“Mysterious Quantum-Inspired Breakthrough at Spanish Startup Attracts €189 Million – What’s Hidden Inside CompactifAI?”

Spanish startup Multiverse Computing announced Thursday that it has secured a significant Series B funding round of €189 million (approximately $215 million), driven by its innovative AI-compression technology, “CompactifAI.”

Drawing inspiration from quantum computing techniques, CompactifAI dramatically reduces the size of large language models (LLMs) by up to 95% without sacrificing performance, according to Multiverse. The startup currently provides highly compressed versions of popular open-source AI models such as Llama 4 Scout, Llama 3.3 70B, Llama 3.1 8B, and Mistral Small 3.1. Soon, it plans to roll out a compact version of DeepSeek R1, along with additional open-source and reasoning models. However, proprietary models, such as those created by OpenAI, aren’t supported.

Multiverse’s “slim” models can be deployed on Amazon Web Services (AWS) or licensed for private, on-premises use. These models run between four and twelve times faster than their uncompressed counterparts, translating into inference cost savings of about 50%-80%. For instance, Multiverse’s Lama 4 Scout Slim model offers inference at 10 cents per million tokens on AWS, compared to 14 cents per million tokens for the standard-sized Lama 4 Scout.

Additionally, the company suggests certain CompactifAI models are compact and energy-efficient enough for deployment on everyday devices—ranging from personal computers and smartphones to vehicles, drones, and even lightweight hardware like Raspberry Pi units.

Multiverse brings substantial technical expertise to the table. Co-founder and CTO Román Orús is a professor at Spain’s Donostia International Physics Center in San Sebastián, recognized for his pioneering research on tensor networks—algorithms inspired by quantum computing principles but operable on classical computing hardware. Tensor networks have notably gained prominence as an efficient means to compress deep learning models.

CEO Enrique Lizaso Olmos, Multiverse’s other co-founder, holds multiple advanced degrees in mathematics and previously served as the deputy CEO of Unnim Bank in Spain. Lizaso Olmos’s professional background spans academia and banking.

The funding round was spearheaded by Bullhound Capital, an investment firm whose past portfolio includes Spotify, Revolut, and Discord, with participation from notable investors such as HP Tech Ventures, SETT, Forgepoint Capital International, CDP Venture Capital, Santander Climate VC, Toshiba, and Capital Riesgo de Euskadi – Grupo SPR.

Multiverse reports owning 160 patents and serving around 100 clients worldwide—including high-profile customers such as Iberdrola, Bosch, and the Bank of Canada. With this latest round, the startup’s total funding to date now reaches approximately $250 million.

More From Author

Ethereum’s Silent Surge: What Lies Beneath Its Steady Ascent Above $2,700?

Unveiling the Shadows: Who is Behind the Spyware Attacks on Europe’s Journalists?

Leave a Reply

Your email address will not be published. Required fields are marked *