On the initiative of the Secretary of State for Digitalization and Artificial Intelligence of the Ministry for Digital Transformation and Public Service, the Spanish Association for Standardization (UNE) has published a new specification pioneering technique in the commitment to a more responsible and sustainable Artificial Intelligence within the framework of the National Program of Green Algorithms (PNAV).
The purpose of the standard is to establish a common framework for the measurement of energy consumption, carbon footprint, water consumption and the performance of AI systems.
Its development has involved the participation of a specialized technical working group, composed of more than 40 members of the research community in green algorithms, along with private companies, both large technology companies and highly specialized AI firms, as well as certification bodies that provide the verification perspective.
The specification provides a detailed guide to quantifying the environmental impact of AI models and algorithms at all stages of your cycle, especially during training and deployment, in both on-premises and cloud environments. It focuses, especially, on generative AI models and large language models (LLMs), due to their greater computational load and consequent environmental impact.
Through this contribution of standardization, a decisive step is taken towards a more transparent, measurable and efficient AI from an environmental point of view, providing a real and unified vision of its impact that allows to compare, optimize and validate the environmental performance of AI models rigorously.
This initiative, which is part of the Artificial Intelligence Strategy 2024, responds to the commitment of the Government of Spain for a responsible AI, evolving from the isolated methods of companies to calculate the consumption of their AI systems.
Sustainable impact
Defined metrics include indicators that help monitor the most critical phases of the process, such as the direct use of energy and water, as well as evaluating the overall efficiency of AI models during training and inference, which are those that concentrate the greatest environmental impact.
In this way, the specification focuses on the key points of the life cycle where most of the resource consumption actually occurs, providing a common and rigorous framework for its measurement and comparison.
International Good Practice
Although the specification has an impact at a national level, it is a key step towards future European standardisation, positioning Spain at the forefront of the definition of international standards in technological sustainability.
In addition, it responds to the demand of the European Union on the need to report on the energy consumption of AI systems established in the ‘Code of Good Practice for General Purpose AI Models’, of August 2025.
At the same time, Spain’s commitment to compliance with the European Artificial Intelligence Regulation is reinforced, contributing to transparency, traceability and sustainability in the design and operation of these technologies.
