Loading...
Loading...
Oklahoma occupies a unique position in U.S. oil and gas — it is simultaneously a major producing state, the headquarters state for three of the country's most influential upstream and midstream operators, and home to Cushing, Oklahoma, the pricing and storage hub that defines the WTI crude benchmark traded globally. Devon Energy and Continental Resources, both headquartered in Oklahoma City, have concentrated their primary development programs on the STACK (Sooner Trend Anadarko Basin Canadian and Kingfisher) and SCOOP (South Central Oklahoma Oil Province) plays in the Anadarko Basin — one of the most intensively drilled and AI-analyzed oil and gas plays in North America. ONEOK, headquartered in Tulsa, operates one of the country's largest natural gas gathering and processing networks with extensive Oklahoma presence, particularly in the Midcontinent region where its infrastructure feeds both Anadarko Basin gas and the liquids-rich production from the STACK's condensate window into its NGL pipeline system. The Oklahoma Corporation Commission, which regulates oil and gas production through its Oil and Gas Conservation Division, maintains one of the country's most comprehensive public well databases and an active enforcement program for spacing rules, shut-in well compliance, and environmental monitoring. The Cushing, Oklahoma storage hub — where roughly 80 million barrels of crude storage capacity sits at the intersection of multiple major pipeline systems — is the physical delivery point for WTI futures contracts and an AI optimization target in its own right.
Updated June 2026
The STACK play in Canadian and Kingfisher counties and the SCOOP play in Grady, Stephens, and McClain counties present complementary but distinct AI reservoir challenges. STACK wells target the Meramec, Woodford, and Osage formations in a stacked-pay configuration with high formation pressure and oil-saturated zones that produce strong initial rates but steep early declines — a EUR prediction problem that ML models handle better than type-curve averaging across broad acreage positions. Devon Energy, which has one of the largest STACK positions among public E&P companies and an active data science program run out of its Oklahoma City headquarters, has been among the most transparent major independent operators about AI integration in reservoir management. Devon's 2023-2024 annual reports and investor presentations have specifically referenced ML completion optimization and production surveillance automation in the STACK/SCOOP context. Continental Resources, Harold Hamm's company, has similar data maturity in the SCOOP play, where its large Grady County position benefits from ML production forecasting and completion design iteration trained on thousands of proprietary wells. In practice, the gap between manually designed STACK completions and AI-optimized designs has narrowed across the industry as operators have published their learnings, but the in-fill spacing and interference modeling problem — determining optimal parent-child well distances in a stacked pay environment — remains a live area where ML models consistently outperform deterministic engineering judgment.
ONEOK's Oklahoma midstream network — gathering systems across the Cana-Woodford, Arkoma Woodford, and STACK/SCOOP producing areas, plus its WesTex and Midcontinent NGL transmission pipelines — operates at a scale where AI optimization has direct financial consequences measured in tens of millions of dollars annually. Compressor station efficiency on ONEOK's gathering systems directly affects the cost-per-Mcf of gas processing — AI-driven compressor scheduling that accounts for inlet pressure variation, processing plant throughput constraints, and NGL market pricing has documented lifting-cost reduction of $0.03-$0.08 per Mcf at comparable midstream facilities, which compounds significantly at ONEOK's volumes. The Cushing hub presents a different optimization problem: storage scheduling across 80+ million barrels of capacity, coordinating injections and withdrawals with WTI calendar spread pricing, pipeline nominations from multiple upstream systems, and refinery demand from the PADD 2 distribution network. AI scheduling models that integrate commodity price signals, pipeline capacity availability, and inventory positioning allow Cushing storage operators to capture contango spreads more reliably than manual trading desks. Magellan Midstream (now part of ONEOK following the 2023 acquisition) operated significant Cushing storage capacity, and the combined entity's optimization capability is one of the more sophisticated pipeline AI programs in the Midcontinent. The Oklahoma Corporation Commission's pipeline safety division, which oversees state-jurisdictional gathering lines, sets the compliance floor for pipeline monitoring and anomaly reporting that AI tools must satisfy.
