The 2026 Nebius Physical AI Awards program is now accepting applications, targeting startups that have successfully transitioned physical AI products from research into practical deployment. This global competition, conducted in partnership with NVIDIA, seeks to support companies grappling with the unique computational challenges that emerge once AI systems are actively operating in the field, such as managing rapid data influx and validating iterative retraining cycles.
Accelerating Innovation with Significant Support
The program is structured to provide substantial resources to innovative ventures. Five distinct categories will each recognize a winner, with each recipient receiving a prize package valued at $150,000 in Nebius compute credits. These credits are designed to directly facilitate intensive training runs, enabling rapid product development and scaling.
Beyond the financial incentive, winners will benefit from joint promotional efforts by both Nebius and NVIDIA across their extensive channels, significantly boosting their market visibility. A Nebius C-level executive will also provide mentorship through working sessions, guiding teams on strategies for scaling their developed solutions. Furthermore, winners will be featured in a dedicated ‘Winners Circle’ content series published by Nebius and will receive two exclusive invitations to an executive dinner next year, connecting them with key leaders in the Physical AI ecosystem. There are no associated entry fees or sponsored categories for participation.
Understanding the Compute Credit Value
The $150,000 in Nebius compute credits translates into tangible computational power. Based on Nebius' published on-demand GPU-hour rates, this award could provide approximately 38,960 GPU-hours on H100 nodes, translating to about 203 8-GPU node-days. For intensive, multi-node training operations, $150,000 on H200s can fund around 521 hours on a 64-GPU cluster, equating to roughly three weeks of continuous work. The bundled vCPU and RAM ensure that the full value is directed towards GPU processing without additional instance charges.
Diverse Categories for Physical AI Excellence
Applicants are encouraged to submit their strongest product to the most relevant of the five competition categories. While companies with multiple products may apply to additional categories, the emphasis is on showcasing a single, impactful solution. The categories include:
- Physical AI Models: Focusing on areas like Vision-Language Models (VLMs), Vision-Language Actions (VLAs), world models, and reinforcement learning.
- Perception & Spatial Intelligence: Encompassing computer vision, Simultaneous Localization and Mapping (SLAM), sensor fusion, and scene understanding.
- Simulation & Synthetic Data: Covering physics simulation, digital twins, sim-to-real transfer, and the generation of synthetic data.
- Systems & Deployment: Addressing innovations in humanoids, industrial robots, Autonomous Mobile Robots (AMRs), and warehouse automation.
- Software, Tooling & Orchestration: Including middleware, MLOps platforms, benchmarking tools, fleet management solutions, and deployment utilities.
Eligibility requirements stipulate a clear Physical AI use case, a Minimum Viable Product (MVP) actively in use or testing, a registered legal entity, and a functional public website. Each application should focus on one specific product.
Rigorous Judging by Industry Leaders
The selection process involves an initial internal screening for quality. Subsequently, a panel chosen by Nebius and NVIDIA evaluates each submission against criteria such as depth of innovation, breadth of implementation, market potential, and real-world impact. Finalists will then advance to a jury vote to determine the category winners.
The 2026 judging committee features prominent figures from the Physical AI and venture capital sectors, including Evan Helda (Head of Physical AI, Nebius), Anu Maheshwari (Head of Venture Ecosystem, Nebius), Sumay Parikh (Robotics & Physical AI Startup Lead, NVIDIA), Amber Yang (Partner, Lightspeed), and several co-founders and CEOs of leading data and tooling infrastructure companies like Foxglove, Voxel51, Encord, and Antioch. Notably, Calvin Zhou, Co-founder and VP of Engineering at RoboForce—a 2025 winner—is also part of the panel, bringing direct experience from a past program participant.
Building on Past Success
The inaugural 2025 program, then known as the Nebius Robotics & Physical AI Awards, attracted 254 applications and ultimately recognized 55 finalists, who collectively received $1.5 million in Nebius compute credits. That year's judging panel included executives from organizations such as OpenAI, ABB, and leading venture capital firms.
Several past winners have demonstrated significant growth following their recognition. RoboForce, the 2025 Industrial Robotics winner, secured a $52 million funding round in March 2026 and has reported over 11,000 robot orders via letters of intent. Gather AI, the Vision AI and Analytics winner, closed a $40 million Series B round in February 2026. While Nebius does not directly attribute these successes to the award, these examples illustrate the caliber of companies the program aims to identify and support. Applications for the 2026 awards close on October 25, 2026, with winners announced in mid-November 2026.
This article is a rewritten summary based on publicly available reporting. For the original story, visit the source.
Source: MarkTechPost
