Energy: where the industry stands with AI
Adds ADNOC's planning agents, Arizona and Connecticut AI reviews, Brazil's AI returns; cuts stale surveys. Freelance, solo tools partial.
Where energy stands with AI
Energy in late 2026 uses AI mostly where it has for years: software that reads sensor data and warns of a failing pump, a damaged line or a leaking pipe before it breaks. What is new is AI agents, programs that carry out a multi-step task on their own, moving into planning, permits and commercial work at the largest operators. The proof still trails the promise. Nearly all the measured money comes from maintenance, forecasting and loss detection, and almost none yet from agents.
Adoption is wide and results are thin. Utility Dive, a US trade publication, reported in late October 2026 that AI agents have barely reached anything that touches physical equipment at US utilities, and that evidence of steady performance at scale is scarce.
What works today is predictive maintenance, image inspection, wind and solar forecasting, theft detection and customer email. What is still promise is agents that plan an oil field or run equipment with little supervision, and any measured saving from them. None of the tools named in this report says which AI model it runs on, so no efficiency claim here can be checked against the best current models.
What changed since late September
Agents reached the oil planning desk. On November 2 ADNOC, Abu Dhabi's national oil company, signed to use OpenAI's agent platform, Frontier, for planning field development and for commercial decisions on supply, prices and margins, with people signing off the consequential ones. ADNOC says development planning could drop from months to hours, its own estimate; no price or measured result was published. The deal came at ADIPEC, the oil and gas industry's annual conference. There, AIQ, ADNOC's AI venture, agreed with Siemens Energy, the German power-equipment maker, to build an offering for running plants autonomously. Its first product, RoboWell, runs a well's controls on its own on more than 500 wells, by AIQ's count. Era Haus covered both in Energy: AI agents reach oil planning; utilities keep people deciding. In the US, Constellation, the largest US nuclear operator, announced on October 6 a five-year alliance with Google to build AI tools for outage and maintenance schedules and for uprates, upgrades that get more power out of existing reactors (POWER magazine, October 2026).
US regulators started asking what is running. On October 21 Arizona's utility regulator, the Arizona Corporation Commission, held a public workshop on AI in utility operations; its October 27 summary says the state's two largest electric utilities stressed that people make the decisions and ruled out autonomous agents. On October 13 Connecticut's attorney general asked the state utility regulator to review how utilities use AI and who oversees it. New York's first inventories of every AI system its utilities use fall due around November 16, under a September 17 order.
What works in practice
Start from the sensors you already have. Equinor, Norway's state-controlled oil and gas company, says AI brought it $130 million of savings and added value in 2025, and that watching more than 700 pumps, compressors and other rotating machines through their sensors accounts for about $120 million of its gains since 2020 (Equinor, January 2026).
Aim at a loss with a cash figure attached. In Brazil, Axia Energia, the former Eletrobras, puts AI's benefit, new revenue plus lower costs, at about 100 million reais a year, and Equatorial, a distributor in the north and northeast, uses cameras and AI to find electricity theft and is extending the project to two more states in 2026 (Reuters, April 2026). Both figures are the companies' own.
Label the machine and keep a human route. Octopus Energy, a UK electricity and gas supplier, let an AI assistant answer about 8,000 routine customer emails a week. Customers rated 76% of its replies satisfactory, against 72% for comparable replies from staff (Octopus, July 2026). Every AI email was labeled, and any customer could ask for a person.
Write down who decides, and check the vendor. NRECA, the US association of electric cooperatives, updated its sample AI policy in July 2026 and is surveying members on their AI uses until November 15. A long contract is only as good as the company behind it: C3 AI, a US AI software company that has run Shell's maintenance program since 2018, saw revenue fall about 36% in its fiscal year to April 2026 and cut about a quarter of its staff (company filings).
On jobs, none of the large oil companies cutting staff in 2025 and 2026 has blamed AI; they cite prices and mergers.
What named companies are doing
PG&E, the northern California utility, opened a monitoring center in May 2026 that runs machine learning over grid sensors and about 5.5 million smart meters. PG&E credits that work with intercepting 17 potential wildfire ignitions in 2025 and about $6 million in operating savings that year (PG&E, T&D World).
