Company About Us Leadership Careers Blog More Technology Data Centres Global Platform EnergyX DY-Building
Contact Us ↗

Data Centres

The AI Wave
Runs on Watts.

Compute is no longer the binding constraint on artificial intelligence. Power is. Data centres concentrate the building-energy challenge: power availability, heat, space and operating cost must be considered together. EnergyX applies Foundry and Intelligence to evaluate the building envelope, on-site generation and energy performance as a connected engineering problem, with project outcomes established against a defined design and operating baseline.

EnergyX DY-Building
01 — THE CONSTRAINT

Roadmaps are contracting, quietly, for want of electricity.

The limiting factor on the largest build-out in the history of computing is not silicon and it is not talent. It is interconnection queues, grid capacity and the price of a megawatt-hour. Models that should exist are deferred because the power to train and serve them is not available where and when it is needed.

A data centre is, structurally, an enormous roof and an enormous façade wrapped around a very large electrical load. Conventionally that envelope does nothing. It is the most valuable unused surface in the industry.

ElectricityGrid interconnection is now measured in years, not months, in most mature markets.
Energy costPower is the dominant lifetime operating cost of an AI facility, ahead of the hardware it feeds.
SpaceEvery square metre of envelope is already paid for. Almost none of it generates.
02 — THE INTEGRATED ANSWER

Two engines, applied to one building.

EnergyX does not supply a component into a data centre programme. It applies both engines — EnergyX Intelligence to evaluate thermal, power, generation, storage, manufacturability and economic options and verify a configuration; EnergyX Foundry to build the envelope that results — to the building, the plant and the dispatch strategy as a single engineering problem. Lifecycle services run beneath both.

01

EnergyX Intelligence

Evaluates the envelope, on-site generation, storage, cooling and cost as one model; simulates the candidate configurations deterministically; verifies the one that meets the facility's engineering requirements before anything is fabricated.

Verified before build

02

EnergyX Foundry

Perovskite absorbers on a sub-micron active layer keep vertical and curved surfaces viable generation area; DFMA panels arrive as finished components so the envelope follows the architecture and fabrication stays off the critical path.

Materials · fabrication · envelope

03

On-site generation

Roof, façade and canopy replace cladding with generation. Power is produced at the point of the highest load density on the site, alongside — not instead of — grid supply.

Generation at the load

04

Dispatch

Generation, demand, storage state and grid price forecast ahead, then dispatched against the workloads the operator makes available for scheduling.

Operator-governed

05

Lifecycle services

Siting and irradiance modelling before the build; performance monitoring, certificate monetisation and grid-services participation after it. Metered results feed back into both engines.

Beneath both engines

03 — DESIGN SCHEMES

Four ways the two engines land on a site.

Form follows the constraint that binds hardest — land, irradiance, planning envelope or proximity to load. The engines underneath do not change. Each scheme is a design concept; performance for a real site is established in a feasibility assessment, not read off a render.

Isometric render of a hyperscale data centre campus with photovoltaic roofs and façades and an elevated solar canopy over the yard.

SCHEME A — CONCEPT

Hyperscale campus

Low-rise halls with a fully activated envelope, and the service yard roofed by an elevated DFMA canopy so circulation space generates as well. Battery storage and substation sit inside the same optimisation loop.

Isometric render of a curvilinear modular data centre with a double-curvature photovoltaic envelope around a central courtyard array.

SCHEME B — CONCEPT

Curvilinear modular

A continuous double-curvature envelope, panelised by DFMA — the geometry a conventional flat module cannot follow. Capacity scales by adding arc segments, and the courtyard carries a ground array.

Isometric render of a stepped vertical urban data centre tower with photovoltaic curtain wall and arrays on every setback.

SCHEME C — CONCEPT

Vertical urban

Halls stacked on a constrained inner-city footprint, close to the load they serve. Every setback becomes an array and every façade becomes a generator, which is the only way a dense site recovers meaningful surface.

Isometric render of a high-irradiance edge data centre with photovoltaic halls and a large single-axis tracker array field.

SCHEME D — CONCEPT

High-irradiance edge

Sited against measured irradiance rather than assumed. The hall envelope and an adjacent tracker field together carry the base load through the solar day, with storage covering the shoulders.

04 — HOW IT COMPOUNDS

How the engines compound on one asset.

Verified design and physical execution, on the same asset, change what the facility is: a load that also generates, forecasts and — where the operator chooses — trades.

Materials

Perovskite makes the surface worth cladding

A sub-micron, full-spectrum absorber performs on vertical and curved surfaces and in diffuse light, where conventional modules fall away. That is what turns a façade from a token gesture into real generation area.

Manufacture

DFMA makes it buildable at data-centre schedule

Factory-laminated panels arrive as finished components. On a programme where every week of delay is deferred revenue, moving fabrication off the critical path is worth as much as the generation itself.

Envelope

BIPV puts generation at the point of load

Power produced on the roof of the hall does not queue for interconnection and loses far less in transmission. It offsets demand at the meter, in the hour it is generated — reducing what the facility draws from the grid, not replacing the grid.

Intelligence

Verified first, then dispatched

Before the build, Intelligence evaluates thermal, power, generation, storage, manufacturability and economic options and verifies a configuration against the facility's baseline. In operation, with a forward view of generation, demand and grid price, deferrable workloads — batch training, checkpointing, non-latency-critical inference — can be scheduled into the hours when on-site power is abundant, where the operator permits and the workload allows.

Lifecycle

Metered results close the loop over the asset life

Predicted output is reconciled against metered output continuously, so degradation and underperformance surface as flagged assets. Efficiency certificates and grid services are monetised rather than left on the table.

Two metrics, kept apart. PUE measures total facility energy divided by IT energy; it is set by cooling and electrical design, climate, load and the measurement boundary, and on-site generation does not lower it. On-site generation is reported separately — annual energy coverage, distinct from peak output or daytime coverage — against a defined baseline. Both are established per project in a feasibility assessment, not shown as headline figures.

Discuss a data-centre feasibility assessment →

A feasibility assessment takes the site, IT load and utilisation profile, planning envelope and grid arrangements as inputs, and returns a verified envelope configuration, expected generation range with its assumptions, cooling and electrical design basis, and whole-facility economics.