How they use Nebius AI Cloud

Cambrium runs protein-design inference on Nebius using on-demand GPUs, Nextflow pipelines and NVIDIA NIM microservices. Its protein ML engineer describes structural design, sequence design and folding workloads. Cambrium reports fourfold faster molecule-design iterations, 40% time saved by cutting infrastructure overhead and zero time lost to resource scheduling. It also reports completing work that previously took two to three weeks within one week. The work supports sustainable biomaterials, including proteins used in consumer products.

Cambrium says it brought a new molecule to market in under 24 months. Nebius compares this with an industry norm closer to seven years. Nebius says Cambrium designs entirely new polymers from first principles. Its broader research loop pairs AI-driven molecular design with wet-lab validation to develop sustainable alternatives to petrochemical and animal-derived materials. Nebius describes Cambrium fine-tuning its own biological foundation models on proprietary lab data, verifying predictions at the bench and feeding experimental results back into the models.