MODULAR INFERENCE BLOCKS

One AI rack, powered and cooled in a single container.

The ID units (Inference Distributed) combine Bluen microturbines, a battery, a DC converter, a dry cooler and an exhaust-driven absorption chiller. They arrive ready in a 40 ft HC container and need only gas and fiber.

Manufacturer specifications at ISO conditions, PCI (LHV). Performance varies with operating conditions.

ID-200 unit in a 40 ft HC container, renderRender
~75 m²of footprint per unit (15 × 5 m base)
PUE 1,22estimated in the design
~20 minof battery autonomy with all turbines stopped

Manufacturer specifications at ISO conditions, PCI (LHV). Performance varies with operating conditions.


HOW IT WORKS

How it works

Gas in.

The ID-70 runs three V70 microturbines with battery backup. The ID-200 and ID-400 configurations include an N+1 turbine spare.

Power and cooling on site.

Electricity goes to an 800 V DC bus with a 70 kWh battery. The rack water (45 → 55 °C) is cooled by a dry cooler; above ~36 °C ambient, an NH₃-H₂O absorption chiller driven by the exhaust handles the fine adjustment.

Tokens leave through the fiber.

The customer runs the rack; Bluen delivers power, cooling and availability as a service or as equipment.

Manufacturer specifications at ISO conditions, PCI (LHV). Performance varies with operating conditions.

Energy flow diagram of the ID-200 unit: gas, electricity, exhaust, rack water and airClick to enlarge
Energy flow of the ID-200 unit (2 × V200): gas, electricity, exhaust, rack water and air. The exhaust is never bypassed: it always passes through the chiller generator and leaves through a single stack at ~100 °C.

Manufacturer specifications at ISO conditions, PCI (LHV). Performance varies with operating conditions.

ID FAMILY

Three units, one 40 ft container.

ID-70, ID-200 and ID-400 product family
ParameterID-70ID-200ID-400
Turbines3 × V702 × V200 (1 + 1)3 × V200 (2 + 1)
AI racks1 NVL72-class rack (~140 kW)1 NVL72-class rack (~140 kW)2 NVL72-class racks (~280 kW)
Unit load (continuous / peak)170 / 207 kW170 / 207 kW341 / 414 kW
Installed generation (ISO / 35 °C)200 / 164 kW380 / 312 kW570 / 467 kW
Battery250 kW · 70 kWh250 kW · 70 kWh450 kW · 130 kWh
Absorption chiller (exhaust)126 kW126 kW237 kW
Gas consumption (normal operation)≈ 55 Nm³/h≈ 59 Nm³/h≈ 111 Nm³/h
Container1 × 40 ft HC1 × 40 ft HC1 × 40 ft HC
Design availability—~99,99%~99,97%

Manufacturer specifications at ISO conditions, PCI (LHV). Performance varies with operating conditions.

Cutaway of the ID-70 unit with three V70 microturbines in lineRender

ID-70: three V70 in line.

Three transverse V70 microturbines in 1.7 m bays, with battery backup. It is the first unit in the family. The image illustrates the interior layout only; roof-mounted cooling and exhaust heat-recovery equipment are omitted.

Manufacturer specifications at ISO conditions, PCI (LHV). Performance varies with operating conditions.

Cutaway of the ID-400 unit with three V200 turbines and two racksRender

ID-400: two racks, three turbines.

Three V200 (2 + 1) power two high-density racks in the same container. The image illustrates the interior layout only; roof-mounted cooling and exhaust heat-recovery equipment are omitted.

Manufacturer specifications at ISO conditions, PCI (LHV). Performance varies with operating conditions.

AI rack room with liquid cooling, illustrative imageIllustrative image

Liquid cooling ready for GPUs.

Water at 45 °C at the rack allows heat to be rejected with a dry cooler for most of the year, with no cooling tower and no water consumption.

Manufacturer specifications at ISO conditions, PCI (LHV). Performance varies with operating conditions.

Exploded render showing the container and roof-mounted equipment assemblyRender
MODULAR CONSTRUCTION

One unit. Integrated equipment.

The assembly view shows how the container and roof-mounted cooling and exhaust equipment fit together in the modular unit.

Engineering illustration of the modular unit.

Why on-site generation

Speed:

no grid connection queue; deployment in a container.

Scale in blocks:

add units as inference demand grows.

Resilience:

All units use a battery for load transients; the ID-200 and ID-400 also include N+1 turbine redundancy.

Location:

close to gas and fiber: wellhead gas, landfill biogas, industrial pipelines.

DOWNLOADS

Want to power AI racks where there is gas?

Talk to Bluen