Shell extended its maintenance program with C3 AI in June 2026. It covers more than 13,000 pieces of equipment, and the new stage adds software that proposes the cause of a fault and the fix. Shell has published no result for it.
Duke Energy, a large US utility, built AI tools with Amazon Web Services for the engineering checks run before a new solar farm or battery can connect to the grid. Duke says data preparation fell from about two weeks to hours, unmeasured by anyone outside the two companies (T&D World, September 2026).
In Argentina, YPF's La Plata refinery runs an AI system, built with the consulting firm BCG, that tracks fuel components in real time to blend gasoline and diesel; YPF said in September 2026 that blending time fell by up to half against an 18-to-20-hour lab routine. In Chile, the Spanish renewables company ACCIONA Energía says an AI-based control system on southern transmission lines raised their capacity by 60%, letting more wind power through (IndustriAmbiente, August 2026). Both figures are the companies' own.
The tools, by size of operator
Corporate. Large operators get AI inside platforms they already license, as Shell does with C3 AI. Emerson announced in July 2026 AI tools for Ovation, the control system in many power and water plants, to sort floods of alarms and explain stalled control sequences; they were not yet released in early November. Enverus ONE, launched in April 2026 by the US oil-and-gas data company Enverus, lets US teams query well and lease data in plain language. In Argentina, YPF renewed its contract with Corva, a US drilling-data software company, in September 2026 for a center where engineers watch live data from its wells.
Small business. The tools arrive through the vendor or the buying group. US municipal utilities get truck-mounted cameras from Noteworthy AI that photograph poles and flag defects, offered at group terms through two buying groups for city-owned utilities, one based in Ohio and, since October 2026, one in Florida that serves 32 of them. In Germany, about 100 municipal utilities can buy a private AI assistant from Deutsche Telekom under a framework agreement signed by Thüga, their shared services network (Telekom, April 2026). Solar installers meet AI inside their design software. Aurora Solar, used mainly in the US, drafts designs and customer proposals, and added heat pumps and batteries in September 2026. Reonic, a Berlin platform for installers in Germany and the UK, added an assistant in June 2026 that answers questions over the firm's own customer, project and price data.
Freelance and solo. Not enough verified evidence this version on what independent energy consultants, engineers and one-person installers use; no survey with a named method covers them yet, and the next update will add it.
What to watch over the next year
First, whether US inventories turn into rules. New York's filings arrive around November 16 and every six months after, and the commission has said it will then decide whether customers need more protection. Watch Connecticut's answer, and NRECA's survey results, the first picture sized for small utilities. If you sell services to a US utility, expect to be asked what AI your work uses and who checks it.
Second, how narrowly Europe draws its line. The AI Act's extra duties for "high-risk" AI in critical infrastructure (risk checks, human oversight, record-keeping) now apply from December 2, 2027, after the Digital Omnibus, a package of changes to EU digital laws, took effect in July 2026. The European Commission's draft guidelines of May 19 would cover energy AI only when it is a safety component at an operator its government has formally listed as essential; the final text is not out. In the UK the energy regulator Ofgem has promised a 12-month AI sandbox, a supervised trial space, from late autumn 2026; it has not opened yet. In Chile, a Senate committee approved the outline of the country's AI bill on October 26, ahead of a line-by-line vote.
Third, the fine print. Since January 2026, ISO, the body that writes standard US business-insurance forms, has offered insurers optional clauses that exclude harm caused by generative AI, and carriers are adding them at renewal (Insurance Journal, August 2026). In EU wholesale energy markets, any firm that trades by algorithm, AI agent included, must already notify its national regulator and ACER, the EU agency of energy regulators.
- 09-Nov-2026Energy: AI agents reach oil planning; utilities keep people deciding
- 28-Sep-2026Energy: New York orders utilities to list every AI tool they use
- 02-Sep-2026Energy: AI reaches the control room as the EU delays its deadline
- 21-Jul-2026Energy: the power grid fell short, and the backlash began