Artificial intelligence (AI) could unlock significant value across upstream oil and gas, but that opportunity is concentrated in a relatively small number of applications, analysis from McKinsey & Co. found. 

In an Aug. 25 release of the report, McKinsey & Co. estimated about $65 billion in annual recurring value in the near-term using today’s technology, $125 billion as proven technologies are adopted more broadly, and $230 billion at full potential, assuming autonomous operating modes and maximum addressability across the upstream value chain. 

The estimates are net of more than $30 billion in annual AI implementation costs. Separately, AI-driven improvements in exploration success could unlock more than $35 billion annually in balance sheet value through reserves accretion.

The analysis, by Bill Ambrose, Giorgio Bresciani, and Spandan, with Priyank Singh, covers more than 550 individual AI use cases across the upstream life cycle. It uses an industry cost base of around $500–550 billion in annual capital expenditure and about $400–500 billion in annual operating expenditure. McKinsey cites Rystad Energy data for global upstream capital and operating expenditure in 2024–2025. The analysis also uses a production base of about $2 trillion in annual production margin at $50/boe. AI can generate value by increasing production, reducing operating costs, improving capital productivity, and changing the risk and balance sheet profile of assets, the analysis showed.

Top AI use cases drive most value

The opportunity is highly concentrated. McKinsey found that the top 10 AI use cases drive nearly half of the identified value, the top 20 roughly two-thirds, and the top 60 about 95%. Much of the value sits in the development and produce stages, with the leading use cases concentrated in production optimization, drilling activity, and reservoir management.

Production, drilling offer largest gains

Production assets generate high-frequency operational data with immediate physical feedback. AI can optimize rod pumps, electric submersible pumps, gas lift, waterfloods, production networks, surface facilities, flow assurance, and chemical programs. Value can be measured through incremental production, lower well downtime, fewer failures, lower intervention costs, and lower energy intensity.

In drilling, AI can improve well planning and equipment selection, reduce tripping and connection time, provide real-time and early anomaly detection, and execute closed-loop drilling automation for optimal well placement within the reservoir. These applications can shorten execution timelines through more efficient operational sequencing, reduced nonproductive time, and improved productive-time management.

AI also can accelerate subsurface interpretation, improve static and dynamic model updates, generate surrogate models for faster simulation, support recovery strategy, improve reserves estimation, and support decision-making. Machine learning, ML-plus-physics models, deep-learning surrogate models, generative AI, and agentic AI systems can execute multistep reservoir-management workflows with limited human intervention, potentially reducing time for exploration and field development from months and years to days and weeks. McKinsey cautioned that realizing this opportunity would require significant changes to reservoir planning and management activities that are currently grounded in highly siloed workflows and experience across organizations.

Some applications are already being deployed. SLB and Vår Energi are pooling resources to build collaborative well-planning workflows for projects on the Norwegian Continental Shelf, with the companies aiming to reduce cycle times from discovery to first oil from months to days. Baker Hughes and Expand Energy announced a multiyear collaboration to deploy Leucipa, an AI-powered automated field production solution, across thousands of natural gas wells and pilot an AI production assistant.

Scaling challenge extends beyond technology

The main scaling challenge, however, extends beyond technology. Incomplete data, legacy systems, limited connectivity, and models insufficiently robust for complex field environments are real barriers, but McKinsey argued that many companies may be tackling the wrong scaling problem. AI creates the most value when it changes decisions, it noted. 

The scaling challenge is therefore to build new ways of working that use AI to make faster, more consistent, and higher-quality decisions. Broad prioritization, excessive internal technology development, underestimated change management, weak value tracking, and misaligned commercial models can all impede deployment. Less than 20% of deployments track AI-generated value and share the KPI with their organizations, according to the report. 

Commercial incentives pose a particular challenge. Operators benefit from faster wells, fewer failures, lower downtime, and fewer interventions, while oilfield services and equipment (OFSE) companies can lose revenue when those efficiencies reduce billable activity. At current AI deployment levels, about $17 billion of oilfield services revenue could be exposed, rising to $60 billion at full potential. At typical industry margins of about 40%, the cash flow impact is closer to $7 billion at near-term potential and $24 billion at full potential. OFSE companies using AI internally could improve their own efficiency by about $10 billion today and up to $20 billion at full potential, according to the analysis. 

Production uplift differs because incremental barrels can create new value without necessarily displacing OFSE revenue. McKinsey's modeled production-uplift use cases total $15 billion–34 billion, making production optimization, artificial lift, and reservoir management natural candidates for performance-linked commercial arrangements.

Such arrangements are already emerging. Helmerich & Payne (H&P), for example, uses performance contracts under which operators receive wells faster and at lower cost while H&P can earn per-well bonuses for exceeding drilling speed benchmarks. Other structures include SaaS, gain-sharing models, and potentially risk-sharing autonomy models in which providers accept greater responsibility for automated decisions in exchange for higher compensation.

For OFSE companies, this creates a strategic decision over whether to remain activity providers or evolve into performance partners willing to accept greater risk in exchange for greater upside. Such partners could combine proprietary workflows, automation, equipment knowledge, data infrastructure, and commercial models designed to translate efficiency into improved profitability for both operators and service providers.