Multiverse Computing targets $570M Series C to shrink AI models and slash inference costs

Spanish AI Startup Bets on Compression Over Scale

Multiverse Computing, a Spanish AI startup based in San Sebastián, has opened a Series C funding round targeting $570M at a $1.7B pre-money valuation, marking a 5x step-up from its previous round 1

1

2

2

. The round is co-led by Forgepoint Capital International, BNPP Solar Impulse Venture Fund, and Bullhound Capital, with participation from Santander Alternative Investments, Tikehau Capital, HP Inc, Orange Ventures, and others. If the round closes at its target, Multiverse Computing would reach approximately $800M in total funding across all rounds, positioning the company at the center of a distinctly European approach to AI infrastructure.

CompactifAI Technology Drives AI Model Compression Strategy

At the heart of the company's pitch sits CompactifAI, a technology that applies tensor networks borrowed from quantum physics to reduce large language models by 80 to 95% with what the company describes as minimal accuracy loss 2

2

. The approach strips redundancy from neural networks, cutting memory requirements, computational costs, and energy consumption per query. Co-founder and chief scientific officer Dr. Román Orús, a physicist whose work on tensor networks underpins the product, developed the compression framework that now allows AI models to run on edge devices including smartphones, factory floors, drones, cameras, satellites, and vehicles without requiring cloud connectivity 1

1

.

Targeting AI Inference Costs as Industry Constraint Tightens

Multiverse's strategy addresses a critical shift in enterprise AI economics: as AI inference costs overtake training expenses as the dominant burden for many buyers, the company sells slimmed-down versions of open models such as Meta's Llama, packaged to run on cheaper hardware or on-premises rather than inside hyperscaler data centers 1

1

. CEO Enrique Lizaso stated that "the AI industry has accepted a false constraint for years, that powerful models require expensive infrastructure. That constraint is gone" 2

2

. Data-center electricity demand has emerged as one of the industry's hardest constraints, and energy efficiency gains from compression offer one of the few levers available to startups without ownership of chips or power infrastructure.

Steep Growth and Enterprise Traction Signal Market Validation

Multiverse Computing reported annualised revenue growth of more than 10 times since its Series B closed in June 2025, with first-quarter 2026 sales up 96 times year-on-year 2

2

. While these figures are self-reported and have not been independently audited, the customer roster includes Iberdrola, Bosch, Telefónica, Allianz, Bank of Canada, Indra, and PwC 1

1

. The company's platform now spans cloud, on-premises hardware, and on-device deployment, coordinated by the CompactifAI Router alongside a software layer aimed at sovereign AI data centers 2

2

.

Expansion Plans Target Sovereign AI and Global Markets

The new capital will fund expansion of Multiverse's model library, continued research and development, investment in sovereign AI infrastructure, and regional buildout in East Asia, Southeast Asia, the Middle East, Canada, and the US 2

2

. Damien Henault, managing director at Forgepoint Capital International, noted that Multiverse has evolved from a downstream compression technology into what he called a complete AI foundry and operating system, describing it as the only company his firm has seen with both the technical foundation and commercial traction to become a critical platform for the industry. The round remains open to further strategic investors, advised by JP Morgan and Santander CIB, though the company has not disclosed when it expects to close or the specific terms attached to lead investors' stakes 2

2

.