Multiverse Computing Raises $215M to Lower AI Costs

Multiverse Computing Secures €189 Million Series B Funding

Spanish firm Multiverse Computing announced on Thursday a substantial Series B funding round, totaling €189 million (approximately $215 million). This investment was driven by the company’s innovative technology, known as “CompactifAI.”

Introducing CompactifAI: Quantum-Inspired Compression

CompactifAI represents a novel approach to data reduction, utilizing principles from quantum computing to compress Large Language Models (LLMs). The technology reportedly achieves up to a 95% reduction in LLM size without compromising performance.

Supported Models and Future Development

Currently, Multiverse provides compressed versions of popular, open-source LLMs. These include models like Llama 4 Scout, Llama 3.3 70B, Llama 3.1 8B, and Mistral Small 3.1.

The company intends to integrate DeepSeek R1 into its offerings shortly and is actively developing support for additional open-source and reasoning models. It’s important to note that proprietary models from providers such as OpenAI are not currently supported.

Deployment and Performance Gains

These “slim” models are accessible through Amazon Web Services or can be licensed for on-premise deployment. Multiverse asserts that its models demonstrate a speed increase of 4x to 12x compared to their uncompressed counterparts.

This enhanced speed translates to a significant reduction in inference costs, estimated between 50% and 80%. As an example, Multiverse states that its Llama 4 Scout Slim version costs 10 cents per million tokens on AWS, compared to the 14 cents for the standard Llama 4 Scout.

Expanding the Reach of LLMs

Multiverse suggests that its compression technology enables the deployment of LLMs on a wider range of devices. The models are designed to be small and energy-efficient enough to operate on PCs, smartphones, automobiles, drones, and even Raspberry Pi single-board computers.

The Technical Foundation of Multiverse

The company’s technical expertise is anchored by co-founder and CTO Román Orús, a professor at the Donostia International Physics Center in San Sebastián, Spain. Professor Orús is a recognized authority in the field of tensor networks.

Tensor networks are computational methods that emulate quantum computers using conventional hardware. They are increasingly utilized for compressing deep learning models.

Leadership and Background

Enrique Lizaso Olmos, the other co-founder and CEO of Multiverse, possesses a strong mathematical background and prior experience as a college professor. He previously held a senior position at Unnim Banc, serving as its deputy CEO.

Investors in the Series B Round

The Series B funding round was spearheaded by Bullhound Capital, with participation from HP Tech Ventures, SETT, Forgepoint Capital International, CDP Venture Capital, Santander Climate VC, Toshiba, and Capital Riesgo de Euskadi — Grupo SPR.

Current Status and Future Plans

Multiverse currently holds 160 patents and serves a global clientele of 100 customers, including prominent organizations like Iberdrola, Bosch, and the Bank of Canada. This latest funding brings the company’s total raised capital to approximately $250 million.