The two sides of AI's emergence in the renewable energy sector

AI is becoming a key tool for improving the efficiency of the global energy system and for ensuring the takeoff of renewables , in particular. "AI allows, on the one hand, to optimize generation forecasting (solar, wind) by adjusting weather and production models in real time; and, on the other, it also helps improve demand and storage management, as well as a near-instant adjustment between supply and demand, which increases system efficiency," explains Ismael Morales, Head of Climate Policy at the Renewable Foundation.

With this in mind, its potential is enormous, "as long as its application aligns with strict sustainability and social equity criteria , along with physical databases and its consumption of natural resources, which must also be supplied with renewable energy."

Miguel Colomo, head of predictive maintenance at Endesa, provides specific examples. "AI is present in many of our processes. For example, it helps us predict energy production for decision-making based on the wind, solar, and hydropower resources we will have at our plants," he begins by emphasizing.

"In predictive maintenance," he continues, "to anticipate failures and reduce unplanned downtime thanks to models based on sensor data." Its usefulness goes further: "In the classification of production losses with classification models that indicate the causes associated with our losses. In image diagnostics, which allows us to automate asset monitoring and bird detection using computer vision; or in the automation of tasks using conventional and generative AI applications." All of this, according to Colomo, improves efficiency, reduces costs, and increases network security.

At Smarkia, a company dedicated to the monitoring and intelligent management of energy data, they apply artificial intelligence to automate and optimize their clients' entire energy management, from data acquisition to participation in flexibility markets. In solar energy alone, this company manages more than 1,500 photovoltaic plants and over 5 GW of capacity. Thanks to AI, they help companies like MN8 Energy (one of the largest independent solar power producers in the US) improve data quality, reduce costs, and operate more efficiently.

"We automate critical alerts, integrate key signals (such as temperature or radiation), and offer comprehensive monitoring from a single environment," explains Marina Salmerón, CMO of Smarkia. "Our clients see immediate results."

He explains that, for example, in the retail sector, savings of an average of 5% per establishment and more than €2 million in rate optimization have been achieved. "In large international hotel chains, we've achieved an ROI of 162%, with 40% savings on industrial cooling and energy contracts. That is, in economic terms, for every euro invested, our client has generated €1.62 in savings."

For clients in the real estate sector, he claims that reductions of up to 40% in heating consumption have been achieved, while for clients in the agri-food industry , "thanks to the platform, savings of up to 28% in final product production costs have been achieved." And, to highlight just one more, "in a leading group in the leisure and entertainment sector, an ROI of 545% has been achieved with a payback of less than a month. In this case, for every euro invested, the return is 6.45 euros." In addition to energy savings, Salmerón emphasizes that AI "helps our clients improve operational organization, increase their ability to anticipate, make data-based decisions... it's real, measurable, and scalable efficiency."

But as AI applications in the energy sector are deployed, so are concerns. "The energy consumption of data centers hosting AI systems is growing faster than the capacity to generate renewable energy," warns Rafael Mayo García, head of Scientific Computing at Ciemat and coordinator of the joint program "Digitization for Energy" of the European alliance EERA. "And that raises an uncomfortable paradox : to digitize energy, we are using more and more energy. This year, artificial intelligence could consume as much energy as Finland. Some more pessimistic reports equate it to the expenditure of all of Japan," he notes.

Ismael Morales goes further. "Data centers will consume 10% of global electricity by 2030. Server cooling requires 660 billion liters of water per year. And accelerated obsolescence will generate more than five million tons of electronic waste." In his opinion, this cannot be sustained without strict sustainability criteria.

In this regard, Endesa has begun taking measures to minimize the impact. "We use edge computing to avoid sending massive amounts of data to the cloud, and we choose suppliers that certify their use of renewable energy and energy efficiency," explains Colomo. Smarkia, for its part, emphasizes that its AI is not generative, like ChatGPT, but rather uses algorithms designed to minimize energy consumption . "Our energy balance is clearly positive," says Salmerón.

Often, the perception that digitization means increasing energy consumption is a barrier. "We need to break the false dichotomy between technology and sustainability," Salmerón insists. "What we see in our clients," he says, "is that, applied judiciously, AI not only offsets their energy expenditure, but actually multiplies it in savings."

However, not all applications are created equal. The Renewable Foundation warns that many companies still do not audit the socio-environmental impact of their algorithms , and that opaque practices or practices focused solely on economic efficiency persist. "Ninety percent of companies prioritize efficiency over equity, and only 12% conduct environmental impact audits of their AI," Morales denounces. Therefore, they propose a series of measures: exclusive use of renewable energy, geographic planning of data centers, consumption and water footprint audits, circular economy equipment, and public transparency of consumption data. "And, above all, fair access to these technologies must be guaranteed. Because otherwise, vulnerable groups will be left out of the new energy model," he warns.

Even with its shadows, artificial intelligence is seen as a pillar for the future of the energy sector. "It will be a key enabler for integrating storage, automating critical decisions, and improving the competitiveness of renewables. It's not just a tool; in Endesa's case, it's a strategic part of our vision for the future," Colomo concludes.

From CIEMAT, Mayo points to the next milestones on the path: designing lighter algorithms, powering systems with clean energy, and bringing advanced computing closer to the entire industry, not just large companies. Ismael Morales places particular emphasis on the social perspective: "AI must serve a just energy transition . It cannot replace reduced consumption or widen inequalities. But if applied with transparency, equity, and ecological awareness, it can be a formidable ally."