Multiverse Computing

Overview

Multiverse Computing is a Spanish deep-tech company that has changed shape as the quantum market has matured. It began in 2019 building quantum and quantum-inspired software for banks and other industrial customers, with products such as Singularity for portfolio optimization and risk analysis. Its fastest-growing business is now CompactifAI, a technology for compressing large language models so that they run faster, use less energy, and can operate on devices such as phones or factory equipment.

This is a useful example for readers: not every company in a list of quantum firms builds a quantum computer. Multiverse applies mathematical ideas from quantum physics, running on ordinary classical hardware, to problems in artificial intelligence.

History and background

The company was co-founded in 2019 by Enrique Lizaso, Román Orús, Samuel Mugel, and Alfonso Rubio-Manzanares. Orús is a physicist known for his work on tensor networks, a way of describing complicated quantum states as chains of smaller components. In 2021 Multiverse announced its first significant funding of about 12.5 million euros, and in 2022 Gartner named it a Cool Vendor for quantum software in financial services.

CompactifAI was launched in 2023. In June 2025 Multiverse closed a 215 million dollar Series B, and in 2025 it released compressed versions of Meta's Llama models as well as very small models named SuperFly and ChickenBrain. In 2026 it launched HyperNova 60B, its first open-source model. On July 27, 2026 it announced a 570 million dollar (500 million euro) Series C at a 1.7 billion dollar pre-money valuation, five times its Series B valuation, bringing total funding to about 800 million dollars. The company had about 480 employees in 2026.

How the technology works

A large language model stores its knowledge in enormous grids of numbers called weight matrices. Tensor-network methods rewrite such a matrix as a chain of much smaller linked pieces, keeping the structure that matters and discarding redundancy. The same mathematics physicists use to describe entangled quantum systems turns out to be effective at finding which parts of a neural network can be simplified.

Multiverse says CompactifAI can shrink a model by 80 to 95% with little loss of accuracy, and that the compressed models run faster and consume less energy. Independent evaluation is important here, since results depend on the model, the task, and how accuracy is measured, and the company's figures come from its own announcements. The platform also includes a router that decides in real time whether a request should run locally on a device or in the cloud.

On the quantum side, Multiverse continues to develop algorithms for finance and industry that can run on quantum hardware or on quantum-inspired classical solvers. Because current quantum computers are small and noisy, many of its commercial results come from the classical, quantum-inspired side of the business.

Products and platforms

Key milestones

Strengths and challenges

Strengths

• A technology that addresses an immediate problem: the cost and energy use of running large AI models.
• Substantial funding and a growing list of investors, including HP Inc., Orange Ventures, Santander, and Tikehau Capital.
• Deep expertise in tensor networks and quantum algorithms.
• A product line that spans compressed models, software, and quantum applications.

Challenges and open questions

• Compression results come largely from company reports, and they can depend on the model and task.
• The AI model market is crowded and moves quickly, with strong competition from larger AI labs.
• As the business shifts toward AI, the connection to quantum hardware becomes weaker.
• The valuation step-up creates high expectations for revenue growth.

Why it matters

Multiverse is one of the clearest examples of quantum ideas paying off before quantum computers are ready. Tensor networks were developed to describe quantum systems, and now they are used to make AI more efficient. That shows that research in quantum information can create commercial value through its mathematics, not just through hardware.

It is also a reminder to read funding headlines carefully. A company on a list of quantum firms may earn most of its revenue from AI, and this page therefore separates what Multiverse does with quantum methods from what it does with classical computers.

Latest news

• Multiverse Computing raises a 570 million dollar Series C at a 1.7 billion dollar pre-money valuation — July 2026
• Multiverse launches HyperNova 60B, its first open-source AI model — 2026
• Multiverse releases the SuperFly and ChickenBrain nano models — August 2025
• Multiverse closes a 215 million dollar Series B — June 2025

Frequently asked questions

This profile is written for educational purposes and summarizes public announcements, company filings, and published research. Last reviewed: September 30, 2026. It is not investment advice. Check the company's official channels for the latest figures and claims.

← Back to all quantum companies