Equinor is advancing its adoption of artificial intelligence, building on significant successes in seismic data interpretation and predictive maintenance, and is now targeting the development of generative and agentic AI systems.

Established AI Successes

The company's initial large-scale AI deployments have yielded substantial, tangible benefits. In seismic interpretation, AI has accelerated processing capacity by a factor of ten, enabling Equinor to re-analyze vast areas of the Norwegian Continental Shelf. In 2025 alone, the company processed 770,000 square kilometers of data, far exceeding the 70,000 square kilometers geologists could manage manually. This capability, which reduces tasks from months to hours, has directly contributed to identifying 37 new high-quality prospects and leads based on refreshed seismic imaging.

In operations, predictive maintenance programs have generated over $100 million in value by reducing unplanned downtime and optimizing intervention schedules. The strategic goal is to evolve this system from predictive to prescriptive maintenance, aiming to fully automate the workflow from early warnings to issuing work orders and managing spare-parts logistics.

Next Frontier: Generative and Agentic AI

Equinor's next phase focuses on generative AI and agentic systems. The company has already implemented internal generative AI tools, such as a "training box" that provides engineers with instant access to the full historical database of wells and drilling operations for pattern analysis and data comparison.

The current focus is on developing agentic AI across six priority domains:

• Maintenance
• Supply Chain
• Software Development
• Shared Services
• Subsurface
• Marketing & Supply/Trading

Equinor is working to define the highest-value use cases for agentic AI, with some projects expected to require years of development. However, the company adopts a cautious "fast follower" approach to operational deployment, emphasizing the need for transparency and human oversight in autonomous, self-adapting systems.

Specific Applications and Broader Vision

A key breakthrough has been in optimizing well trajectories. AI now screens hundreds of thousands of possible drilling paths to identify the optimal route, which applied to the Johan Sverdrup field Phase 3 development, yielded estimated savings exceeding $12 million.

Overall, the shift toward prescriptive maintenance and automated workflows is part of a broader strategy to create safer, data-driven offshore environments. The onshore operations centre is actively exploring the automation of alarms and routine decisions, while carefully defining where human involvement remains essential. The company prioritizes initiatives based on value, acknowledging a substantial portfolio of potential AI applications.