--- title: "ADNOC Built AI to Squeeze Its Oil Wells, Then a War Made It a Product" description: "AIQ’s well-optimizing software became wartime infrastructure; now the UAE wants to sell it to the world" author: "Nate Ledger" published: 2026-10-03T23:00:00Z modified: 2026-10-04T05:01:48Z url: https://rews.cc/a/adnoc-built-ai-to-squeeze-its-oil-wells-then-a-war-made-it-a-775f28 language: en tags: ["ai", "oil", "energy", "war", "infrastructure", "business", "tech"] publisher: "Rews (https://rews.cc)" --- # ADNOC Built AI to Squeeze Its Oil Wells, Then a War Made It a Product *AIQ’s well-optimizing software became wartime infrastructure; now the UAE wants to sell it to the world* By Nate Ledger · October 3, 2026 · https://rews.cc/a/adnoc-built-ai-to-squeeze-its-oil-wells-then-a-war-made-it-a-775f28 ## In brief - UAE crude exports fell from ~5.1m to 1.9m bpd after the war began, recovering to 3.236m bpd by mid-September - AIQ’s RoboWell runs on 500+ ADNOC wells, lifting output 5% and cutting interventions by up to 50% - Neuron 5 predictive maintenance covers 1,200 pieces of equipment and has halved unplanned shutdowns - ADNOC awarded AIQ a $340m three-year ENERGYai contract in March 2025 covering 28+ fields - AIQ is targeting the US, Canada and North Sea and has identified roughly 100 acquisition targets For years, the pitch for artificial intelligence in the oil business has been a margin story: drill a little faster, catch a pump before it fails, squeeze a few extra percent out of a field you’ve already paid for. Nice to have. The UAE’s state oil company, ADNOC, recently got to run the other experiment — the one where AI stops being a rounding error on efficiency and becomes the thing that tells you how to run an oil company when a war has just removed two-thirds of your export route. According to Reuters and Kpler data cited by Oilprice.com, UAE crude exports fell from about 5.1 million barrels per day before the conflict to just 1.9 million bpd in March. By mid-September they had recovered to 3.236 million bpd, up from 2.886 million in August and 2.871 million in July, via a patchwork of the Habshan–Fujairah pipeline, expanded tanker operations and ship-to-ship transfers in the Gulf of Oman to keep crude out of Hormuz. Reuters has also reported ADNOC became a major buyer of discounted Iraqi crude, processing much of it at Ruwais to free up more UAE barrels for export. The company making the decisions inside that patchwork is AIQ, ADNOC’s AI venture, which by June had developed roughly 200 AI use cases inside the oil giant. The way to think about its product line is that each piece replaces a human loop that suddenly became too slow for wartime. Take RoboWell: it continuously adjusts producing wells using real-time data, rather than waiting for an engineer to intervene, and is now deployed across more than 500 wells. ADNOC says it has lifted well production by 5% and cut well interventions by as much as 50%. In peacetime, that means margin. In a war, it means that when a pipeline goes away, the question “which wells do we shut, which do we keep, where do we send the flow” can be answered in hours rather than weeks. AIQ’s CEO, Dennis Jol, put it plainly at a Semafor event in June: > We could have decided which wells we’re going to turn off today, which ones we’re going to pump, which pipelines were not working, and be able to reroute everything. Then there’s the machinery itself, which it turns out is even less tolerant of wartime surprises than the logistics. Neuron 5 watches pressure, temperature and vibration data from compressors, valves and generators to predict failures before they happen. Its initial deployment covered hundreds of pieces of equipment at the Northeast Bab field and the Taweelah gas compression plant, where pilots suggested it could cut unplanned shutdowns by 50% and stretch planned maintenance intervals by 20%. ADNOC expanded it to 1,200 pieces of critical equipment by the end of 2024 — full coverage is scheduled for 2027 — and says it has in fact halved unplanned shutdowns. The system also nailed its first overseas test: in Egypt, it flagged an impending failure in an electrical submersible pump 45 days in advance. Below ground, the numbers get more theatrical. In a 90-day trial of AIQ’s ENERGYai platform across two fields, one AI agent did seismic interpretation 10 times faster than conventional workflows, while another produced well-pressure predictions in 15 minutes. ADNOC was persuaded: in March 2025 it signed a three-year, $340 million contract to roll ENERGYai across its upstream operations, eventually covering more than 28 producing fields and thousands of wells. And even before the war, ADNOC said its 30-plus AI applications generated $500 million in additional value in 2023 through lower capital, operating and working-capital costs plus improved production. Now comes the part of the story that is really about business models. If you are AIQ, you have spent six years building software inside one of the world’s largest oil companies, with privileged access to producing assets and decades of proprietary operating data — the kind of training environment money cannot easily buy. And you have just acquired the best case study in the industry: “our systems helped run an oil major through a war that cut its exports by two-thirds.” The obvious next step is to stop being an internal tools department and start being a vendor. So that is the plan. Per Semafor, AIQ’s next product is Genesis, an operating system meant to run large-scale agentic AI across upstream and downstream operations — and deliberately model-agnostic, so customers aren’t married to OpenAI or Anthropic. The sales tour is underway: technology is being tested in Kuwait, India, Malaysia and Vietnam, while Egypt is discussing a dedicated “AIQ Egypt” venture that would pair the Egypt Upstream Gateway’s data with AIQ’s stack, with possible applications in hydraulic fracturing, horizontal drilling, well design and exploration. The rich-market push is where it gets telling. AIQ is targeting the US, Canada and the North Sea; according to Semafor it has made its first UK hire, is considering Houston, and has identified roughly 100 potential acquisition targets. Jol told Reuters the company has substantial cash to deploy and that spending it is “up front and center.” There is an irony available here for anyone who wants it: the oil industry spent a decade being told that Silicon Valley would disrupt it, and instead one of the more credible enterprise-AI sellers to emerge lately is the in-house software arm of a national oil company, battle-tested by an actual battle.