Spanish artificial intelligence firm Multiverse Computing has secured €189 million ($217 million) from a group of global investors that includes Bullhound Capital, HP, Forgepoint Capital, and Toshiba. The company specialises in compressing large language models (LLMs), claiming that its proprietary approach may reduce model sizes by up to 95% while maintaining accuracy and performance.

According to the corporation, this innovation has the potential to reduce AI model operation expenses by up to 80%. The startup uses principles from quantum physics and machine learning, but it does not require a quantum computer to function. This efficiency establishes Multiverse as a key player in the emerging AI field.

Multiverse has already used its compression methods on models such as Meta’s LLaMA, DeepSeek, and Mistral. CEO Enrique Lizaso Olmos stated that their primary focus is on compressing open-source models for enterprise use.

This approach is consistent with broader AI tendencies towards decreasing computing burden, energy usage, and environmental effects. Training huge models requires a lot of resources; however, compression techniques provide long-term solutions. Multiverse’s huge funding mirrors a broader increase in investment in European AI businesses, with Spain joining France, Germany, and the United Kingdom in encouraging high-impact AI innovation.