Vultr and AMD support University of Cambridge AI project for environmental monitoring
Vultr and AMD are supporting the University of Cambridge’s TESSERA project, an artificial intelligence (AI) foundation model designed to monitor environmental changes using satellite imagery at a global scale.
As part of the collaboration, Vultr is providing cloud infrastructure powered by AMD Instinct MI325X GPUs to support the development and deployment of the model. According to the companies, the infrastructure will be used to generate AI embeddings covering the Earth’s land surface between 2017 and 2025.
Developed by the University of Cambridge’s Energy & Environment Group, TESSERA processes imagery from the European Space Agency’s Sentinel satellites to create AI representations of the Earth’s surface at a 10-metre resolution. The project is intended to support applications including agricultural monitoring, biodiversity conservation and renewable energy infrastructure mapping.
The collaboration will use Vultr’s GPU infrastructure for model training, data processing and long-term storage of the generated datasets. The University of Cambridge also plans to release the embeddings and the model’s training pipeline under an open licence to enable researchers and organisations to build region-specific applications.
“By providing the computational resources required for TESSERA and embracing open source technology, Vultr is enabling research that has immediate real-world impact,” said Kevin Cochrane, Chief Marketing Officer at Vultr.
“TESSERA shows what becomes possible when world-class research teams can scale AI with the right compute and an open software foundation,” said Stephanie Dismore, Senior Vice President, EMEA at AMD.
Professor Anil Madhavapeddy, Professor of Planetary Computing at the University of Cambridge, said the project aims to make planetary-scale environmental monitoring more accessible by making both the AI embeddings and training pipeline publicly available.
The collaboration began in November 2025, and the first global AI embeddings have been available for public access since early 2026.