Megawatts in months,
not a queue position in years.
Reference configurations of our REF-01 hybrid microgrid, sized for AI load. Each block is engineered to your site and load profile — fuel-flexible, phased, and built to bypass a multi-year interconnection queue.
Edge Block · 1 MW
Stand up a megawatt while the queue says 2031.
- Generation
- Gas genset or fuel cell + PV tie-in
- Storage
- 0.5–1 MWh BESS buffer
- Cooling
- Air / liquid-ready PDU interface
- Footprint
- 1× 40ft container + pad
- Redundancy
- N (optional N+1)
Fits: Edge inference, pilot clusters, single-rack-row sites
Pod Block · 5 MW
A self-sufficient power pod for a GPU hall.
- Generation
- Gas + fuel-cell hybrid, PV-augmented
- Storage
- 2–4 MWh BESS
- Cooling
- Liquid-to-rack CDU interface
- Footprint
- 3–4 containers, skid-mounted
- Redundancy
- N+1
Fits: Mid-size GPU pods, training annexes, colo expansion
Cluster Block · 20 MW
Bypass the interconnection queue at cluster scale.
- Generation
- Multi-unit gas + fuel cell, DC-native coupling
- Storage
- 8–12 MWh BESS, bidirectional
- Cooling
- Direct-liquid, 400–800V DC bus option
- Footprint
- Containerized farm, ~0.5–1 acre
- Redundancy
- N+1, phased capacity
Fits: Dedicated training clusters, neocloud build-outs
Campus Block · 50+ MW
Hyperscale power, energized years before the grid.
- Generation
- Phased hybrid plant, fuel-flexible
- Storage
- 20+ MWh BESS, grid-forming
- Cooling
- Full DC distribution, per-rack telemetry
- Footprint
- Multi-acre, modular phased build
- Redundancy
- N+1 to 2N, islandable
Fits: Hyperscale training campuses, sovereign AI sites
Start with a Power Audit
We size the block to your actual load, site, and grid position — and tell you honestly if behind-the-meter is the wrong answer. Go / no-go in 10–14 days.