Multiverse Computing, a San Sebastian-based startup that compresses AI models using quantum physics-derived mathematics, announced Monday a $570 million Series C round at a $1.7 billion pre-money valuation. The raise represents a five-fold valuation jump from the company’s $215 million Series B closed in June 2025 and brings total funding to approximately $800 million.

The Round

Forgepoint Capital International, BNPP Solar Impulse Venture Fund, and Bullhound Capital co-led the round, according to SiliconAngle. Additional investors include Santander Alternative Investments, Tikehau Capital, HP Inc., Orange Ventures, Scania Invest, NAventures (National Bank of Canada’s venture arm), Qatar Development Bank, Zouk Capital, and the Basque Government’s Hazten Scale-Up Fund. JP Morgan and Santander CIB advised on the transaction. The round may remain open to additional strategic investors, the company said.

CompactifAI and the Compression Thesis

Multiverse’s core product, CompactifAI, applies tensor networks, a mathematical framework from quantum physics, to shrink large language models by 80% to 95% with what the company describes as minimal accuracy loss. The practical result: models that would normally require GPU clusters can run on standard CPUs. Multiverse recently demonstrated Meta’s Llama 3.3 70B running on an Intel Xeon 6 processor after CompactifAI compressed the model from 130 gigabytes to 65 gigabytes on disk.

“The AI industry has accepted a false constraint for years, that powerful models require expensive infrastructure,” CEO Enrique Lizaso said in the announcement. “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.”

The company’s platform includes CompactifAI Router, which decides in real time whether a workload runs locally or gets sent to the cloud. Multiverse is also building a software layer for AI data centers that bundles model compression, GPU orchestration, and governance controls into a single stack, per SiliconAngle.

Traction and Customers

Revenue growth since the Series B has been significant: 10x annualized growth overall, with 96x year-over-year sales growth in Q1 2026, according to the company. Customers and partners span manufacturing, finance, energy, aerospace, cybersecurity, and defense, including Allianz, Bank of Canada, Bosch, Iberdrola, Indra, PwC, and Telefónica. Multiverse models are deployed across drones, cameras, satellites, vehicles, and telecom infrastructure.

Where the Capital Goes

The $570 million will fund expansion of Multiverse’s compressed model library, continued R&D on proprietary compression algorithms, investment in AI data center infrastructure, and geographic expansion into East Asia, Southeast Asia, the Middle East, Canada, and the United States, the company said.

The Edge Deployment Race

Multiverse’s raise lands as the economics of running AI agents at scale become a central infrastructure question. Autonomous agents running frontier-class models through cloud APIs accumulate significant per-query costs; compression that enables on-device or on-premises inference changes the unit economics of every agent interaction. The investor consortium, which includes sovereign funds (Qatar Development Bank), device manufacturers (HP Inc.), and industrial conglomerates (Scania, Telefónica), reflects a bet that efficient local inference will be the deployment default for enterprise AI rather than an edge case.