Multiverse Computing is a Spanish scale‑up that has engineered a breakthrough to slash the compute requirements of large AI models. The company announced it is closing a €500 million Series C round, positioning it as a leading contender in Europe’s AI race.

Its proprietary compression layer can be inserted into existing transformer‑based architectures, cutting inference latency and energy use by up to 70 % without compromising accuracy, according to internal benchmarks. The technology is described as a hardware‑agnostic optimisation that makes high‑performance AI more accessible to mid‑market firms.

With backing from a consortium of European venture funds and strategic corporate investors, the deal brings the total valuation to roughly €2.5 billion. €367 million has already been committed, leaving €133 million to be raised in final tranches. The firm plans to roll out the technology to its flagship product, AI‑Edge, in Q4 2024, and aims to license the core engine to major cloud vendors by 2026.

Analysts see this as a strategic shift that could reduce the carbon footprint of AI workloads across Europe. Multiverse’s CEO, Javier Ortega, said the funding will accelerate research into hardware‑agnostic optimisations. The startup has already secured pilot projects with automotive and fintech clients, demonstrating real‑world gains.