Spanish AI startup Multiverse Computing has raised $570 million (€500 million) in a Series C funding round, bringing its pre-money valuation to approximately $1.7 billion and making it one of Europe’s newest AI unicorns.
The round was led by Forgepoint Capital International, with participation from a broad group of global institutional and strategic investors.
Rather than building larger AI models, Multiverse Computing focuses on making existing large language models significantly smaller, faster, and more cost efficient. Its flagship platform, CompactifAI, helps enterprises reduce inference costs and hardware requirements while maintaining high model accuracy, addressing one of the biggest challenges facing AI deployment at scale.
Founded to Solve AI Efficiency
Multiverse Computing was founded in 2019 in San Sebastián, Spain, by Enrique Lizaso, Román Orús, Samuel Mugel, and Alfonso Rubio.
The founding team brought together expertise in quantum computing, mathematics, artificial intelligence, and enterprise technology with a shared goal of making advanced AI more practical for real-world use. Instead of competing to build larger language models, they focused on solving one of the biggest barriers to AI adoption: the high computing cost required to deploy and run these models at scale.
Today, Multiverse Computing works with enterprises across finance, manufacturing, telecommunications, energy, healthcare, and government. Its customer base includes Bosch, Iberdrola, Telefónica, Allianz, Moody’s Analytics, PwC, the Bank of Canada, and Indra, demonstrating the broad demand for AI model optimization across industries.
These organizations use the company’s technology to reduce AI infrastructure costs, improve deployment efficiency, and run advanced AI workloads with fewer computing resources, positioning Multiverse Computing as one of Europe’s leading AI infrastructure companies.
Building AI That Runs Smaller, Faster and Cheaper
Multiverse Computing’s core business focuses on helping enterprises run artificial intelligence more efficiently instead of building new foundation models.
Its flagship platform, CompactifAI, compresses large language models by up to 95% while maintaining near-original accuracy.
By reducing model size, organizations can use significantly less GPU memory, lower inference costs, reduce electricity consumption, speed up deployment, and run advanced AI applications on edge devices such as AI PCs, smartphones, industrial systems, and connected hardware.
The technology behind CompactifAI is built on tensor networks, a mathematical framework that originated in quantum physics rather than conventional neural network optimization techniques. This allows the company to remove unnecessary model parameters while preserving performance.
As enterprise AI adoption grows, Multiverse Computing is addressing one of the industry’s fastest-growing challenges. While many AI companies continue investing in training larger models, the company is focused on reducing the cost and computing resources required every time those models are used in production.
Powering the Next Phase of Enterprise AI
GPU shortages and rising inference costs continue to be two of the biggest barriers to large-scale AI adoption. While enterprises want to deploy powerful AI across more products, teams, and devices, expanding GPU infrastructure is expensive and often difficult. This is creating strong demand for technologies that make existing AI models more efficient instead of requiring more computing power.
Multiverse Computing is positioned differently from companies such as OpenAI, Anthropic, and Google. Rather than developing competing foundation models, it focuses on optimizing the models that businesses already use.
That means its technology can add value regardless of which large language model an enterprise chooses. As AI adoption moves from experimentation to production, companies that reduce infrastructure costs and improve deployment efficiency are likely to play an increasingly important role in the AI ecosystem.
