Energy AI: 89% expect returns, fewer than 20% scale with measurable impact

The ADNOC, XRG and Microsoft report Powering Possible 2026 (almost 400 senior energy and finance leaders, published 29 October 2026) finds that 89% increased AI spending over the past year. 63% expect returns on existing AI investments to cover or exceed costs, and 89% expect the same of future investments. Organizations running integrated agentic AI workflows rose from 15% to 33%.

Bain's ENR Executive Survey 2026 (859 executives across oil and gas, utilities, chemicals, mining and agribusiness) is more cautious: 68% expect AI to significantly reshape performance within 5 to 10 years, but under 20% report scaling AI with measurable impact. Nearly half name unclear outcomes or weak links to business value as the top obstacle.

The samples differ, so these are not a like-for-like trend, but together they place the gap between expectation and a measured P&L line. The nature of the value is also moving: supply stability and reliability is now the most-cited AI contribution (39%, up from 26%), while cost savings fell from 19% to 13%. Reliability is harder to put on an income statement. Bain's own example, an AI-enabled diagnostic on a mature offshore asset, produced more than a dozen initiatives worth over $15 million per year, each with a timeline, decision points and a named owner.

Cardan-AI recommendation: before the next budget cycle, give every operational AI use case one owner, one baseline and one P&L line, or a reliability metric converted into cost of downtime. Discount expected returns in business cases until a 90-day measurement confirms them. Read the full point of view.