Polaron is a London-based AI startup spun out of Imperial College London's Dyson School of Design Engineering, building the intelligence layer for materials science. The company's platform connects process, structure, and performance by training AI models on real microscopy images and measured material properties, enabling engineers to characterise, design, and optimise advanced materials at unprecedented speed and scale.

Polaron's core product suite includes Polaron Segmentation (automated identification and quantification of microstructural features such as pores, cracks, phases and defects), Polaron Reconstruction (3D volume generation from 2D images to unlock deeper structural insights), and Polaron Design (generative AI tools to explore process-structure-performance relationships and optimise material configurations in silico). The platform is deployed as a secure enterprise SaaS, with full traceability from image data to engineering decision.

The company serves customers in automotive, energy storage, metals, ceramics, composites, and pharmaceuticals. Its technology is already in use with global EV manufacturers accounting for over a third of world EV production, and has demonstrated greater than 10% improvements in battery electrode energy density in early deployments. Polaron is also a partner in the UK government-funded ReCAM battery recycling consortium, alongside AVL for battery cell design, and has launched academic partnerships with leading universities.

Founded in 2024 and winner of the inaugural £1 million Manchester Prize for AI in Science, Polaron raised $8 million in seed funding in February 2026, bringing total disclosed funding to approximately $8 million. The team has grown from nine to a target of around 18 and publishes in leading journals including Nature Machine Intelligence and Advanced Materials.