AI returns reach the energy sector, but grids remain the toughest test
Spending on artificial intelligence in the energy sector is rising, and its leaders say the return has begun to cover or exceed the cost. But the figure that sets limits on celebration is different: only 52 percent of participants in a new survey trust that the current infrastructure can meet the additional electricity demand that AI could generate.
The report is the third edition of “Empowering Capabilities” issued by ADNOC, XRG and Microsoft, titled “Resilience and Robustness of the Energy Sector in the Age of Artificial Intelligence”. It is based on a survey of roughly 400 leaders in energy and finance. Its core finding is that use is shifting from pilot projects to daily operations, not as a separate experiment within the company.
The return is advancing, but not aloneEighty-nine percent of participants said they increased their investment spending on AI during the past year, and 63 percent expect the returns on their current investments to cover or exceed the costs. Sixty-two percent placed AI among their top two priorities, compared with 42 percent in 2025. These figures illustrate the shift in focus from asking whether to try the tool to asking whether it improves supply reliability and supports decision-making.
The shift is also evident in assisted AI, i.e., workflow paths that execute decisions with limited human supervision. The share of organizations that deploy integrated pathways rose from 15 percent to 33 percent, while the share testing separate applications fell from 35 percent to 15 percent. In field operations, 60 percent of energy-sector participants said they have begun trialing or deploying self-operating and robotic systems, from infrastructure inspection to hazardous tasks.
The grid is the bottleneckSeventy-four percent expect AI to increase energy consumption by 2030. At the same time, confidence in infrastructure readiness did not exceed 52 percent. Consequently, electricity grid upgrades and smart-grid technologies ranked as the top investment priorities needed to support growth for 67 percent of participants, with energy storage second at 42 percent.
The shift in the value metric is notable as well. The share of those who see stability and reliability of energy supply as AI’s most important contribution rose from 26 percent to 39 percent, while the share prioritising cost reduction fell from 19 percent to 13 percent. Eighty-six percent say global energy supplies have become harder to forecast over the past five years, and 57 percent place improving energy-resource flow, load distribution and grid balancing at the forefront of areas where AI can drive transformation.
For the Gulf, where ADNOC frames the report within its operations and the ENACT council meets in Abu Dhabi ahead of ADIBK 2026, the findings leave a clear operational message: buying smart tools or measuring their financial return is not enough. Their safe expansion requires the electricity and digital infrastructure, cybersecurity, trustworthy data, standards, governance and skills to move together. This is not a decorative add-on to an AI project, but the conditions the report cites for its widespread and reliable deployment.
The next testThe survey does not rule that grids will fail to meet demand, but it reveals a clear confidence gap. Accordingly, the sector’s most important metric will be whether investments in grids, storage and digital infrastructure advance at the same pace as AI adoption.