MINING DARK GRID CAPACITY.

Dark fibre powered the MODERN internet economy. Dark grid capacity will power the physical AI economy.

NASA, nighttime aerial view of New York

In the late 1990s, the future of the internet was quietly determined not by websites or software, but by something far more prosaic: dark fibre.

Vast quantities of unused, unlit optical cable lay beneath cities, dismissed at the time as excess capacity. Yet that dormant infrastructure became the foundation on which the modern internet economy was built. When demand finally arrived, scale was possible because the physical substrate was already there.

Without dark fibre, there would have been no economically viable path to companies such as Google, Amazon Web Services, Netflix, Facebook, YouTube, or modern cloud computing itself. Cheap, abundant, low-latency bandwidth enabled search at planetary scale, video streaming, global SaaS, hyperscale data centres, and ultimately the AI revolution. The internet of the 2000s and 2010s was built on fibre laid in the 1990s.

Today, the physical AI economy faces an analogous moment.

When people talk about grid capacity, they tend to picture the national electricity system: power stations, transmission lines, and gigawatts flowing across the country. But as with fibre two decades ago, the decisive constraint — and the decisive opportunity — lies elsewhere.

The true bottleneck in modern cities is not the transmission grid.

It is the distribution grid.

Beneath every city sits a dense lattice of primary substations, feeders, and local circuits. These networks were designed for a different era: predictable demand curves, modest peak loads, and limited electrification. They are now under growing strain — not because they are uniformly exhausted, but because their remaining capacity survives only in fragmented, highly localised pockets.

These pockets are the energy equivalent of dark fibre.

They are what we call dark grid capacity.

WHAT IS “DARK GRID CAPACITY”

Dark grid capacity is firm, connectable headroom on the distribution network — typically at or near primary substations — that exists but is not yet fully utilised.

It is:

• too small to interest hyperscale data centres

• too distributed to appear in national statistics

• too local to be visible from transmission-level analysis

But in aggregate, it represents GWs of latent value embedded inside cities.

Dark grid capacity exists because distribution networks were historically built with redundancy, uneven demand growth, and conservative planning assumptions. Electrification is now catching up — but not evenly. What remains is scattered and  precious.

WHY DARK GRID CAPACITY MATTERS NOW

Two structural forces are converging at the distribution layer.

1. Physical AI is moving into cities

The next wave of AI is not confined to distant hyperscale campuses.
It is physical AI:

• robots

• autonomous vehicles

• drones

• sensors

• cameras

• city-scale inference

• real-time coordination

This intelligence must operate:

• close to people and economic centres

• with ultra-low latency

• continuously

That means compute must move into cities, onto the distribution grid, not out to remote transmission-connected sites.

BUTTRESS is designed precisely for this reality: a dense mesh of urban compute nodes, each operating at the scale of dark grid capacity, aggregated into a city-wide intelligence fabric.

ELECTRIC HEATING AND COOLING DEMAND WILL SURGE AND CROWD-OUT PHYSICAL AI

At the same time, electricity demand from heating and cooling is accelerating rapidly.

Global energy agencies consistently project that:

• space cooling will be one of the fastest-growing sources of electricity demand

• urban HVAC loads will increasingly dominate peak demand

• electrification will concentrate stress at the distribution layer, not the transmission system

In cities, peak cooling demand and peak compute demand collide at the same time, at the same substations, on the same feeders.

THE INEFFICIENCY HIDING IN PLAIN SIGHT

Today, urban electricity is often used twice:

Traditional data centres:

• consume electricity for compute

• force buildings to draw additional power for HVAC

• reject this thermal energy (heat) into the atmosphere

Traditional buildings:

• consume electricity for operations

• consume even more electricity for heating and cooling (this will intensify as we move off gas)

WHY BUTTRESS CHANGES THE EQUATION: ONE MEGAWATT INTO TWO MEGAWATTS OF VALUE

BUTTRESS is built to exploit dark grid capacity efficiently.

Each unit of dark grid capacity consumed by BUTTRESS:

• powers physical-AI inference

• converts the power used by the compute to directly supply heating or cooling to buildings

• displaces separate electric heating and cooling

In other words, one megawatt of energy produces two megawatts of value.

From the grid’s perspective, this is not simply a new load.
It is load substitution, peak mitigation, and infrastructure leverage.

DARK GRID CAPACITY AS A STRATEGIC MOAT

Dark grid capacity is not renewable in any practical timeframe.

Once it is consumed:

• new capacity requires substation upgrades

• feeder reinforcement

• transformer replacement

• land acquisition

• permitting

• public consultation

• multi-year delivery timelines

• and often tens of millions of pounds per location

In dense cities, these upgrades are:

• slow

• politically fraught

• capital-intensive

• and frequently blocked by physical constraints

Every unit of dark grid capacity leveraged by BUTTRESS strengthens our moat.

As electrification accelerates ( EVs, heat pumps, cooling, robotics, autonomy) the remaining dark grid capacity will vanish. But the nodes that already control it will be structurally advantaged, insulated from future scarcity, and increasingly valuable to cities, utilities, and markets.

CONCLUSION: THE OVERLOOKED RESOURCE OF THE NEW URBAN ECONOMY

Cities are not running out of electricity.
They are running out of grid capacity.

Every industrial era has had a resource that mattered before anyone named it.

·       Coal seams before railways.

·       Ports before empires.

·       Spectrum before mobile networks.

Dark grid capacity is that resource for the age of physical AI.

It is invisible to most policymakers, ignored by most investors, and unusable by most hyperscale infrastructure models. Yet it is precisely this overlooked, distribution-level capacity that will determine where intelligence can exist, where autonomy can operate, and which cities can accelerate their economies.

Once dark grid capacity is consumed, it is often simply gone forever.

BUTTRESS is not betting on (or waiting on) more nuclear power stations, more transmission lines, or more distant hyperscale campuses. It is betting on the reality that the future economy — autonomous, robotic, AI-driven, and grid-constrained — will be shaped by where power can still flow.

BUTTRESS performs much more like an exoskeleton than traditional infrastructure because it converts dark grid capacity into both intelligence and thermal value, multiplying the usefulness of electricity at the point of scarcity.

This is how we win.