The energy company TotalEnergies, together with Dell Technologies and NVIDIA, has contracted the design and installation of Pangea 5, a next-generation supercomputer to be hosted at the Jean Féger Scientific and Technical Center (CSTJF) in Pau, France. This program represents an investment of over 100 million euros and will multiply the company’s available computing power by six times. The project is intended to support a broad set of digital and scientific activities, from advanced subsurface analysis to ambitious AI initiatives, and is scheduled for first commissioning in 2027.
The new system is designed to respond to two converging pressures: growing data volumes from seismic surveys and the rising computational demands of industrial-scale artificial intelligence. Pangea 5 will speed processing workflows and shorten experimentation cycles, enabling teams to iterate faster on models used to image the subsurface and to develop integrated power and energy models. By combining raw throughput with improved energy performance, the machine aims to provide both operational gains and lower carbon intensity per calculation.
Table of Contents:
Architecture and energy performance
Pangea 5 relies on specialized, highly parallel hardware to handle large scientific workloads. The design uses modern GPUs and optimized interconnects to achieve high concurrency, which is central to the system’s performance. At equal computational output, TotalEnergies expects roughly a 40% reduction in overall energy consumption compared with its prior platform, while the cooling plant’s electricity draw will be reduced by a factor of five. These improvements are driven by a combination of denser compute units and more efficient thermal management.
Processors and system topology
The chosen configuration integrates high-throughput GPUs, contemporary CPUs and a low-latency network fabric to accelerate both simulation-oriented and data-parallel tasks. This mix targets workloads such as advanced seismic imaging, which require sustained floating-point throughput, and machine learning training, which benefits from parallel matrix operations. By adopting this architecture, TotalEnergies positions Pangea 5 to handle the most demanding industrial and research scenarios while maintaining energy-conscious operation.
Heat recovery and sustainability
Energy efficiency is complemented by a practical reuse strategy: the residual heat produced by the supercomputer will be recovered and fed into the heating system for buildings at the CSTJF, which host more than 2,500 employees. This residual heat recovery approach reduces waste and turns computing by-product into campus energy, aligning the deployment with corporate sustainability goals and delivering a tangible local benefit.
Operational and strategic impacts
Operationally, a sixfold expansion in compute capacity shortens the turnaround time for complex seismic processing and modeling tasks, enabling faster decisions in exploration and field development. The enhanced compute capability also expands R&D horizons for AI, allowing teams to train larger models and explore integrated power systems with finer granularity. Company leaders have framed Pangea 5 as a tool to maintain leadership in high-performance computing and to support low-cost, lower-emission hydrocarbon projects while scaling renewables and electrification efforts.
Regional role and organizational context
The Jean Féger center remains a cornerstone of TotalEnergies’ technical know-how and will host one of the firm’s most significant supercomputing investments to date. Executives have noted that the facility will continue to anchor research activity in the region while the company also develops capabilities elsewhere, including expanded digital centers internationally. Local employment and the campus experience for staff are expected to benefit from the upgrade, and management has indicated that the new installation will place Pangea 5 among the largest systems globally, reflecting a step change compared with prior refresh cycles.
Partners emphasize that the collaboration with Dell Technologies and NVIDIA is intended to accelerate discovery and operational efficiency across energy activities. With the planned commissioning in 2027, TotalEnergies will begin to leverage the new platform to reduce time-to-insight for exploration, drive more ambitious AI experiments and capture efficiency gains through both hardware improvements and campus-level heat reuse. The deployment represents a mix of technological ambition and practical sustainability measures that together aim to support the company’s evolving energy strategy.