Oklahoma's seismicity surge between 2012 and 2017 — driven by large-volume saltwater disposal well injection in the Arbuckle formation — resulted in the Oklahoma Corporation Commission's seismic response protocol, which created a regulatory mechanism to restrict disposal well operations when seismic activity exceeds defined thresholds. This protocol creates a specific AI compliance demand: real-time seismic monitoring correlated with active disposal well operations, automated injection volume and pressure logging against OCC threshold criteria, and predictive risk scoring for disposal wells operating near known fault structures. AI platforms integrating Oklahoma Geological Survey seismic data with operator disposal well records have become standard tools for larger operators managing high-volume water disposal programs in the Anadarko Basin. The OCC's electronic data submission requirements for drilling permits, production reporting, and environmental compliance create additional AI compliance workflow value — operators with 100+ active wells report that manual OCC reporting consumes several FTE positions that AI-assisted reporting tools can partially replace. The University of Oklahoma's Mewbourne School of Petroleum and Geological Engineering in Norman is the state's primary oil and gas technical research institution, with ongoing work on induced seismicity, completion optimization, and reservoir characterization that informs commercial AI model development across the Anadarko Basin.
Connecting AI systems to existing business infrastructure and workflows
Predictive models, data analysis, and ML pipeline development
Image recognition, object detection, video analysis, and visual inspection systems
Bespoke AI solutions, model fine-tuning, and custom model development
Devon Energy has concentrated AI investment in STACK completion optimization and real-time drilling decision support — its Oklahoma City data science team has been building proprietary ML models on Meramec and Woodford well data for several years, with investor-reported results on completion design. Continental Resources has applied AI more broadly to SCOOP production surveillance and in-fill prioritization across its large Grady County position. Both companies participate in Anadarko Basin technical conferences where their AI program details occasionally surface, making those venues the best source of current program intelligence.
Cushing is the physical delivery point for NYMEX WTI futures contracts, meaning storage scheduling decisions at Cushing directly interact with commodity futures pricing. The contango or backwardation of the WTI forward curve determines whether injecting or withdrawing storage inventory is profitable — AI models integrating price signals, storage capacity, and pipeline nomination windows optimize those decisions better than manual trading analysis. Cushing's connectivity to 15+ incoming and outgoing pipeline systems makes the nomination and scheduling optimization problem uniquely complex compared to standalone storage terminals.
The Oklahoma Corporation Commission's seismic response protocol can trigger injection restrictions on disposal wells operating above threshold seismic activity levels, which directly affects operators managing high-volume water disposal in the Anadarko Basin. AI platforms monitoring OGS seismic feeds against individual well injection data give operators advance warning before OCC action — allowing proactive injection adjustments that avoid formal restrictions. Operators in Grady, Stephens, and Kingfisher counties with large disposal volumes have reported that AI seismic correlation monitoring has reduced their OCC-directed downtime materially versus operators who responded to restrictions reactively.
A focused STACK completion optimization AI engagement — covering 50-200 existing wells as training data and delivering a completion design model for new drilling — runs $75,000-$200,000 for initial deployment. Ongoing access to a commercial completion AI platform with STACK/SCOOP-trained models runs $10,000-$25,000 per month. The ROI case in the STACK is similar to other multi-zone plays: a 10-15% EUR improvement on a $7M-$9M STACK well carries a 5-to-1 or better return on a $100,000-$150,000 AI investment, particularly when oil prices are above $60/barrel.
The OU Mewbourne School of Petroleum and Geological Engineering in Norman is the state's primary petroleum engineering research institution, and its faculty and graduate students produce research directly relevant to Anadarko Basin AI applications — including Woodford Shale fracture characterization, STACK reservoir simulation, and induced seismicity from disposal wells. Several commercial AI firms in the Oklahoma City market have advisory relationships with Mewbourne faculty, and OU's annual petroleum engineering symposium is a venue where academic and commercial AI work on Oklahoma formations intersects. OU's partnership with the Oklahoma Geological Survey gives researchers access to state production data that underpins some of the most rigorous Anadarko Basin ML model validations.
Get your practice in front of the right clients.