Multiverse Computing raises $215M for tech that could radically slim AI costs

On Thursday, the Spanish startup Multiverse Computing made headlines by securing a remarkable €189 million (approximately $215 million) in a Series B funding round. This impressive capital influx is driven by their innovative technology named "CompactifAI," which leverages quantum-computing principles to significantly compress large language models (LLMs). According to Multiverse, their technology can reduce the size of LLMs by an astounding 95% without sacrificing performance. The company specializes in offering compressed variants of popular open-source LLMs, focusing on smaller models such as Llama 4 Scout, Llama 3.3 70B, and Llama 3.1 8B. Additionally, a new version of DeepSeek R1 is expected to be launched soon, with more open-source and reasoning models anticipated in the near future. However, proprietary models from OpenAI and other entities are not included in their offerings. The company’s so-called "slim" models are available through Amazon Web Services or can be licensed for on-premises applications. Multiverse boasts that its models operate 4x to 12x faster than their uncompressed counterparts, resulting in a substantial reduction of 50% to 80% in inference costs. For example, the cost of using the Lama 4 Scout Slim model on AWS is 10 cents per million tokens, compared to 14 cents for the standard Lama 4 Scout. Furthermore, these models are designed to be so compact and energy-efficient that they can even run on personal computers, smartphones, vehicles, drones, and even tiny devices like the Raspberry Pi. This vision sparks creativity, as one could imagine upgrading festive Raspberry Pi setups to include LLM-powered interactive displays. Multiverse Computing boasts a robust technical foundation, co-founded by CTO Román Orús, a professor at the Donostia International Physics Center in San Sebastián, known for his pioneering research on tensor networks—tools that emulate quantum computing on classical systems. The company’s CEO, Enrique Lizaso Olmos, who holds multiple degrees in mathematics and has a rich background in banking, including a tenure as deputy CEO of Unnim Bank, complements this expertise. The Series B funding round was led by Bullhound Capital, recognized for its investments in high-profile companies like Spotify and Discord, with additional backing from HP Tech Ventures, SETT, Forgepoint Capital International, CDP Venture Capital, Santander Climate VC, Toshiba, and Capital Riesgo de Euskadi – Grupo SPR. With this latest round, Multiverse Computing has raised nearly $250 million to date and holds an impressive portfolio of 160 patents and a global customer base of 100, including notable names like Iberdrola, Bosch, and the Bank of Canada.