Paris, 4 November – At the Euronaval exhibition, Multiverse Computing, the European artificial intelligence company specialising in quantum-inspired model compression, and Capgemini, a global business and technology transformation partner for organisations in the AI era, are exploring opportunities for technological cooperation within OMNISS (Open Mission Network for Interoperability, Security and Sovereignty). This initiative aims to make advanced AI easier to deploy on drones, robotic platforms and other autonomous systems operating with constrained computing resources and limited connectivity. As Europe seeks to strengthen its technological sovereignty in defence, this collaboration addresses a very practical question: how to make AI work where the mission takes place.
Bringing advanced AI on board without relying on the cloud
Modern defence operations increasingly rely on drones, ground robots, distributed sensors and software-defined platforms. These systems must process information in environments where access to cloud infrastructure may be restricted, unstable or entirely unavailable. Yet the most sophisticated AI models can require more memory, computing power and energy than onboard hardware can provide.
Reducing these requirements while preserving the accuracy, responsiveness and reliability needed for each task is therefore a major technical challenge, particularly in contested or disconnected environments, where local data processing can offer a real operational advantage.
Opportunities for cooperation within OMNISS
OMNISS aims to promote interoperability between autonomous platforms, sensors and mission applications within an open and sovereign technology ecosystem. Multiverse Computing and Capgemini have identified several areas to explore:
- Demonstrating compact AI models on an existing drone platform;
- Investigating FPGA (field-programmable gate array) acceleration for onboard AI processing;
- Deploying optimised models on resource-constrained edge devices;
- Assessing a capability for making models more resource-efficient for the OMNISS marketplace;
- Potentially organising joint demonstrations or military exercises to assess their operational relevance.
These areas remain exploratory: specific activities, timelines and deliverables will be subject to further discussions and agreements.
Multiverse Computing’s contribution
Multiverse Computing brings its expertise in AI model compression and optimisation through CompactifAI, a technology based on quantum-inspired tensor networks. These mathematical representations help reduce the size and computational requirements of neural networks while seeking to preserve their useful performance.
Applied within OMNISS, this approach could enable the evaluation of compact AI models for onboard perception, information processing and other mission-related tasks. Potential benefits include a smaller memory footprint, lower computing requirements and greater flexibility to process information locally. Performance on specific defence platforms has not yet been established as part of this cooperation.
A shared ambition for European defence innovation
“Advanced AI will only deliver operational value if it can run on the platforms that actually need it, including systems with very limited computing resources. Through OMNISS, we want to explore how model compression can help bridge the gap between increasingly capable AI and the practical constraints of autonomous defence systems, while promoting interoperability and freedom of technological choice,” says Michel Kurek, Managing Director France at Multiverse Computing.
Next steps
At a later stage, the partners could define technical priorities, identify suitable models and platforms, and assess opportunities for experimentation. No deployment, performance target, demonstration timeline or commercial commitment is being announced at this stage.
About Multiverse Computing
Multiverse Computing is a European company specialising in artificial intelligence. Drawing on its expertise in quantum-inspired algorithms and tensor networks, it develops technologies that reduce the computational requirements of AI. Its CompactifAI technology compresses and optimises AI models to make deployment easier. Headquartered in Donostia–San Sebastián, Spain, the company has operated in France since 2021 through its Paris-based French subsidiary. In France, Multiverse Computing works with major industrial and financial organisations, including Crédit Agricole CIB and Allianz France, and participates in strategic initiatives alongside companies such as Orange. Its position in the French ecosystem is further strengthened by support from leading French investors, including Quantonation, BNP Paribas Solar Impulse Venture Fund, Tikehau Capital and Orange Ventures. With a presence in the United States, Canada and several European countries, Multiverse Computing serves more than 100 clients worldwide.
Further information: multiversecomputing.com