
Data centres and AI inference
1.12–1.26 MW of IT per module: 8 liquid-cooled AI racks or one NVIDIA DSX POD. 45 °C water at the rack; ~1.0 MW of rack heat rejected without a compressor.
2 MW CCHP module
One containerised natural-gas module generates firm power, cooling and heat on your own site, without waiting for the grid.
Concept in development · design targets, to be validated on the pilot block

Why a turbine and not a genset: the heat leaves in a single hot stream that drives absorption cooling and still leaves ~1.7–1.9 MWt for steam or hot water. For loads of hundreds of kW or raw biogas, see the V70 and V200 microturbines.
Read why →How trigeneration (CCHP) works
One fuel produces electricity, cooling and heat. What the turbine would send up the stack goes back into the cycle or becomes a utility on site.
A 2 MW-class industrial turbine, running on natural gas, drives the generator.
Returns ~2.3 MWt of exhaust heat to the cycle: electrical efficiency rises from ~23% to ~33%.
After the recuperator, ~1.7–2.2 MWt is still left above 100 °C. That heat drives an absorption chiller, which makes chilled water without a compressor.
Whatever the chiller does not use becomes hot water or steam for the process, the hospital or a neighbour.
Efficiency: 31–33% electrical; up to ~80% when all the heat is put to use (cogeneration).


What one module delivers
Cooling and heat come from the same exhaust: what goes to the chiller does not go to hot water, and the other way round. Each site's design sets the split.
≈1,630 kW at 15 °C and ≈1,460 kW at 25 °C.
On hot days, exhaust heat cools the air the turbine draws in.
Available as hot water or steam.
In the data-centre configuration, from a 352 kW chiller driven by exhaust heat.
36–42% counting the cooling; up to ~80% with cogeneration.
Absorbs load steps, starts the turbine without the grid (black start) and carries the module for ~7 min if the turbine stops.
One module's consumption, from a pipeline or the distribution network.
About 340–430 TR, if all the remaining exhaust heat drives absorption chillers (COP ~0.7).
configuration estimate — sized per projectDesign targets per module, natural gas, sea level; the recuperated version will be validated on the pilot block.
Applications
The machine is the same in every case. What changes is where the electricity, the cooling and the heat go.

1.12–1.26 MW of IT per module: 8 liquid-cooled AI racks or one NVIDIA DSX POD. 45 °C water at the rack; ~1.0 MW of rack heat rejected without a compressor.

Electricity plus process steam or hot water from the same gas. Heat that would leave through the stack goes into the plant.

Absorption air-conditioning and electricity on site. The module shaves the peaks in grid demand.

Electricity, hot water or steam, and cooling on the same site, with continuity of supply for loads that cannot stop.

One plant for power, cooling and heat across several buildings, tied into the campus chilled- and hot-water networks.

Firm capacity and backup at points in the grid that need reinforcement.
NVIDIA, DSX POD, Vera Rubin and NVL72 are trademarks of NVIDIA Corporation.
Hundreds of kW
For smaller loads, with redundancy inside the block itself.
Distributed energy →MW range
For megawatt loads, with cooling and heat from the same exhaust. Several modules side by side serve larger loads.
Fuel
From a pipeline or the distribution network. Biomethane that meets the grid specification can also be used.
In development — already accepting up to 30% hydrogen blended with natural gas.
Timing
On a ready site: gas connected, environmental licence issued, foundations in place.
Built and tested at the factory before shipping.
Send us the electrical load, the cooling and heat demand, the gas available and the address. We'll do the first analysis.