OpenNebula Systems and Nokia have announced a collaboration to develop a European open technology stack for sovereign AI infrastructures. The first step will be integrating Nokia’s data center network solutions with the OpenNebula management platform, aiming to coordinate compute resources, GPUs, storage, and networking from the same layer in future AI Factories.
The key points of the OpenNebula-Nokia partnership in 20 seconds
- OpenNebula will contribute its open cloud and AI infrastructure management platform, while Nokia will provide its data center networking technology.
- The initial integration aims to coordinate GPUs, compute, storage, and connectivity.
- Both companies are considering reference architectures and validations for AI Factories, Gigafactories, HPC, and Neoclouds.
- The agreement coincides with the European deployment of 19 AI Factories and plans for new Gigafactories.
The collaboration is especially timely. Europe is moving beyond the debate over technological sovereignty to a more tangible phase: it needs data centers, accelerators, storage, and networks capable of supporting large-scale AI training and inference workloads.
Currently, the European Commission is deploying 19 AI Factories across 16 Member States, most of which are expected to be operational by the end of 2026. Simultaneously, the AI Gigafactories initiative aims to further increase capacity for developing and running advanced models.
It is precisely in this infrastructure layer that OpenNebula and Nokia intend to position their proposal.
Networks Are Starting to Be as Important as GPUs
The race to build AI infrastructure is often measured by the number and types of accelerators available. However, an AI Factory requires thousands of GPUs to exchange large amounts of data with low latency, as well as access to storage and communication with the rest of the infrastructure.
That’s why networking has become much more prominent in the design of new data centers for AI.
The announced collaboration combines two different components. OpenNebula offers an open and vendor-neutral infrastructure management platform, while Nokia supplies its data center networking portfolio.
| Component | Intended Role |
|---|---|
| OpenNebula | Cloud and AI infrastructure management and orchestration |
| Nokia | Data center networking and fabric solutions |
| GPUs and CPUs | Training, inference, and HPC workloads |
| Storage | Data, models, checkpoints, and datasets |
| Networking | Node and resource communication |
| Management layer | Coordination of various components |
The initial collaboration will focus on integrating Nokia’s networking solutions into the OpenNeula management platform.
The goal is for networking not to remain a component managed separately, but to be integrated into an orchestration that encompasses compute, GPUs, storage, and networking itself.
Following this initial integration, both companies plan to work on technical validations, demonstrations, and reference architectures.
The announcement explicitly mentions several scenarios: EuroHPC AI Factories, AI Gigafactories, Neoclouds, high-performance computing (HPC) environments, and AI-oriented private clouds.
Technological Sovereignty Also Depends on Infrastructure
The other component of the agreement is European-centric.
OpenNebula Systems is headquartered in Madrid and develops an open source platform for virtualization, cloud, and AI infrastructure. Nokia, based in Finland, maintains extensive activity in telecommunications and networking.
Both companies frame their collaboration around a recurring idea in European projects: reducing technological dependence while avoiding the creation of sovereign infrastructure that leads to reliance on a single vendor.
OpenNebula advocates for a provider-neutral architecture. Its proposal for AI Factories aims to serve as a control layer that enables integration of different technologies and maintains flexibility in hardware and software choices.
This does not mean that an AI Factory built with these technologies is automatically considered “100% European.”
Accelerators used for AI remain a vital component, and Europe continues to rely heavily on foreign hardware in several layers. Digital sovereignty, therefore, does not necessarily mean that all components are manufactured within the European Union.
The announced goal is clearer: to provide an open European layer for managing AI infrastructure and networking, avoiding unnecessary dependence on a single proprietary platform.
Constantino Vázquez, head of Systems Engineering at OpenNebula Systems, explains that organizations building sovereign AI factories need to be able to integrate the technologies that best suit their needs without being tied to a single vendor.
Nokia aligns with this thinking. Matthieu Bourguignon, Vice President of Network Infrastructure Europe at Nokia, emphasizes that European data centers require infrastructure that operators can control and expand while maintaining an open foundation.
Both perspectives highlight goals expressed by the companies themselves, not yet proven results of the announced integration.
Europe Is Preparing a Much Larger AI Infrastructure
This collaboration should also be viewed within the context of European investment plans.
The European Commission recognizes a deficit in large-scale computing infrastructure compared to other regions and aims to expand capacity for training, tuning, and inference of AI models.
The AI Continental Action Plan envisions at least 19 AI Factories leveraging European supercomputing infrastructure and up to five large-scale Gigafactories. The InvestAI initiative plans to mobilize €20 billion specifically for these AI Gigafactories, as part of broader AI investment efforts.
On July 30, 2026, the European Commission announced the launch of a call for proposals for AI Gigafactories, positioning these facilities as key to expanding European computational capacity.
Scale differences are significant.
An AI Factory mainly leverages Europe’s supercomputing infrastructure to supply resources for researchers, companies, and startups. Future Gigafactories aim for higher scale, necessary to train and operate frontier models and other highly intensive workloads.
That also increases demands on the surrounding infrastructure for GPUs.
It’s not enough to install accelerators; they need to be powered, cooled, connected to high-performance storage, and linked via networks capable of moving enormous volumes of data between nodes.
This is where the Nokia and OpenNebula partnership takes on its most interesting technical dimension: treating the network as part of the infrastructure that must be coordinated alongside compute and storage.
OpenNebula has been progressively adapting its latest versions toward this market. Throughout 2026, the platform has added capabilities related to accelerated infrastructure, Kubernetes, and operation of AI Factories, while developing architectures tailored for sovereign and distributed AI environments.
The partnership with Nokia extends this strategy into networking.
Details on the technical integration, initial Nokia products supported, automation levels, and reference architectures resulting from this collaboration remain to be seen.
Currently, this is a collaborative development agreement, not an announcement of a specific AI Factory built jointly by the companies nor a contract awarded within European initiatives.
However, it underscores an increasingly important point: as Europe moves from announcing AI capacity to deploying it, technological independence will depend not only on GPU access but also on who controls the software, networks, data, and operation of the surrounding infrastructure.
Frequently Asked Questions
What have OpenNebula and Nokia announced?
They will collaborate to integrate OpenNebula’s infrastructure management platform with Nokia’s data center network solutions. The goal is to facilitate open infrastructures aimed, among other scenarios, at AI Factories and private clouds for AI.
What will each company contribute?
OpenNebula provides the cloud and AI infrastructure management and orchestration layer. Nokia supplies its data center networking solutions.
Is this agreement part of the EU’s AI Factories?
Both companies reference initiatives like EuroHPC AI Factories and AI Gigafactories as areas of potential collaboration, but the announcement does not imply the award or construction of a specific European Factory.
How many AI Factories is Europe deploying?
The European Commission indicates that 19 AI Factories are being deployed across 16 Member States, most expected to be operational by late 2026. Europe is also preparing for a new generation of larger AI Gigafactories.
