Multiverse Computing Secures $215M to Cut AI Costs with CompactifAI Compression Tech

Multiverse Computing, a Spanish tech startup, has raised a massive $215 million (€189 million) in Series B funding to revolutionize AI with its cutting-edge compression technology, CompactifAI. The technology, inspired by quantum computing, can shrink large language models (LLMs) by up to 95% without sacrificing performance.

The company currently compresses open-source models such as Llama 4 Scout, Llama 3.3 70B, Llama 3.1 8B, and Mistral Small 3.1. A version of DeepSeek R1 is also on the way. Proprietary models like OpenAI’s are not supported.

Available via Amazon Web Services and on-premise licensing, these “slim” models deliver 4x–12x faster performance and 50%–80% lower inference costs. For example, the Slim version of Llama 4 Scout costs just 10 cents per million tokens, compared to 14 cents for the standard model.

CompactifAI models are also incredibly efficient—small enough to run on PCs, phones, drones, and even Raspberry Pi devices. It’s based on tensor networks, a quantum-inspired tool that works on classical hardware.

Bullhound Capital, HP Tech Ventures, Forgepoint Capital, and Toshiba are among the investors supporting Multiverse. With this round, the company’s total funding has reached $250M, boasting 160 patents and 100+ global clients, including Bosch, Iberdrola, and the Bank of Canada.

This funding positions Multiverse Computing as a major player in the race to make AI faster, cheaper, and more accessible.