TotalEnergies and Mistral Partner to Develop Frontier AI for Geoscience

Energy major TotalEnergies and French AI developer Mistral have announced a three-year joint research program representing an investment exceeding €100 million. The partnership will develop a new generation of frontier AI models engineered to assist TotalEnergies’ geoscience and reservoir engineering specialists in exploring, characterizing, and developing oil and gas reservoirs.

Context of the Cooperation
Upstream oil and gas exploration and reservoir management require interpreting massive, heterogeneous datasets—including seismic surveys, well logs, geological formations, and dynamic production histories. Analyzing these complex subsurface environments has traditionally involved labor-intensive, multidisciplinary workflows that can prolong exploration cycles and constrain reservoir optimization.

To address these computational and analytical challenges, cooperation was established between TotalEnergies and Mistral. The alliance combines nearly a century of proprietary geoscience data and subsurface expertise from TotalEnergies with Mistral’s foundational AI research and digital infrastructure, advancing European technological sovereignty by establishing proprietary frontier models within a European ecosystem. Patrick Pouyanné, Chairman and Chief Executive Officer of TotalEnergies, noted that exploration and reservoir engineering represent areas where AI can generate significant value by assisting specialists in evaluating complex datasets to support decision-making. Arthur Mensch, Co-founder and CEO of Mistral, highlighted the initiative's ability to adapt advanced AI models to demanding scientific, industrial, and energy challenges while safeguarding intellectual property.

Technical Solution and Responsibilities
The initiative centers on developing specialized, domain-tailored frontier AI models—leveraging agentic AI frameworks—to autonomously synthesize, analyze, and interpret large-scale subsurface data. Responsibilities and resources are divided between the partners to align energy sector expertise with foundational AI architecture:

TotalEnergies provides strategic subsurface data, domain expertise in geology, geophysics, and reservoir engineering, and contributes nearly a century of accumulated operational data.

Mistral delivers the AI foundational architecture, agentic AI algorithms, high-performance training workflows, and sovereign digital infrastructure.

At a system level, the collaboration will establish a dedicated joint scientific laboratory bringing together TotalEnergies' subsurface teams and Mistral's AI engineers. The models will process approximately 10 petaflops of subsurface and geological data, enabling the system to evaluate reservoir structures, simulate fluid dynamics, and autonomously generate multiple development scenarios. This analytical capability is designed to help technical teams evaluate high-potential exploration acreage while optimizing recovery and extending the production lifespan of existing assets.

Deployment or Implementation
The implementation strategy spans a three-year development and deployment roadmap managed directly through the joint scientific laboratory.

During the initial phase, teams will curate and train models on TotalEnergies' proprietary subsurface datasets, creating specialized tools tailored to upstream workflows. As the frontier models mature, they will be embedded into TotalEnergies’ day-to-day reservoir engineering and exploration decision pipelines, providing geoscientists with real-time scenario generation to accelerate project timelines, improve asset recovery, and strengthen digital competitiveness within a sovereign European framework.

Edited by Romila DSilva, Induportals Editor, with AI assistance.