Multiverse Computing targets €500m raise at €2bn valuation

Multiverse Computing, a Spanish scaleup that compresses large AI models using techniques from quantum physics, is raising €500m in Series C funding at a €2bn post-money valuation. The round, still open to additional strategic investors, would be one of the largest ever for a Spanish private tech company and would push Multiverse into unicorn territory.

The funding is co-led by Forgepoint Capital International, UK-based Bullhound Capital, and BNP Paribas's climate-focused Solar Impulse Venture Fund. The European Innovation Council Fund, Spanish state-owned investor SETT, Qatar Development Bank, and NAVentures (National Bank of Canada's venture arm) are also participating. HP Inc, Orange Ventures, and Santander Alternative Investments round out the investor list.

The deal comes 14 months after Multiverse closed a $215m Series B. If completed at target, total funding will exceed €700m.

What does Multiverse Computing actually build?

Multiverse's core product is CompactifAI, a compression technology that uses tensor network methods inspired by quantum physics to shrink large language models. The company claims it can reduce LLM size by up to 95% with minimal accuracy loss. Smaller models need less compute, which translates to lower infrastructure costs and the ability to run AI on edge devices without cloud connectivity.

The pitch is straightforward: customers can run powerful AI models on a smartphone, a factory floor sensor, or a drone, rather than relying on expensive data center GPUs from US hyperscalers. Multiverse says its compressed models are already embedded in drones, cameras, satellites, vehicles, and telecom infrastructure.

“The AI industry has accepted a false constraint for years — that powerful models require expensive infrastructure. That constraint is gone.”

— Enrique Lizaso, CEO and co-founder, Multiverse Computing

Beyond compression, Multiverse is building a deployment platform that bundles model selection, GPU management, and control systems. The goal is to help enterprises pick the right model for each workload rather than defaulting to one-size-fits-all deployments.

The commercial traction behind the valuation

Multiverse claims more than 100 customers across Europe and North America, spanning manufacturing, finance, energy, aerospace, cybersecurity, defence, and life sciences. Named clients include the Bank of Canada, Bosch, Telefónica, and Allianz.

Lizaso previously told Sifted the company is on track for €200m in annual recurring revenue in 2026. Multiverse says annualized revenue has grown more than 10x year over year.

If that €200m ARR figure holds, the €2bn valuation implies a 10x revenue multiple. That's aggressive by current SaaS standards but not unusual for deep-tech AI companies with strong enterprise traction.

Where the money goes

Multiverse plans to use the capital to expand its library of compressed models, accelerate R&D, and invest in AI infrastructure. Geographic expansion is also on the agenda, with a focus on Asia, the Middle East, and North America.

What the announcement leaves out

Multiverse hasn't disclosed how much of the €500m target has actually closed. The round is described as "in its final stages but still open," which could mean anything from 80% committed to a soft-circle term sheet. The 10x revenue growth claim is also annualized, not trailing twelve months, a common tactic for making early traction look more impressive than it is.

The company's claim that compressed models maintain accuracy "with little loss" deserves scrutiny. How much is "little"? For a chatbot, 1% accuracy drop might be acceptable. For a medical diagnostic model, it might not. Enterprises evaluating CompactifAI will need to run their own benchmarks.

Source: Sifted