Deployments

BBUs in the field, counted in racks and events.

A sampling of programs running VoltBridge hardware, described by type, region and scale. Client names stay under NDA — the numbers below are deployment scales, not marketing projections.

Browse deployments Featured program

Case index

Six programs, six operating profiles

Representative deployments; details under NDA. Each entry names the hardware, the region and the scale actually deployed.

Case 01 · North America

Hyperscale AI cluster

2,400 x VB-5125 across 200 AI racks, delivered in phased batches matched to rack integration. Modules bridge each rack bus while upstream power stabilizes, keeping training jobs alive through grid events.

Case 02 · Europe

Colo retrofit

480 x VB-5125 installed into live colocation rows during migration windows. Legacy strings retired row by row, so customer SLA exposure never extended past a scheduled window.

Case 03 · Southeast Asia

Edge deployment

90 x VB-E3 as per-node bridge in unattended edge sites. The sub-millisecond transfer carries nodes through grid instability and generator handoff with nobody on site to reset them.

Case 04 · Middle East and Africa

Telecom hybrid

1,200 x VB-5125 on -48V boards per ETSI EN 300 132-2, at hybrid sites pairing batteries with generators. Batteries take the fast transients; gensets are spared for the long hours.

Case 05 · Australia

University HPC

36 x VB-5125 bridging an on-campus HPC build-out where the grid connection waits in the interconnection queue. The bank protects compute runs through source switching on a constrained feed.

Case 06 · Japan

Utility pilot

24 x VB-H400 on a ±400VDC evaluation bay, validating the 800VDC-ready architecture and reviewing UL 9540A test data ahead of wider rollout.

Representative deployments; details under NDA

Container-scale AI deployment with rack rows equipped with VB-5125 battery backup shelves

Featured program

200 racks, one migration window at a time

The North America hyperscale program ships in batches, not a single wave. Each batch is commissioned rack by rack: parallel groups set in BMS configuration, CAN addressing mapped into DCIM, and protection-response checks run against the site's own procedures before a rack takes live load.

2,400 modules across 200 racks, phased with the integration schedule
Parallel groups and CAN addressing configured before units ship
Site acceptance repeats the same end-of-line checks as the factory
Per-batch QC records retrievable by serial number for every rack

FIG.P3 — Container-scale AI deployment, shown without site-identifying detail

Operator feedback

What operations teams report

Quotes are shared by role with permission; names and sites stay under NDA.

Bridged 14 grid events in 12 months, zero dropped racks.

Head of DC Operations — EU colocation

Commissioning held to the migration-window schedule — 200 racks handed over without a slipped window.

Program Director — US hyperscale cluster

The document pack was audit-ready on arrival: test summary, MSDS, transport papers, batch records.

Procurement Lead — SEA edge network