
Which is why ERCOT's NOGRR 282 is now in force, and other operators are following:
- Ride-through and recovery — mandatory for large computational loads in ERCOT since 1 August 2026.
- Failure to ride through — triggers a root-cause investigation and a remediation plan; ERCOT can require disconnection until compliance is shown.
WHAT WE BUILD
XMight POM — active power orchestration at the rack.
XMight builds two products on POM, the open rack-level coordination interface it contributes to OCP:
Facility-level layer. Sets the power envelope for each rack from grid conditions and site policy, and records the per-rack evidence grid operators increasingly ask for.
Rack-mounted module in the ORv3 power shelf's PMI-class slot, with its own control plane. Coordinates BBU discharge and PSU output in real time, and keeps operating autonomously if the facility layer is unreachable.
XMight's POM implementation provides the intelligent power-modulation layer that ORv3 / ORW racks are missing — actively orchestrating BBU discharge to absorb workload transients, and coordinating PSU restart to deliver recovery inside the windows that NOGRR 282 and the grid codes following it define.
WHY THIS APPROACH
Three technical paths. Where each one fits.
| Dimension | Traditional centralized UPS | Software GPU throttling | XMight POM |
|---|---|---|---|
| Intervention layer | Facility-level | Compute-level | Power-level (rack) |
| Impact on AI compute | Zero impact | Forced GPU throttling — straggler effects and NCCL communication failures in synchronous training | Zero impact |
| OCP ORv3 / ORW compatibility | Decoupled from rack hardware | Requires deep integration with specific server vendor BIOS / BMC | 100% compatible, plug-and-play |
| PFAPR recovery window (ERCOT: 2 s) | Compliant | Probabilistic in software; physically difficult to guarantee | Deterministic hardware-level compliance |
| CapEx / facility rework | Significant — requires battery room build-out plus double-conversion space | Zero hardware cost (but with implicit compute interruption risk) | Native at the rack; lets facility UPS / BESS be sized for the services that need bulk energy |
| Retrofit downtime (existing sites) | Months to over a year — disconnect main switchgear, rebuild distribution panels, transformers, and battery rooms; for hyperscale AI sites the hourly revenue loss alone is substantial | No physical downtime (pure software deployment) | Rack-level hot-swap — the XMight POM Module is a POM-class module in the power shelf; install time measured in minutes per rack, no facility-level disconnection required |
XMight's choice: solve ride-through at the power layer, inside the rack — and let facility UPS or BESS do the jobs that need bulk energy. IT and OT fully decoupled.
INDUSTRY CONTRIBUTION
Where we sit in OCP.
-
OCP Startup Program 2026 member
Admitted to the OCP Startup Program 2026 cohort.
-
POM in review with OCP Rack & Power
POM (Power Orchestration Module) — an open rack-level coordination interface for grid ride-through and recovery — is in review with the OCP Rack & Power project. The XMight POM Module is XMight's implementation of POM.
-
Contributor, OCP DCF LVDC Power Distribution white paper v1.1
Named contributor to the OCP Data Center Facility LVDC Power Distribution white paper v1.1.0 (September 2026), section 8.1 Utility Grid — ride-through and recovery requirements as they reach the rack power shelf.
-
Presenter, OCP Global Summit 2025 Future Technologies Symposium
"The Active BBU: Dynamic Power Orchestration for Stable and Efficient ORv3 AI Racks" — the technical foundation of POM.
-
OCP Speaker, Staff Program member and Energy SME, 2025–26
Young Liang serves in the OCP Speaker and Staff programs and as a Subject Matter Expert in the Energy domain.
-
Exhibitor, OCP Global Summit 2026 Innovation Village
San Jose, 12–15 October 2026.
WHAT WE BELIEVE
Open standards win
We contribute to OCP because rack power specs should be open. Power orchestration belongs in a community-maintained spec, not a vendor moat.
Software closes the loop
The gap between GPU power transients and grid protection relays is a coordination problem. XMight's POM implementation fills it with rack-level coordination and the evidence to prove it.
Compliance is a product
ERCOT NOGRR 282 turns grid compliance from a checkbox into a continuous operational requirement. XMight's POM implementation closes the loop on it.
Founded by Young Liang, 2022 · LinkedIn ↗
Reach the team at service@xmight.com or visit the contact page for grid framework coverage and a direct line to engineering.