Microagi Partners with Google Cloud to Train Robotic AI Models
Munich-based startup Microagi announced on July 22, 2026, a strategic partnership with Google Cloud to accelerate the development of AI models and robots capable of understanding and interacting with physical environments. Microagi will have access to Google Cloud's GPU infrastructure and the NVIDIA Blackwell platform to train its specialized models.
Access to Google Cloud's Blackwell Infrastructure
Founded in 2025 and based in Munich, Microagi will benefit from a highly optimized infrastructure including NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs (G4 VM instances) and NVIDIA GB300 NVL72 rack systems (A4X Max instances). This infrastructure will power the startup's training and inference workflows, complemented by Google Cloud's Gemini Enterprise Agent platform and AI models designed to process multimodal information such as video. This will enable the startup to scale its applications and technology to major corporate clients globally.
Custom Models for Targeted Sectors
Microagi has developed a business model centered on training models specific to each robotic platform. Its solution, Atlas, refines models based on client-specific operations using its proprietary data corpus. This partnership will allow Microagi to offer customizable software packages for corporate robotics: sectors such as hospitality or industry will thus be able to acquire robots preconfigured with Microagi models tailored to their specific operational roles.
Collaboration Integrating NVIDIA into the Ecosystem
The close collaboration between Google Cloud and NVIDIA supports Microagi in training new AI workloads and optimizing its training and inference portfolios. The engineering teams from both providers work closely with Microagi to optimize its workflows on the Google Cloud platform and accelerate the availability of new products to customers.
