Latos files Manchester neural-edge data centre
·

Latos files Manchester neural-edge data centre

Latos has submitted plans for a 28,000 sq ft Manchester data centre on a three-acre brownfield site, advancing its proposed network of smaller high-density facilities aimed at low-latency AI workloads.

Latos files Manchester neural-edge data centre
Summary
  • Latos has filed a Manchester planning application for two data centre buildings off Bower Street in Newton Heath.
  • The 28,000 sq ft proposal forms part of a broader plan for a distributed network of smaller AI-focused facilities.
  • The application advances the site into formal planning, although its IT load, grid connection, and detailed cooling design have not yet been disclosed publicly.

Latos Data Centres has submitted plans for a 28,000 sq ft data centre in east Manchester, moving a proposed three-acre brownfield site in Newton Heath from consultation into the formal planning system.

Latos Data Centres is seeking consent for two buildings off Bower Street, close to Ten Acres Lane, the Rochdale Canal, and the Lowry Park industrial area. Manchester City Council lists the proposal under planning reference PLA-2026-001350.

The developer describes the project as a “neural edge” data centre — its term for relatively compact facilities intended to place high-density compute closer to users and applications rather than concentrating every workload in a remote hyperscale campus.

The phrase is proprietary marketing language rather than an industry-standard facility class, but the underlying engineering proposition is familiar: reduce network distance, build at smaller urban or regional sites, and equip the facility for power-dense AI hardware.

Latos has said its wider neural-edge design can support high-density racks, advanced cooling, and modular deployment. Its first Stockton-on-Tees project is designed for up to 9.5MW of IT capacity in a 1,750 sq m footprint, with the company claiming rack densities as high as 250kW in some configurations.

The Manchester application does not yet disclose an equivalent IT load in the public material reviewed by DataCentral. Nor has Latos publicly specified the site’s final grid connection capacity, detailed cooling architecture, or the customer commitments that would underpin the build.

Those omissions are important because the physical constraints on an edge data centre do not disappear with a smaller footprint. High rack density can reduce the amount of floor space required for a given compute load, but it transfers pressure into electrical distribution and heat rejection.

An urban or near-urban facility must therefore solve a different geometry from a remote hyperscale campus. Land requirements are lower, but neighbouring uses, noise, plant screening, backup generation, transport access, and electrical infrastructure can become more sensitive planning issues.

Latos’ Manchester consultation material says plant and service areas would be screened and that the project would use energy-efficient cooling and modern technical systems. The site is currently vacant, giving the developer a brownfield-regeneration argument alongside its digital-infrastructure case.

The company is targeting delivery by 2030. It has separately set out ambitions for a national network of 40 facilities, including sites in Manchester, Birmingham, Leeds, and Glasgow, although the maturity of those individual projects varies.

The application also arrives in a more supportive national planning environment for data centres. DataCentral reported this week that England’s planning framework has strengthened policy support for data centre development.

National support does not remove local tests. Manchester still has to assess the scheme against ordinary planning considerations, while the developer has to demonstrate that the project can actually secure power, equipment, finance, and customers on a timetable that justifies construction.

That is particularly relevant for distributed AI infrastructure. Hyperscale campuses are typically underwritten by very large cloud companies or long-term colocation customers. A network of smaller facilities relies on demand being sufficiently distributed to support repeated investments across many cities.

If that demand develops, edge sites could complement rather than replace large central campuses — handling latency-sensitive inference, local processing, and selected sovereign workloads while training and bulk compute remain concentrated elsewhere.

The Manchester filing is therefore a more meaningful step than Latos’ earlier consultation because it places a defined site and design into a statutory process. The unanswered questions are now less about the concept and more about the engineering numbers: how much IT load the site can support, where the electricity comes from, and what cooling system will be required to keep high-density hardware within operating limits.


Stay updated with the latest insights and trends in the data centre industry by subscribing to our newsletter.

← Back

Thank you for your response. ✨