Spanish artificial intelligence company Multiverse Computing has raised €189 million in funding from Bullhound Capital, Hewlett Packard Enterprise (NYSE:HPE), Forgepoint Capital and Toshiba to advance its work on compressing large language models.
The firm claims its technology can shrink the size of these models by as much as 95% while cutting computing costs by up to 80%, all without compromising performance.
Its approach draws on principles from quantum physics and machine learning but operates on classical systems rather than quantum computers.
Multiverse has already released compressed versions of widely used open-source models, including Meta’s Llama, DeepSeek and Mistral, with more expected.
The compression tool is available through Amazon Web Services’ AI marketplace.
With this investment, Multiverse becomes Spain’s largest AI startup and joins the ranks of top European firms in the sector. The company says it is focused on optimising models already in use by enterprise clients.