AI's Eye on Earth: How Open-Source Tech is Mapping Our Planet's Future

CAMBRIDGE, England – July 22, 2026

In an era defined by data, our ability to understand and protect our planet hinges on translating a firehose of satellite imagery into actionable intelligence. A groundbreaking collaboration between the University of Cambridge, cloud provider Vultr, and chipmaker AMD is poised to do just that. Their project, TESSERA, is not just another environmental monitoring tool; it's a revolutionary AI foundation model designed to create the most detailed, open-access digital representation of Earth's land surface ever conceived.

By processing petabytes of European Space Agency satellite data, TESSERA is generating a dynamic, 10-meter resolution map of the globe, tracking changes from 2017 to 2025. This initiative represents a pivotal moment where cutting-edge AI infrastructure meets a pressing planetary need, democratizing access to environmental data on an unprecedented scale.

A New Foundation for Planetary Science

At the heart of the project is TESSERA, which stands for Temporal Embeddings of Surface Spectra for Earth Representation and Analysis. Unlike traditional AI models that require extensive, task-specific training data, TESSERA is a "foundation model." Pre-trained on vast, diverse datasets from Sentinel-1 (radar) and Sentinel-2 (optical) satellites, it learns a general, nuanced understanding of Earth's surface. This allows it to compress a full year of observations for every 10-meter square of land into a compact, 128-dimensional "embedding"—a smart, compressed AI representation.

This approach solves one of the most persistent problems in satellite remote sensing: cloud cover. By analyzing temporal patterns, the model effectively sees through the clouds, filling in data gaps that have historically plagued environmental analysis. The result, as described in a peer-reviewed paper presented at the 2026 CVPR conference, is the "most accurate pixelwise foundation model ever created from Sentinel satellite data."

Powering this monumental computational task is a strategic partnership providing the necessary muscle. Vultr, the world’s largest privately-held cloud company, is supplying the infrastructure, specifically configurations featuring AMD's latest Instinct™ MI325X accelerators. These powerful GPUs, managed by the AMD ROCm™ open software stack, are essential for processing the petabytes of raw satellite data and training the complex AI model.

"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. She noted that the infrastructure allows Cambridge researchers to "efficiently process petabytes of satellite data and generate planetary-scale embeddings that help turn environmental observation into actionable insight.” This isn't just about providing hardware; it's about enabling a new paradigm of research.

From Pixels to Policy: The Open-Source Revolution

Perhaps the most transformative aspect of TESSERA is its radical commitment to openness. In a field where proprietary models and paywalled data are common, the Cambridge team is taking a different path. The entire dataset of global embeddings is being released under a permissive CC-BY license, making it free for anyone to use, adapt, and build upon, even for commercial purposes.

"Our goal is to democratize access to planetary-scale environmental monitoring," emphasized Professor Anil Madhavapeddy of Cambridge's Department of Computer Science & Technology. "By making TESSERA's embeddings freely available... we're ensuring that any researcher, government, or organization worldwide can deploy this technology for their specific conservation needs."

This open approach stands in stark contrast to other major initiatives, such as Google DeepMind's closed AlphaEarth Foundations model. While both aim to create powerful environmental models, TESSERA's transparency is designed to foster a global community of innovation. The team is also publishing the complete, reproducible training pipeline, allowing others to verify their results or even fine-tune the model for specific regional challenges on the same Vultr infrastructure. This focus on reproducibility is critical for building scientific trust and enabling equitable access for researchers in developing nations who may lack the resources to build such models from scratch.

The embeddings themselves are remarkably lightweight, capable of being analyzed on a standard laptop. This accessibility removes a significant barrier, empowering a new generation of non-AI experts—from local conservation groups to city planners—to leverage state-of-the-art environmental intelligence without needing a supercomputer.

Real-World Impact Across Continents

The true measure of TESSERA lies in its application. The project is already being deployed across three critical verticals, demonstrating its versatility and immediate real-world value.

In agriculture, the 10-meter resolution data enables precision farming on a new scale. It can monitor crop health, predict yields, and manage risks, with a particular focus on supporting smallholder farmers in developing regions who are most vulnerable to climate change.

For biodiversity and conservation, TESSERA is a powerful new weapon. It can track subtle habitat changes, monitor deforestation, and assess the health of critical ecosystems. In the UK, for example, a project is already using the data to evaluate the effectiveness of government nature protection schemes in Cumbria by monitoring changes in heather, pine forests, and vital peatlands. Its high "label efficiency"—achieving strong performance with minimal training data—makes it ideal for monitoring diverse and resource-constrained environments, from the Amazon rainforest to the hedgerow networks vital for British wildlife.

The third vertical is renewable energy. By accurately detecting and mapping solar and wind farm installations, TESSERA provides crucial data for planners and policymakers working to accelerate the global energy transition. It offers a clear, data-driven view of infrastructure deployment, helping to track progress and identify new opportunities.

The 'Why Behind the Buy': A Strategic Bet on Purpose-Driven AI

For Vultr and AMD, supporting TESSERA is more than corporate philanthropy; it is a shrewd strategic investment. By aligning with a high-profile, high-impact "AI for good" project, both companies are showcasing the power and scalability of their technologies beyond the traditional enterprise market.

"This collaboration positions Vultr at the intersection of cutting-edge AI infrastructure and planetary conservation," stated Kevin Cochrane, Vultr's Chief Marketing Officer. He highlighted that by providing the computational resources and embracing open-source technology, Vultr is enabling research with "immediate real-world impact." It serves as a powerful, public-facing case study of Vultr's mission to make high-performance cloud infrastructure accessible for world-changing innovation.

For AMD, it's a prime opportunity to demonstrate the prowess of its Instinct accelerators and ROCm open software stack in a demanding, real-world scientific workload. As the AI hardware market becomes increasingly competitive, proving capability in planetary-scale data processing gives them a significant differentiator. This partnership exemplifies how strategic innovation in the private sector can accelerate public-interest research, creating a virtuous cycle where technological advancement directly serves the global good. The collaboration, which began in late 2025, has already made its first global embeddings available, marking a new chapter in our ability to watch over our world.