Multiverse Computing raises $570 million: Revolutionizing AI model compression
Spanish startup Multiverse Computing has made a significant breakthrough in the artificial intelligence market, closing a Series C funding round of up to $570 million. The company's valuation has reached $1.7 billion, confirming high investor interest in solutions that optimize AI computing costs.
The startup's core technology is based on tensor networks — a mathematical framework that allows compressing machine learning models by up to 95% of their original size. The key advantage: with such a radical reduction in data volume, the accuracy and quality of the models' performance remain virtually unaffected. This paves the way for the adoption of "efficient AI" — systems that require fewer computing resources, energy, and infrastructure costs.
The raised funds will be allocated to three key areas. First, the development of the CompactifAI platform, which is already positioned as a standard for neural network compression. Second, the creation of sovereign AI infrastructure — this is particularly relevant for nations striving for technological independence. Third, global expansion: the startup is targeting the markets of the US, Asia, and the Middle East, where demand for energy-efficient AI solutions is growing exponentially.
In my opinion, Multiverse Computing demonstrates a mature approach to the problem of the "energy hunger" of modern AI systems. While giants like OpenAI and Google spend billions on scaling data centers, the Spanish startup offers an elegant mathematical solution. If tensor compression truly maintains model quality at 95%+, this could become one of the most significant breakthroughs of 2025 in the field of efficient computing. However, the main question remains whether the company can scale the technology to industrial volumes without losing performance.