A rigorous and fair cost recovery process#

Session 1, Room 2 Chair: David Sarmiento-Perez (Alan Turing Institute)


  • How do you fund the operation of TREs at your institution?

  • What activities and costs do you include in the budget?

  • Do you pass on costs to projects using TREs?

    • How do you determine the costs for each research project?

  • What happens when a project has a limited budget and paying for a TRE will impact its capacity to deliver?

  • How do we ensure projects include TREs in their application budgets?

  • Should TREs be centrally subsidised?

  • What are the restrictions/conditions that most affect your funding (costs and recovery) model?


  • Context: ATI ran a project for the codebase of TRE, which led to data access projects. There was no dedicated service team and the use cases created a need for cost recovery for set-up, ongoing support etc. It was difficult to balance proportionally against user-budget.

  • Flat fee for service provision, heavily subsidised, will not recover full cost, balance as a research institution not looking to run at profit. Also balance between developing and providing services.

  • At UCL, longitudinal social studies

    • all charges are with the funders, free for the researchers. Funding body against cost recovery (frowned upon)

    • The TRE is the one making the application for funding (other projects with heavy use do apply)

    • In favour of TREs being centrally subsidised (as a public service)

  • Worked on the cost recovery model, by projecting costs. It would be good to have clarity from the funders on cost recovery expectations (ie how much they will be funding projects)

  • Is access to a TRE done on a one by one basis or monthly or yearly?

    • ATI does so in a per TRE and month basis

  • The more we use AI how do we account for an increasing use of GPUs; and a second part, how will we evaluate model security?