Category

Can I claim my cloud and data costs in my R&D claim?

Our clients often ask if they can claim their cloud computing or data costs, and up to now the answer has disappointingly been no. But this has finally changed with the government’s announced update to the qualifying expenditure rules, allowing innovators to claim for the cloud computing that enables their work, and the data that powers their projects. 

Cloud computing has become vital for many businesses, in tasks such as hosting web applications or training machine learning models, and data is feeding the machine learning models used in many industries. Expensive data has also found applications in fields as diverse as high-fidelity digital twins for cities, modelling wind loads on skyscrapers in wind tunnels, and functional genomics. 

cloud, cloud computing, RD tax relief software, Novel

What were the rules before? 

Prior to the recent announcement, companies have been able to claim for their software and consumables costs. But due to data and cloud computing hardware not being  ‘consumed’ during a project or classed as software, these costs have been excluded from the claimable cost categories. This oversight has left the U.K. lagging behind other nations in providing funding for the industries which use data and cloud computing heavily in their R&D; notably in Artificial Intelligence (AI) and BioTechnology.

The definition of software costs in particular has become increasingly outdated, with current legislation designed for a world where software was installed on physical floppy disks and was tied to a single purchase. Modern software is typically downloaded directly from the internet, or accessed remotely through a web application, typically paid via a subscription. This Software as a Service (SaaS) model is treated ambiguously in the legislation, with many clients feeling unable to claim for software that is vital for their R&D projects. 

Further complicating the software rules is the rise of Infrastructure as a Service (IaaS)— where users rent servers for compute and storage in the cloud—which does not fall neatly into the category of hardware or software. HMRC’s recent announcement seeks to address this problem by enabling some cloud computing costs to be claimable for R&D relief. 

What costs will be claimable after the changes? 

The exact details of the qualifying data and cloud costs are yet to be revealed, but HMRC’s published consultation on the subject highlights several examples when data cost may be claimed: 

  • Data acquisition
  • Data cleaning and manipulation
  • Data storage—tied to cloud costs.
  • Cloud costs for training machine learning models
  • Cloud costs used in development phases of software project

Datasets vary in their cost of acquisition, with many being publicly available for free, or can require specialist providers such as in genomics where data is collected from expensive medical samples. The feedback received from HMRC’s consultation on qualifying expenditure suggests that data acquisition costs are likely to be included as a qualifying expenditure, to the extent to which it is part of a larger R&D project.  

Often the acquired data is unusable in its raw format, requiring significant resources to transform and clean the data into a format appropriate for the task. HMRC has given little indication of whether the costs associated with this task will qualify—although we would argue that processing data is essential for the R&D project to succeed, and would therefore qualify as an ‘indirectly’ qualifying activity. To learn more about indirectly and directly qualifying activity in a research and development project, click here.

We expect that once an R&D software project has been completed and a commercial product produced, the continuing cloud costs associated with hosting the web platform or storing data will no longer be claimable. Here it will be important to make a distinction between the development phases of a project and the deployment phase—a distinction that does not mesh well with the modern Continuous Integration, Continuous Deployment (CI/CD) method of software development. 

Data can be acquired by businesses in several ways, affecting how the costs might be classified. For example a company may set out to gather data themselves, thus incurring a staff cost. Or the company may contract another company to gather the required data, classifying it as a subcontractor cost. HMRC’s consultation also highlighted that a company may also pay for data under a licensing agreement, which enables them to use the data for the specific purpose outlined in the project, after which the data may be deemed as consumed—leading it to be classified as a consumable. In the end, the inclusion of data as a allowable cost may require a new cost category, or an update to the ‘consumables’ or ‘software’ category, to catch the various methods companies may use to acquire data.

Potential Pitfalls and Considerations 

The addition of cloud computing to the categories of qualifying expenditure does raise some new questions about the lack of inclusion for non-cloud hardware costs, such as the installation of a company’s own hardware for hosting data or running computations. Whilst capital expenditure does not attract R&D tax relief, there are other incentives on offer such as the recently announced Super Deduction which we expect hardware costs to qualify for.

Datasets that are purchased outright are less simple to classify into existing cost categories, as the data is generally still available for use again after being used, although it may lose much of its utility if the data is time-sensitive. This may require HMRC to deploy a new cost category, rather than shoehorning data costs into the existing cost categories.

The update to the qualifying expenditure rules for cloud and data costs is a significant win for many UK businesses. Although this revision is the correct step forward to ensure that the incentive reflects modern R&D processes, we are still waiting for draft legislation—expected in the 2022-23 Finance Bill—to confirm many implementation details that will determine the success of the proposed update. The changes are expected to take effect from April 2023 and we will continue to update our guidance as more information is published. 

More
articles