SÃO PAULO, April 28 (Reuters) – Large electrical companies in Brazil, such as Axia and Equatorial, are expanding the use of artificial intelligence, applying new tools in activities ranging from the operation of large infrastructures to administrative processes, with perceived annual gains reaching hundreds of millions of reais, executives told Reuters.
But this is not yet a widespread reality for the entire Brazilian electricity sector, consider experts, who point out uses that are still immature in general, in addition to a lower degree of specialization among professionals in the sector when using AI.
The largest energy company in Latin America, Axia (formerly Eletrobras) has implemented several AI-based solutions in recent years, mainly aiming to gain efficiency and increase the resilience of its enormous asset base, which includes dozens of hydroelectric plants and 74 thousand km of transmission lines.
Continues after advertising
AI applications at Axia involve, for example, climate models that assess risks associated with fires and extreme winds, aiming at preventive actions to avoid shutdowns of power transmission lines and other interruptions. They also perform administrative functions, such as auditing contracts and calculating legal liabilities.
All this work already produces a positive annual impact of R$100 million for Axia, considering new revenues, lower or avoided costs, the company’s vice-president of Technology and Innovation, Juliano Dantas, told Reuters.
Part of these developed solutions could be sold in the future to other companies or agents in the sector, he said, considering that there is still no clarity on how this could be done.
‘We are in the phase of consolidating internal use, and it is natural that these tools, especially those that contribute to the sector as a whole, are made available to other players.’
A step in this direction has already been taken by the electricity company with the inauguration, in Rio de Janeiro, of a ‘neocloud’, a cloud infrastructure that provides computing and processing capacity for highly demanding activities, such as machine learning.
The ‘AI factory’, which Axia says is the first in Latin America and was supported by the Electric Energy Research Center (Cepel), will have 96 GPUs (parallel processing units) and can be accessed by third parties, such as startups.
Continues after advertising
AI in energy distribution
One of the most transformational uses of AI for the electricity sector is in energy distribution, especially in combating commercial losses, a problem caused mainly by theft (energy ‘cats’) which leads to billion-dollar revenue losses for concessionaires, according to Falconi, a business management consultancy.
The Equatorial group, which has one of the largest energy distribution platforms in the country, works with a portfolio of more than 40 projects and innovation initiatives and is in the process of consolidating some uses of generative AI, said Maurício Velloso, director of Innovation, Customers and Services.
The main strategic fronts for the company in the area are linked precisely to the distribution business, with AI applications both for asset management (network monitoring, carrying out predictive maintenance) and to improve interaction with customers.
Continues after advertising
In the case of energy theft, the company has been using computer vision and satellite images to automate inspections and identify irregularities, said Velloso, stating that just one of the initiatives has already identified more than 415 thousand cases of fraud in all of Equatorial’s concession areas.
He further estimated that innovation initiatives have already contributed to more than R$185 million in direct financial benefits for Equatorial, between increased revenue and reduced costs, in the last two years.
Still limited use
Despite the growing adoption of AI tools by companies in Brazil, most of them still use them ‘very immaturely’, observing efficiency gains that are in fact far below the real potential of these technologies, according to Falconi’s Vice President of Operations, Marina Borges.
Continues after advertising
‘Including large organizations, they say they are adopting AI practices and when we come to understand what is being done, these are still very basic adoptions… Every time you still need a human to intervene in a process, we see a lot of opportunity for improvement’, he highlighted.
Specifically regarding the electricity market, she assesses that the use of artificial intelligence is still less mature than in other sectors, such as finance, and that companies have had difficulty capturing real cost reductions, transforming initial results into recurring ones.
This is also a reflection of less specialization of professionals in the energy sector to work with the new tools.
Continues after advertising
An analysis by the International Energy Agency (IEA), based on self-declared skills by professionals on LinkedIn, showed that the utilities, oil and mining sectors have lower levels of AI skills than other activities, such as education and financial services, in 43 countries.
In the case of Brazil, ‘utilities’ professionals show fewer skills linked to AI than in the manufacturing, oil and mining, financial services, education and IT and media segments, according to the IEA report, published last year.
