Castilla-La Mancha and UCLM promote measuring software energy use

The Government of Castilla-La Mancha and the University of Castilla-La Mancha have launched a collaboration to treat software sustainability as a measurable engineering concern. The agreement, announced on July 20, includes more than €140,000 in regional funding and focuses on training professionals, creating assessment tools and studying the energy consumption of applications used by the regional administration.

The initiative has four lines of work. The first covers training in software sustainability and sustainable artificial intelligence. The second proposes a framework for assessing whether a software program is more or less sustainable. The third includes support for the regional government when evaluating the energy consumption of its applications. The fourth consists of scientific and technology outreach for companies, professionals and citizens.

The project’s technical interest lies in moving the conversation beyond the data centre and toward application behaviour. The energy used by a service depends not only on the cloud provider or hardware type. It is also affected by the amount of work performed per request, algorithm efficiency, database queries, response size, memory use, execution frequency and the need to keep processes active. Artificial-intelligence systems add model size, inference duration and the cost of moving data between components.

Turning those variables into a useful assessment requires clear boundaries. A measurement should record which software version was tested, under what workload, on which hardware or infrastructure, for how long and with which energy metric. Without that context, comparing two applications can lead to misleading conclusions: one may use less energy per request while serving many more requests or completing a less demanding task. The work announced by UCLM and the regional government could be valuable precisely if it turns sustainability into a repeatable criterion within the development lifecycle.

The regional administration also provides a concrete testing ground. Evaluating public applications could connect architectural decisions with operational and environmental costs, and help prioritize improvements such as removing redundant work, optimizing queries, tuning scheduled workloads or choosing AI models that match the task. The goal is not to pursue a single number, but to include efficiency and consumption in design, procurement and maintenance decisions.

Editorial caution is important: the announcement describes an agreement and its workstreams, not published measurement results. The applications assessed, baseline metrics and savings are not yet known. The advance is the recognition that software has an operational footprint and that teams need instruments to observe it. If the framework becomes public and reproducible, it could help other organizations and companies in Castilla-La Mancha compare improvements using shared technical criteria.