A Spanish AI startup just secured three-quarters of a billion dollars to solve one of the industry's most pressing problems: making powerful artificial intelligence models small enough to run on your phone, your factory floor, or a satellite, without needing constant internet connection to a data center. Multiverse Computing announced a $570 million Series C funding round at a $1.7 billion pre-money valuation, marking a dramatic acceleration in the race to bring efficient AI inference to edge devices.

The funding round, co-led by Forgepoint Capital International, BNP Paribas SIVF, and Bullhound Capital, reflects a fundamental shift in how the AI industry thinks about deployment. Rather than assuming all AI workloads must flow through hyperscaler data centers, investors and device manufacturers are betting that the next generation of AI will run locally, on the devices themselves.

What Is Model Compression and Why Does It Matter?

At the heart of Multiverse's technology is CompactifAI, a compression platform that applies mathematical techniques borrowed from quantum physics to shrink large language models (LLMs) by 80 to 95 percent while preserving accuracy. Large language models are the AI systems that power chatbots and content generation tools; they typically require enormous amounts of computing power to run. CompactifAI makes them small enough to fit on consumer hardware.

The practical implications are significant. A compressed model can run faster, consume far less energy, and operate in environments where sending data to a cloud server is too expensive, too slow, or simply not allowed. This matters for industries where real-time decision-making and data privacy are critical, such as manufacturing, aerospace, defense, and healthcare.

A Czech SME is also pursuing similar compression techniques for edge-based behavioral safety analysis, developing neural network compression pipelines that reduce computational demand by 60 to 80 percent while preserving detection accuracy. Their technology has been validated for detecting anomalous behavior on low-power devices without cloud connectivity, including dangerous proximity detection, unauthorized access alerts, and fall detection.

How Is On-Device AI Being Deployed Today?

• Transportation and Infrastructure: Multiverse models are already deployed across drones, cameras, satellites, vehicles, and telecom infrastructure, with plans to embed compressed models into AI-enabled personal computers.
• Enterprise Sectors: Customers span manufacturing, finance, energy, aerospace, cybersecurity, defense, and health and life sciences, including major organizations like Allianz, Bank of Canada, Bosch, Iberdrola, and Telefónica.
• Safety Applications: Edge-based systems are being validated for public safety use cases in transport infrastructure and industrial worksites, where reliable AI inference is needed even during network outages or in areas without 5G coverage.

Since closing its Series B in June 2025, Multiverse Computing has grown annualized revenue by more than 10 times, with 96 times year-over-year sales growth in the first quarter of 2026. The company is now positioned among the fastest-growing AI infrastructure businesses in Europe.

Why Are Investors Betting So Heavily on Edge Inference?

The funding thesis behind Multiverse's Series C reflects two converging beliefs. First, device manufacturers and infrastructure operators who collectively reach hundreds of millions of end users believe that powerful AI can and should run directly on devices rather than exclusively through data centers. Second, sovereign and infrastructure investors see the path forward for AI running through facilities that accomplish twice the computational work while consuming half the energy.

"The AI industry has accepted a false constraint for years, that powerful models require expensive infrastructure. That constraint is gone. We have proven that AI can run at full performance on a smartphone, inside a sovereign data center, on a factory floor with no cloud connection," said Enrique Lizaso, co-founder and CEO of Multiverse Computing.

Enrique Lizaso, Co-founder and CEO at Multiverse Computing

The energy efficiency argument is particularly compelling. Unlike generic edge AI frameworks, compression-based approaches achieve structural reductions in the energy intensity of large language model workloads through advanced model compression techniques. This translates to lower GPU utilization, reduced energy consumption, and decreased infrastructure requirements while preserving comparable performance and accuracy.

For enterprises, the efficiency gains translate directly to cost savings. Rather than routing every workload through a hyperscaler and replacing existing infrastructure, Multiverse's platform adds an efficiency layer that enables organizations to match each workload with the right model and backend, potentially saving hundreds of millions in infrastructure costs.

How to Evaluate On-Device AI Solutions for Your Organization

• Accuracy Preservation: Assess whether the compression technology maintains detection or inference accuracy comparable to uncompressed models, particularly for safety-critical applications where accuracy loss could have real consequences.
• Hardware Compatibility: Verify that compressed models can run on the specific devices or edge hardware your organization already uses, whether that's ARM Cortex processors, mobile devices, or industrial equipment costing under $200 per unit.
• Offline Capability: Confirm that the solution maintains full detection and inference capability during network outages, in areas without 5G or LTE coverage, and under adverse environmental conditions relevant to your deployment environment.
• Integration Scope: Evaluate whether the solution covers the full deployment spectrum you need, including compressed models for cloud, on-premises, and on-device environments, plus orchestration tools that decide in real time whether workloads should run locally or be sent to the cloud.

The Series C funding will support research and development of Multiverse's model library, continued algorithmic innovation, strategic investments in sovereign AI infrastructure, and expansion into key international markets including East Asia, Southeast Asia, the Middle East, Canada, and the United States.

For organizations preparing proposals under the European Union's Horizon Europe research funding program, the Czech SME offering edge-based behavioral safety analysis is actively seeking research partnerships with consortia focused on mobility and industrial safety applications. The technology is available for demonstration and has been field-tested in railway safety and transport infrastructure contexts.

The broader implication is clear: the era of assuming all AI computation must happen in distant data centers is ending. As model compression techniques mature and edge hardware becomes more capable, the competitive advantage will belong to organizations that can deploy powerful AI inference locally, securely, and efficiently, without constant dependence on cloud connectivity or hyperscaler infrastructure.