About the project

North Wales AI Infrastructure: accelerating productivity, innovation and growth

A Welsh Government–funded project building the region's first shared, business-facing AI facility — and an open invitation for Welsh SMEs and researchers to build on it.

The opportunity

North Wales businesses were locked out of shared AI compute

The UK's national AI compute for Wales — Supercomputing Wales — sits in Cardiff and Swansea. Bangor academics can reach it remotely, but businesses cannot access it directly as independent users.

Meanwhile, local companies increasingly rely on overseas commercial AI APIs, paying per token and sending confidential data outside their control. No equivalent shared AI development and deployment facility exists anywhere in North Wales.

This project closes that gap: an enterprise-grade AI platform owned and managed by Bangor University, made available to Welsh businesses and university research teams across Wales and beyond through structured collaboration.

Wales-hosted modelsAccess

Managed open-weight models — Qwen, GLM, DeepSeek or suitable successors — served securely from Bangor University.

Isolated workspacesPrivacy

Project-specific controls protect confidential data and prevent it being used to train external third-party models.

Edge AI trialsDeployment

Four portable DGX Spark systems rotate through partner sites for local, low-latency and offline testing.

Objectives

What the project sets out to achieve

Give business a direct route to advanced AI

Provide Welsh companies with advanced compute, hosted models, secure deployment trials, training and reusable workflows — without buying specialist hardware or depending solely on overseas APIs.

Convert demand into measurable productivity

Eleven partner-led cases apply the platform to concrete improvements across software, healthcare, environmental monitoring, professional services and education — each with baselines and before/after measurement.

Advance frontier AI research & talent

Support large-model fine-tuning, autonomous agent architectures, and postgraduate research — driving collaborative UKRI, EPSRC and Horizon Europe bids with academic partners across Wales and beyond.

Build a lasting regional capability

Bangor will retain and redeploy the assets beyond March 2028 — extending benefits across North Wales and the wider Welsh economy through case studies, benchmarks and new collaborations.

Funding & delivery

The project at a glance

Project title: North Wales AI Infrastructure: Accelerating Productivity, Innovation and Growth
Funding source: Welsh Government — SMART Flexible Innovation Support, Capital Equipment Fund
Lead organisation: Bangor University — School of Computer Science and Engineering (Registered Charity No. 1141565)
Project lead: Professor Zengbo (James) Wang
Equipment location: Bangor University, with four edge AI PCs on rolling deployment at approved Welsh partner sites
Industry partner: Eleven committed business and clinical partners, with access expanded during delivery
University delivery team: Eight named Bangor academic leads, supported by final-year undergraduate, master's, PhD and postdoctoral researchers

Timeline

Sept 2026
Funding secured
Successful SMART Capital Equipment Fund application led by Bangor University.
Nov 2026
Procurement begins
Compliant procurement of the platform and edge systems starts on 2 November 2026.
By Jan 2027
Delivery & commissioning
Installation, testing and commissioning at Bangor complete by 22 January 2027.
2027
Service goes live
Shared model-access service expected live within three months of commissioning, starting with pilot partners.
→ Mar 2028
Partner projects deliver
Rolling edge deployments and partner case studies reach measurable targets; assets return to the central pool by 31 March 2028.
Beyond 2028
Lasting capability
Infrastructure, website and engagement programme continue, supporting new collaborations and follow-on investment.

Expected impact

Productivity gains partners are targeting

Each partner project defines a baseline and a measurable target for March 2028. Selected examples from the project plan:

~30%

fewer low-value EEG recordings targeted in an epilepsy-pathway case

50%

less staff time per suspected pollution event, with faster intervention

67–75%

reduction in staff time for building energy analysis in a partner case

3–4×

increase in service capacity for an energy-analytics SME at the same headcount

These are the committed partners' own project targets — not guarantees for future participants. New SME projects set their own measurable goals during scoping.

Circular by design. Shared infrastructure avoids underused systems; rotating edge PCs, right-sized models and local processing reduce hardware demand, data transfer, energy and waste — while keeping AI skills and expenditure in Wales.

Want to see where AI could move the needle in your business?

Find out how the infrastructure works — or start the conversation today.

Infrastructure & technology Register your interest