Microagi, a Germany-based robotics AI startup, has announced a collaboration with Google Cloud to accelerate the development and deployment of artificial intelligence models for robotics. The partnership will see the company leverage Google Cloud's AI infrastructure alongside NVIDIA's Blackwell computing platform to train embodied AI models capable of enabling robots to understand and interact with physical environments.
The collaboration comes as global technology companies intensify investments in physical AI and robotics, an emerging segment that combines foundation models, cloud infrastructure and high-performance computing to enable autonomous machines for industrial and commercial use.
Under the agreement, Microagi will use Google Cloud's AI stack, including its AI infrastructure and enterprise AI services, together with NVIDIA Blackwell-powered cloud instances to process large-scale multimodal datasets and train robotics models. The infrastructure is expected to support both model training and inference workloads while enabling the company to scale deployments for enterprise customers globally.
Microagi develops task-specific AI models that help robots perform real-world operations across industries. Rather than building robot hardware, the company focuses on software that fine-tunes AI models for individual robotic platforms, allowing businesses to deploy robots configured for specific operational tasks. The company's technology is already used by robotics manufacturers including Unitree and UBTECH, according to the announcement.
As part of the partnership, Microagi will gain access to NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs through Google Cloud's G4 virtual machines, along with NVIDIA GB300 NVL72 rack-scale systems available through A4X Max instances. These resources are intended to accelerate training pipelines for embodied AI models that combine visual, spatial and operational data to improve robotic decision-making in physical environments.
The company said it plans to offer configurable software packages that enable enterprises to deploy robots tailored to industry-specific workflows. Businesses in sectors such as hospitality, manufacturing and industrial operations will be able to implement robotic systems pre-configured with AI models optimised for their operational requirements.
Bercan Kilic, Founder of Microagi, said the development of embodied AI requires extensive computing resources and sophisticated AI infrastructure capable of supporting large-scale model inference. He said Google Cloud's AI platform, combined with NVIDIA's hardware, provides the technical foundation needed to expand the company's enterprise robotics offerings.
Google Cloud said its AI platform is designed to support multimodal AI workloads such as robotics by combining scalable cloud infrastructure with foundation models and enterprise AI services. NVIDIA added that robotics is becoming one of AI's most compute-intensive domains because training intelligent machines requires massive physical-world datasets and accelerated computing resources.
Founded in 2025 and headquartered in Munich, Microagi develops Atlas, a hardware-agnostic robotics platform that fine-tunes AI models using customer-specific operational data. The platform is designed to help organisations deploy robots without being tied to a single AI model or hardware provider. The company also operates research facilities in Zurich, with additional offices in London and New York.
The announcement reflects a broader industry shift towards embodied AI, where technology companies are extending generative AI beyond digital assistants into autonomous machines capable of performing physical tasks. As enterprises increasingly invest in robotics for manufacturing, logistics and service operations, cloud providers are positioning AI infrastructure as a key enabler of next-generation robotic systems.
Related Articles
