As AI systems move from generating answers to performing tasks, the infrastructure supporting them is also beginning to change. European AI infrastructure company Granite is building a cloud platform designed specifically for this shift, combining computing, storage, networking and AI agent management into an integrated system. The company has now secured €4 million in funding to bring that technology to customers and expand its commercial operations.
The funding round was led by Bifrost Studios, alongside private investors. Granite plans to use the investment to commercialise its platform and further integrate its software and hardware offering.
Building for the Agentic Era
Granite describes its platform as an “agentic cloud”, infrastructure designed around AI agents rather than conventional applications. The company’s proposition combines physical infrastructure with software that allows organisations to deploy and manage AI workloads.
Its current platform brings together compute, storage, GPU capacity and an agentic runtime, while allowing organisations to retain control over the infrastructure on which their AI systems operate. Granite says the platform is designed to support open source and licensed AI models, making it model agnostic.
The company argues that this approach becomes increasingly important as AI agents begin handling more complex business processes and require persistent access to data, computing resources, tools and permissions.
From Cloud to Physical Infrastructure
A key part of Granite’s model is its physical infrastructure. Rather than relying exclusively on conventional hyperscale data centres, the company develops containerised data centre units that can be deployed in specific locations.
These units combine computing, storage and networking infrastructure in a deployable package. Granite says its systems are designed for in region deployment, with liquid cooled infrastructure intended for high density AI workloads.
The company offers two approaches through its GraniteNet and GraniteCloud products. Organisations can consume sovereign computing, storage and an agentic runtime through the network, or deploy and own a GraniteCloud unit themselves.
Putting Control at the Centre
Granite’s strategy is also closely linked to data sovereignty. The company argues that organisations operating sensitive AI workloads increasingly need greater control over where their data, computing and AI systems are located.
Its architecture is designed to keep workloads within the chosen European region, while giving customers ownership or control over the infrastructure and encryption keys. Granite also highlights requirements arising from regulations such as NIS2, DORA and the EU AI Act as factors increasing demand for infrastructure with stronger governance and localisation capabilities.
The company says its platform includes compliance controls intended to support these requirements, although organisations remain responsible for their own legal and compliance assessments.
Managing AI Agents
Beyond hardware, Granite is building software to manage the AI systems running on its infrastructure.
The platform includes an agentic runtime and tools for managing AI workloads, with features covering permissions, memory, model selection and auditability. This is intended to give organisations greater visibility into what their AI agents can access and how they operate.
Granite’s infrastructure is also designed to remain model agnostic, allowing customers to use different AI models without having to rebuild the underlying platform each time.
Scaling European AI Infrastructure
The €4 million investment gives Granite capital to move from technology development toward wider commercial deployment. The company says its infrastructure is already being deployed across Europe, with its first sovereign container scheduled to ship in 2026.
Granite’s broader ambition is to create an alternative infrastructure layer for organisations that want AI capabilities without giving up control of the underlying technology. As businesses increasingly use AI agents to perform operational tasks, the company is betting that infrastructure designed specifically for those systems will become an important part of the next generation of cloud computing.