Multiverse Computing funding round targets $570 million at $1.7 billion valuation

Multiverse Computing funding is being targeted at up to $570 million, with the Spanish scaleup seeking a $1.7 billion valuation as enterprises look for ways to cut AI compute costs. The company said the round remains open, so the final amount has not yet been disclosed.

Forgepoint Capital International, BNPP SIVF and Bullhound Capital are co-leading the financing, according to Multiverse Computing. The company said other committed backers include Santander Alternative Investments, Tikehau Capital, Orange Ventures and Scania Invest, among others.

If the round closes at the targeted size, Multiverse Computing said its total funding would be around $800 million. The proposed $1.7 billion valuation would be about five times higher than the valuation attached to its Series B, when the company raised $215 million, according to the company. The company did not say whether the stated valuation is pre-money or post-money, a distinction that changes dilution and ownership math in venture rounds and is explained in this guide to pre-money versus post-money valuation.

What does Multiverse Computing do?

Multiverse Computing sells technology at the overlap of AI and quantum physics that it says can shrink large language models so they are cheaper to run. Its central product, CompactifAI, is pitched as a way to reduce model size enough for deployment on edge devices such as phones, cameras, drones, satellites, vehicles and telecom infrastructure.

The company claims CompactifAI can cut LLM size by 80% to 95% with immaterial loss of accuracy. That is the key technical and commercial claim behind the round, because inference cost and energy use remain constraints for enterprises trying to deploy AI beyond centralized data centers. Multiverse Computing did not disclose revenue, gross margins, customer concentration or independent benchmarks supporting the compression figures.

The company’s thesis is that more AI workloads will shift toward devices that process data locally, rather than sending every request to large data centers. Edge AI can reduce latency and bandwidth costs, and it can be relevant where data transfer is expensive, regulated or technically difficult. The trade-off is that smaller devices have far less compute and memory than cloud infrastructure, which makes model compression a practical requirement.

Multiverse Computing said its technology is already used across several device and infrastructure categories, including drones, cameras, satellites, vehicles and telecom systems. It also said its customers and partners span manufacturing, finance, energy, aerospace, cybersecurity, defense, health and life sciences, with named organizations including Allianz, Bank of Canada, Bosch, Iberdrola, Indra, PwC and Telefónica.

Damien Henault, managing director and partner at Forgepoint Capital International, said Multiverse Computing has moved from LLM compression into what he described as a broader AI foundry and operating system. He said Forgepoint views the company as having both technical foundations and commercial traction. Those are investor claims from a lead backer in an open financing, and the company has not disclosed the operating metrics that would let outsiders assess the scale of that traction.

The size of the targeted round shows continued investor appetite for infrastructure plays that promise to make AI deployment less expensive, even as many enterprise AI vendors are still proving durability and ROI. For Multiverse Computing, the open question is whether model compression based on its quantum physics approach can hold up across production workloads, not just whether the market wants cheaper inference.

This story draws on original reporting from Tech.eu.