Multiverse Computing and Zylon are partnering to deliver efficient artificial intelligence within Europe’s highly regulated industries, the company says. The collaboration embeds Multiverse Computing’s compressed AI models directly into Zylon’s on-premise platform, offering a fully European solution for organizations prioritizing data sovereignty and cost control.
This pairing eliminates the need to rely on third-party cloud services, a critical benefit for sectors unable to share data externally, and delivers inference speeds four to twelve times faster without new hardware investment. “Sovereign AI is not about closing borders,” explains Enrique Lizaso, co-founder & CEO of Multiverse Computing, “It is about giving organizations a real choice over where their data lives and who controls the models running on it.”
Zylon Platform Enables On-Premise, Sovereign AI Deployments
Zylon achieves up to 84 percent greater energy efficiency when running Multiverse Computing’s AI models, a benefit particularly attractive to industries bound by strict regulations and sustainability goals. This improvement stems from the reduced computational demands of Multiverse Computing’s models, allowing for significant reductions in power consumption and operational costs, as independently verified by a Sopra Steria study showing 30 to 50 percent savings in energy and carbon dioxide emissions while retaining accuracy.
The combination addresses a critical challenge for organizations prioritizing both performance and environmental responsibility. Zylon distinguishes itself as the sole on-premise AI platform specifically engineered for regulated industries, guaranteeing data sovereignty, the absence of sub-processors, and complete cost predictability. This air-gapped architecture, combining a ready-to-use AI workspace with agentic capabilities and a platform API, provides a secure environment for sensitive data and applications.
Built-in governance, security, and observability tools scale from a single GPU to a full enterprise cluster, offering a comprehensive solution for organizations unable to utilize cloud-based AI services. Multiverse Computing contributes the model layer, completing an end-to-end European AI stack free from reliance on external cloud APIs. Multiverse Computing’s compression technology enables a four to twelve times faster inference speed on Zylon’s platform, allowing customers to support a significantly larger number of concurrent users without investing in additional hardware.
Multiverse Computing has secured over $605M in funding to further develop its CompactifAI platform, a quantum-inspired large language model compression technology aimed at the AI inference market, according to the company. Recent partnerships, including a collaboration with Axelera AI to integrate compressed models with edge devices and a strategic alliance with FIOR Group Ltd for secure AI deployments, demonstrate the company’s commitment to expanding its reach and capabilities, the company says.
“For years, regulated organizations faced a hard trade-off: run frontier AI in someone else’s cloud, or run something weaker on the hardware they own,” declared Daniel Gallego, co-founder & CEO of Zylon. “Multiverse’s compression removes that trade-off. Their models run inside Zylon’s air-gapped platform on the infrastructure a bank or a defense organization already has, at a fraction of the GPU, energy and cost footprint, and not a single byte ever leaves their walls.”
For years, regulated organizations faced a hard trade-off: run frontier AI in someone else’s cloud, or run something weaker on the hardware they own. Multiverse’s compression removes that trade-off. Their models run inside Zylon’s air-gapped platform on the infrastructure a bank or a defense organization already has, at a fraction of the GPU, energy and cost footprint, and not a single byte ever leaves their walls.
Daniel Gallego, co-founder & CEO of Zylon
This unique positioning allows customers to maintain full control over both data and infrastructure, eliminating external dependencies and addressing long-standing challenges in sectors like finance, energy, and defense.
Sovereign AI is not about closing borders. It is about giving organizations a real choice over where their data lives and who controls the models running on it. We are offering European defense, government and regulated enterprises a fully European stack, model and infrastructure, that they can adopt on their own terms, without surrendering their data, their autonomy or their energy budget to anyone.
Enrique Lizaso, co-founder & CEO of Multiverse Computing
See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
