Sharing a facility should improve the work, not just the floor plan.

The starting point might be an instrument, a technical service, a training environment or a research-support capability. It should not be an equipment shopping list. We propose asking which resource potential users struggle to access, what that delay costs them and whether existing providers already offer a suitable alternative. UKRI's infrastructure context includes services and resources as well as facilities, which supports examining the need before deciding that a new building is the answer (UKRI, n.d.).

A shared facility should be compared with outsourced services, existing university or commercial provision and individual ownership where feasible. The comparison would include the user's full experience: access arrangements, travel, training, sample handling, reporting and the ability to repeat the work. A lower hourly charge may not be the decisive advantage. We want to identify where shared provision could create a meaningful improvement in research or development decisions, and where collaboration with an existing operator would be stronger.

Technical staff, maintenance, calibration, consumables, downtime and administrative effort can determine whether a shared resource is sustainable. NPL's measurement guidance supplies context for treating the quality of the result as part of the service (National Physical Laboratory, n.d.). The proposed economic model would distinguish capital cost from recurring operations and replacement. We would examine utilisation under conservative scenarios and make clear which costs depend on uncertain demand. A grant-funded purchase would not automatically create a viable long-term operation.

A staged approach could begin with access coordination, a limited service partnership or a small demonstrably needed capability. The first module should produce evidence about demand, scheduling, support burden and willingness to pay. That evidence could inform the later campus programme without requiring all infrastructure to be built together. The investment question is whether a focused resource can serve a real need reliably, rather than whether a large inventory makes the concept appear comprehensive.

The bounded next step is a requirements study and options comparison for one shared capability. Continuation would need identifiable users, a workable service model and credible recurring cost coverage under realistic utilisation. We would stop or partner elsewhere if existing provision is better, demand is too intermittent or the specialist burden cannot be supported. The resulting case should give capital partners a clear account of what the proposed resource would do, what it would cost and what remains uncertain before commitment.