PatterTech
1 / 8
Research

Whitepaper · Energy · Published by PatterPower

The co-located clean-energy park.

Clean generation, storage and flexible demand located together with co-ordinated optimisation and dispatch from a single local control brain.

Bring demand from compute markets to where the power is made, enabling revenue stacking and more granular participation across energy markets.

A whitepaper from PatterPower, a research venture of PatterTech.
Public · Whitepaper · June 2026

In brief

  • Put clean generation, storage and flexible demand on one site, run by a single local control brain.
  • Bring movable demand, like flexible computing, to where the power is made, instead of waiting years for new wires.
  • Co-optimising every asset as one position earns more from the same grid connection.
  • A public whitepaper from PatterPower, free to read, with no sign-up and no gate.

The deeper logic

Bring the demand to the source

Energy is costly and slow to move. Data is cheap and instant to move. Once that is taken seriously, a lot of the logic of the electricity system turns around.

We have spent a century carrying power to wherever demand happens to be. But moving electricity in bulk needs physical wires, and building them takes years and a lot of money. Moving data is close to free and close to instant, over the fibre that already criss-crosses the country.

So for demand that can travel, and flexible computing is the prime example, the cheaper move is to take the demand to the power. Run the work where the clean power is made, and send out only the result, as data.

How the optimisation brain routes energy and earns revenue££off-siteGenerationStorageConsumptionAI computeTradingGridOptimisation

Everything sits in balance. Generation runs the site and the AI compute, and the battery trims the peaks while spare power is sold to the grid.

Energy is hard to move but data is not: Bring demand to the source.

Why now

Three forces, one opening

Three things are happening at the same time, and a co-located park sits exactly where they meet.

The grid is full

The GB connection queue reached 125 GW by mid-2025, against a peak demand near 45 GW. New projects wait years for a connection.

Demand is surging

Global data-centre electricity use is projected to more than double by 2030, to around 945 TWh. The computing has to land somewhere.

Power is wasted

Great Britain curtailed roughly 10 TWh of wind in 2025, at a cost above £1bn. That is clean power switched off because there was nowhere local to use it.

The cheapest, fastest answer to a full grid is not always more wires. For demand that can move, it is to bring the demand to the cheap, clean power, and use it on the spot.

0 GW
GB grid connection queue, mid-2025
0 TWh
Projected data-centre use by 2030
~0 TWh
GB wind curtailed in 2025
>£1bn
Cost of that wasted clean power

Figures are illustrative, drawn from publicly available 2026 sources (Ofgem, IEA and others), and used as directional ranges rather than forecasts.

The building blocks

Four layers, one site

A co-located park stacks up to four layers. Each one is a real business in its own right, and all four are worth more when a single control brain runs them together.

Layer 01

Generation & storage

Core

Solar makes cheap power on site, and a battery shifts and firms it. Self-generated power beats buying at retail, and the battery is the firm base everything else stands on.

Layer 02

Flexible demand

Upside

Interruptible computing soaks up the cheap surplus, and EV charging adds a steadier use. Compute can be the most valuable activity on site, treated as upside on a sound base.

Layer 03

Markets & flexibility

CoreUpside

The battery and a controllable load trade power and sell flexibility to the grid, earning across the idle hours. Real but volatile, so it is treated with care.

Layer 04

The local control brain

The edge

Software that decides, minute by minute, what every asset should do, and trades the whole position. Local-first, so the site keeps working and keeps its data on site.

Each layer is useful alone. Run as one position by the control brain, they are worth more together. The compute hardware is typically funded by the computing tenant, not the site.

The edge

Why co-location wins

Run the layers separately and they can work against each other. Run them as one and the same kit earns more, from the same connection.

One co-optimised brain

  • One state, one decision: optimal outcomes across all assets
  • The same battery firms supply, arbitrages price and sells flexibility
  • Import when power is cheap, defer when advantageous
  • Every action weighed together, optimising for the whole system

Four separate optimisers

  • Sells excess power when computing could have used it later
  • Each optimiser acts for itself, blind to the others' intentions
  • Two assets end up competing for the same resources
  • Value leaks away without sight of the bigger picture

Four assets run by four systems are a collection. The same four, run by one brain that co-optimises and trades the whole position, are a single, more valuable asset. The combination, not any one layer, is the opportunity.

The tailwind

The rules are moving the right way

The rules keep moving in our favour. The recent changes all reward the same thing: flexible, behind-the-meter sites that can trade without running a supply business.

Connection reform

A tougher, faster queue

Ofgem's reform, advancing through 2026, brings tougher readiness tests and a use-it-or-lose-it approach. Flexible, ready sites are favoured over fixed new loads.

AI Growth Zones

Five zones, faster yes

The government has designated five AI Growth Zones in the UK, with electricity-cost support and faster planning. They signal where flexible demand and on-site clean power are most welcome.

P415, Nov 2024

Trade without a licence

A 2024 market change lets an independent operator trade a site's flexibility into the wholesale market, without running a licensed supply business.

Reform timetables slip and details change, so none of this is a guarantee. What matters is that several independent changes are all pushing the same way, toward flexible sites that can generate behind the meter and trade on their own terms.

In the world

Where it works

The idea is general, but it takes a handful of concrete shapes, built at different sizes and for different reasons.

The compute-led campus

Generation and storage built to feed a large, flexible computing load. The energy is there to power the work, and the surplus is traded.

The energy-led park

A generation and storage site that adds flexible demand to use power that would otherwise be exported cheaply or curtailed.

The behind-the-meter host

An existing large power user, like a factory or a data centre, that adds its own generation and flexibility to cut costs and earn from the market.

The temporary or mobile site

A containerised version assembled for a fixed period or a specific event, then moved. The same logic, packaged to be portable.

An honest note

This model is not universal. It suits demand that can flex, and sites where a strong connection and good generation come together. A site should stand up on its energy fundamentals first, with flexible computing as the upside rather than the foundation.

The research

Read the whitepaper

A public paper sets out the full picture. Free to download, with no sign-up and no gate.

In brief · 14 pages

Summary whitepaper

A short, visual run through the idea and why the pieces are worth more once a single brain runs them together. The quickest way in.

Download PDF

Published by PatterTech Ltd, Belfast · Classification: Public.

Built in Belfast

Who's behind it

PatterPower is a research venture of PatterTech, a Belfast technology company building local-first platforms for energy and AI.

Our work sits at the intersection of both energy and hardware optimisation, with both plugged into live markets.

The co-located clean-energy park is the first idea we find most worth exploring, however it is only the first of many.

Alongside this research we develop tools under the WiseWattage and PatterOS names.

If you'd like to talk

We'd welcome a conversation with:

  • Investors who want to understand the opportunity in more depth
  • Operators and developers exploring co-located sites
  • Partners across energy and computing
  • Researchers and policymakers working on the same questions

For readers who want to go further than this public whitepaper, a more detailed proposition and material are available on request under the appropriate terms.

Working on the same problem from another angle?

If the idea is useful to you, or you see it differently, we'd like to hear from you.

This is a public research paper, not an offer of securities, an invitation to invest, or investment advice. It describes a general concept through a worked example, and does not set out the specific plans, economics or methods of any single business. Every figure is illustrative, drawn from publicly available 2026 sources and used as a directional range rather than a forecast.