ADNOC’s War-Time Export Challenge Puts Its AI Systems to the Test

A system that once helped ADNOC get more oil out of its wells is now being tested against a different problem: keeping production moving during war. UAE oil exports initially fell from about 5.1 million barrels a day before the conflict to 1.9 million in March, according to OilPrice. For the state-run producer and its technology partner AIQ, tools built to improve efficiency have taken on a more urgent role in managing disrupted operations.

By mid-September, UAE crude exports had climbed to 3.236 million barrels a day, from 2.886 million in August and 2.871 million in July, OilPrice reported, citing Kpler data carried by Reuters. The recovery involved physical changes to the export network: ADNOC routed crude through the Habshan–Fujairah pipeline, expanded tanker operations and transferred cargoes between ships in the Gulf of Oman to move oil outside the Strait of Hormuz. Reuters also reported that ADNOC bought discounted Iraqi crude for processing at Ruwais, freeing more UAE crude for export.

From better wells to disruption management

Alongside those changes, AIQ’s systems supply real-time information on drilling, production and facilities. AIQ chief executive Dennis Jol told Semafor that the technology could help determine which wells to stop, which to keep running and how to redirect flows when pipelines become unavailable. That describes a decision-making capability, rather than a measured contribution to the export recovery. The source provides no figure separating AI’s impact from the new shipping and pipeline arrangements.

AI could help ADNOC choose which wells to shut and reroute flows when pipelines are unavailable, AIQ chief executive Dennis Jol told Semafor.

The technology already has a substantial operating footprint. By June, AIQ had developed roughly 200 uses for AI at ADNOC. Its RoboWell system uses live data and AI models to adjust wells continuously. OilPrice reported that it had reached more than 500 wells, lifting production by 5% and reducing well interventions—work needed to adjust or maintain wells—by as much as 50%. Those automatic adjustments allow production settings to change without waiting for an engineer’s intervention each time.

Maintenance is another test. Neuron 5 monitors pressure, temperature and vibration in machinery to flag problems before equipment fails. By the end of 2024, it covered 1,200 pieces of critical equipment, with company-wide deployment scheduled for completion by 2027. ADNOC says it has cut unplanned shutdowns by 50%. In Egypt, AIQ’s technology identified an approaching failure in an electrical submersible pump 45 days ahead, giving operators time to act.

A technology business beyond ADNOC

ADNOC is also expanding AI work below ground. During a 90-day trial of ENERGYai across two fields, one AI agent—a software tool assigned a task—interpreted seismic data used to map underground formations 10 times faster than conventional workflows. Another produced well-pressure predictions in 15 minutes. In March 2025, ADNOC awarded AIQ a three-year, $340 million deployment contract, with the eventual rollout intended to cover more than 28 producing fields and thousands of wells.

AIQ now wants other producers to adopt technology developed over six years inside ADNOC, using operating assets and decades of proprietary data. According to Semafor, its next step is Genesis, an operating system intended to coordinate task-performing AI across production and refining activities without locking customers into one underlying model provider. AIQ is already testing technology in Kuwait, India, Malaysia and Vietnam. Egypt is separately discussing a local AIQ venture.

The next milestones are wider deployment and international deals. AIQ’s target markets include the United States, Canada and the North Sea. Jol told Reuters that the company has substantial cash available for acquisitions. Semafor reported that it has made its first UK hire, is considering Houston and has identified roughly 100 potential acquisition targets. Its challenge is to persuade producers beyond ADNOC that systems built around one company’s fields and data can work in their operations too.