Nebius Physical AI Awards Offer $150K Compute Prizes

Nebius has launched its 2026 Physical AI Awards, a global competition designed to support startups that have moved beyond the research phase and possess a demonstrable physical AI product with significant market traction and a clear plan for scaling. The program, co-hosted with NVIDIA, aims to address the evolving compute needs of companies once their AI products are deployed in the field, facing challenges such as rapid fleet data ingestion, rigorous retraining validation, and the critical sim-to-real loop transition.

Applications for the awards are currently open and will close on October 25, 2026. The finalists and subsequent winners are scheduled to be announced in mid-November 2026. The competition features five distinct categories, with each category winner receiving a substantial prize of $150,000 in Nebius compute credits. These credits are intended to be directly applied to the next training runs for the winning teams' AI models. In addition to the compute credits, winners will benefit from joint promotional support across the marketing channels of both Nebius and NVIDIA, executive mentorship sessions with Nebius C-level leaders focused on scaling their ventures, and a "Winners Circle" feature published by Nebius highlighting their team and product. Furthermore, two representatives from each winning company will be invited to an exclusive executive dinner next year, providing networking opportunities with key figures in the Physical AI ecosystem.

There is no entry fee for participation in the Nebius Physical AI Awards, and the competition does not feature any sponsored categories, emphasizing a focus on merit and innovation. The value of the $150,000 compute credit prize is made concrete by Nebius's published GPU-hour rates, allowing participants to understand the exact amount of computational resources they would receive. For instance, at on-demand rates, $150,000 can purchase approximately 38,960 GPU-hours on an 8-GPU H100 node, equating to about 203 node-days. For the more advanced H200 GPU, this prize translates to roughly 33,330 GPU-hours, or 174 node-days. The B200 GPU would yield approximately 20,980 GPU-hours (109 node-days), and the B300 GPU around 19,100 GPU-hours (99 node-days). For teams engaged in multi-node training, $150,000 in H200 credits could sustain a 64-GPU cluster for approximately 521 hours, or about three weeks of continuous operation. Utilizing preemptible H200 rates, the prize extends to roughly 61,200 GPU-hours, which is particularly beneficial for checkpointable reinforcement learning tasks and data generation workloads. The Nebius compute rates include bundled vCPU and RAM, simplifying the calculation of available compute power.

The judging panel for the 2026 Nebius Physical AI Awards comprises nine distinguished experts from the Physical AI sector, ensuring a rigorous and informed evaluation process. These judges will assess the applications based on the demonstrated product, market traction, and scalability plans of the participating startups. The competition's structure, with its focus on post-deployment compute challenges, highlights Nebius's strategic positioning in supporting the operationalization and scaling of advanced AI technologies in the physical world.