> ## Content Index
> Fetch the complete content index at: https://unhyd.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Drax’s £20 Billion AI Data Center Proposal Has a 100MW Reality Check
- URL: https://unhyd.com/article/drax-ai-data-center-yorkshire-100mw/
- Published: 2026-11-04T22:33:07.000Z
- Updated: 2026-11-04T22:33:07.000Z
- Description: A proposed Yorkshire campus could eventually exceed one gigawatt, but the first phase still needs a tenant, approvals and a final investment decision.
- Author: Unhyd Editorial
- Tags: Business, Drax, Brookfield, Data Centers, United Kingdom, #sidebar-popular-posts, AI Infrastructure, Electricity, #home-featured-slider

The **Drax AI data center** proposal could ultimately bring more than £20 billion in investment to North Yorkshire, according to the companies behind it. But a gigawatt-scale ambition is not the same thing as an approved project. The immediate test is a first phase of up to 100 megawatts that still needs a customer, regulatory decisions and final investment commitments.

On November 3, British electricity generator Drax [announced a strategic partnership](https://www.drax.com/news/strategic-partnership-for-gw-scale-data-centre/?ref=unhyd.com) with investment firm Brookfield and its UK-headquartered AI infrastructure platform Radiant. Brookfield's partnership with Qatar's Qai is expected to provide financing capability; Drax would contribute land and power-related infrastructure at its existing power-station estate near Selby. The proposal is important because electricity access and suitable sites can limit where new AI computing facilities are built. None of those advantages, however, guarantees that every planned phase will proceed.

## What the Drax AI data center deal actually includes

The [partners describe three stages](https://www.drax.com/news/drax-brookfield-and-radiant-announce-20bn-partnership-to-deliver-gw-scale-data-centre-capacity-at-drax-power-station/?ref=unhyd.com). Phase one aims to deliver up to 100MW of computing capacity with operations targeted for the first quarter of 2028\. Phase two is an ambition to add as much as another 500MW between 2028 and 2031\. A later phase would expand the site further, with more than one gigawatt of capacity the eventual goal. These are targets, not facilities already built.

Brookfield estimates the complete development could involve more than £20 billion in investment, including semiconductor equipment. It identifies a potential £2.5 billion associated with the first phase. The two figures are estimates for different scopes; neither should be described as cash already spent on a functioning data center. The agreement establishes partners and a development strategy, not an unconditional obligation to complete a £20 billion project.

The formal announcement makes the near-term conditions explicit: planning permission, regulatory and network approvals, an acceptable tenant, commercial agreements and final investment decisions by the parties. Drax said it was preparing a planning application and expected community consultation to begin in November. Those are more informative checkpoints than the ultimate capacity headline.

![Graphic comparing first-phase up-to-100MW capacity targeted for Q1 2028, up to 500MW additional between 2028 and 2031 and a long-term total exceeding 1GW.](https://storage.ghost.io/c/df/e4/dfe4d080-643c-453a-8a60-0452bdbc0efd/content/images/2026/11/drax-ai-data-center-phases-unhyd.png)

Drax and its partners propose a first phase of up to 100MW, an additional 500MW target, and later expansion beyond 1GW total. These are proposals subject to approvals and investment decisions.

## Why a four-gigawatt grid connection is not four gigawatts of AI capacity

Drax says its estate has access to a 4GW grid connection and 2.6GW of dispatchable electricity generation. The connection figure describes existing power infrastructure at a generation site, not a new data-center entitlement to consume four gigawatts. A computing campus needs its own agreed electrical arrangements, equipment, cooling, construction and commissioning before it can deliver services to customers.

The first 100MW phase is proposed to draw electricity directly from the grid. For phase two, the partners are exploring a different model in which generating equipment at Drax would supply the campus more directly under a long-term power agreement. These arrangements should not be conflated: phase-one grid imports do not prove that a future behind-the-meter supply deal is permitted or economical.

Unhyd has examined this distinction before in its [report on Nscale's Loughton project](https://unhyd.com/article/microsoft-nscale-loughton-data-center-deal-exit/), where a planned anchor customer and power-connection schedule became central questions. The underlying projects are different, but the analytical lesson is the same: available land, proposed capacity, a tenant contract and operating compute are separate milestones.

## Direct biomass power supply faces an additional government test

There is a specific policy complication beyond ordinary development approvals. In a [January 2026 written parliamentary answer](https://questions-statements.parliament.uk/written-questions/detail/2026-01-13/105278?ref=unhyd.com), the UK government said Drax must request permission before directly supplying data centers from its subsidized biomass units during the 2027–2031 support period. Without that permission, such direct supply would not be allowed under the arrangements described by the government.

That statement concerns direct supply from biomass generation, not the grid-powered first phase as announced. It matters because the later phases contemplate more direct use of the power station's generation assets. The November 3 announcement does not establish that those future permissions have been granted.

The station's biomass sourcing and emissions have also faced scrutiny. The House of Commons Public Accounts Committee [criticized weaknesses in the government's sustainability assurance regime](https://publications.parliament.uk/pa/cm5901/cmselect/cmpubacc/715/report.html?ref=unhyd.com) in 2025, while acknowledging that government treats sustainably sourced biomass as a lower-carbon alternative to fossil fuels. The debate cannot be resolved by describing the planned campus as powered by an existing energy site. Any credible environmental assessment must identify what electricity is actually consumed, when and from which sources.

## What a previous policy setback tells us

The partnership follows an earlier setback. In October, [Land and Climate Review reported](https://landclimate.org/drax-aigz/?ref=unhyd.com) that the government had rejected a proposed AI Growth Zone designation involving Drax. Drax told the publication that its data-center plans had never depended on winning that designation. The November partnership supplies a new development and capital partner, but does not reverse the reported designation decision or remove the planning requirements.

The companies say the campus could support local jobs and propose a voluntary community benefits fund, subject to consultation. These are stated intentions; the value, eligibility rules and administration of such a fund have not yet been established in the announcement. Residents will also need specific information about land use, water, traffic, cooling and electricity arrangements as planning proceeds.

## Four milestones that would turn the proposal into a project

The most useful evidence to watch next is concrete. First, an identified anchor tenant would establish a buyer for the initial computing capacity. Second, a formal planning application would define the first phase's physical footprint and environmental assessments. Third, documented grid and regulatory approvals would clarify how electricity reaches the facility. Fourth, final investment decisions would show that the parties have committed to build on agreed terms.

Independent [Reuters reporting](https://www.reuters.com/business/uks-drax-signs-data-centre-deal-with-brookfield-radiant-2026-11-03/?ref=unhyd.com) on November 3 corroborated the partnership and noted the project's potential value to Drax beyond the current biomass support arrangements. That commercial significance is real, but analysts' expectations and share-price movements cannot substitute for completed approvals or construction.

For wider context, Unhyd's examination of [data-center electricity forecasts](https://unhyd.com/article/data-center-electricity-forecast-2030-explained/) explains why proposed power capacity, actual consumption and future energy demand should be measured separately. The Drax announcement illustrates the distinction particularly clearly: a substantial grid connection may make an industrial site attractive, but the project still has to secure a customer and convert that advantage into permitted, financed and commissioned computing capacity.

The November 3 deal is therefore a meaningful move in Britain's AI infrastructure race, not proof that £20 billion of construction is now inevitable. Its first 100MW phase—and the conditions attached to it—will provide the earliest measurable test of that ambition.