Portugal is increasingly establishing itself as an interesting destination for the installation of large-scale data centres and Artificial Intelligence (AI) infrastructures, but there is a decisive factor shaping these investments: the capacity and reliability of the power grid. To address this challenge, INESC TEC has developed and validated a tool that identifies where the national transmission network can safely connect new large-scale data centre clusters, in the range of 500 MW to 1 GW. The goal is to turn a technical problem into an opportunity, making it possible to estimate the connection capacity available for energy-intensive consumers at each node of the grid.
The year 2025 ended with the news that Microsoft would be making a major investment in a data centre in Portugal. The technology company is set to install more than 12,500 processors in a mega data centre in the south of the country, in an investment reaching around 10 billion US dollars. Why this region? The concentration of renewables – Portugal, Spain and Greece are among the 10 countries worldwide where solar photovoltaic energy is expected to account for at least 1/3 of electricity generation by 2030 – and the strategic location are important factors. However, it is the capacity of the power grid that ultimately determines whether these projects move forward or stall. “With this validated tool, we can clearly and robustly indicate where the transmission network can support hundreds of megawatts today and where it will be ready by 2030,” said João Peças Lopes, Director and researcher at INESC TEC, and Emeritus Professor at the University of Porto.
But what exactly does this solution consist of and how does it differ from traditional approaches? The goal is to transform a technical problem into an opportunity by enabling the estimation of connection capacity for energy-intensive consumers at each node of the power grid. This advance moves away from generic or theoretical capacity estimates and instead analyses the entire national transmission network, classifying substations according to their real ability to accommodate large electrical loads.
The methodology combines rapid screening with detailed simulations, complying with the technical criteria of the transmission system operator, including contingency scenarios. This analysis is complemented by a second key element: an advanced model for generating electricity consumption profiles associated with data centres and AI computing infrastructures. This is essential because the impact of such infrastructures on the grid does not depend solely on maximum installed capacity, but on how energy is consumed hour by hour.
“Finding the right location for a 500 MW data centre does not depend only on the size of a transformer. It depends on how the grid behaves during critical hours and even under fault conditions. Our methodology tests these realities, translating them into a viable list of connection options,” explained Ricardo Bessa, researcher at INESC TEC, where he leads the Power & Energy Systems domain.
With this data, the solution will help to accelerate investment decisions and safeguard the security of the national electricity system, at a time when many projects face delays or increased risks due to uncertainty over grid access. Investors in digital and AI infrastructures will have access to clear information on where to build, at what capacity and within what timeframe, while grid operators will benefit from a strategic overview of where new demand is viable and where reinforcements should be prioritised.
The solution ultimately brings positive news for all stakeholders. In addition to investors, the electricity supply company and distribution system operators (DSOs) also gain insight into where network reinforcements should be prioritised, while renewable energy stakeholders are able to anticipate future demand clusters.
“This tool will allow us to quickly analyse the entire transmission network and then confirm, through full simulations, where connection is genuinely secure. We do not make assumptions; we demonstrate. By integrating realistic consumption profiles, the results will reflect the true behaviour of data centres,” emphasised Ignacio Hernando Gil, researcher at INESC TEC and the lead developer of the tool.
INESC TEC’s methodology has already been applied in real consultancy projects concerning the Portuguese power grid – for example, to identify the connection capacity of large electrolysers for green hydrogen – and is now being adapted and further developed to meet the needs of large-scale data centres, in collaboration with several industrial partners.
News, current topics, curiosities and so much more about INESC TEC and its community!
