The AI world has been dominated for years by increasingly large and energy-intensive language models, but a Spanish company is making waves with a radically different approach.
Based in San Sebastian, a beautiful coastal city in Spain, Multiverse Computing, founded two years ago, attracted attention at the VivaTech event. Why? Thanks to its innovative technology, CompactifAI, capable of compressing large language models (LLMs) by up to 93% without compromising their performance.
With a recent funding round of €189 million, the startup is positioning itself as a key player in making AI more accessible, greener, and suited to today’s needs.
CompactifAI: A Revolution in AI Model Compression?
At a time when tech giants, from OpenAI to xAI, are competing to develop increasingly complex models, Multiverse Computing is taking the opposite approach: optimizing what already exists.
CompactifAI, their flagship technology, drastically reduces the size of LLMs, sometimes removing up to 80% of parameters, while maintaining nearly identical performance. This compression makes models faster, less expensive to deploy, and above all less energy-intensive—a direct response to criticism about the environmental footprint of generative AI.
“Why keep creating gigantic models when we can make them smarter and more compact?” says Enrique Lizaso Olmos, CEO and co-founder of Multiverse Computing.
By carefully analyzing model structures, CompactifAI eliminates redundant parameters and adapts algorithms to users’ specific needs.
The result: lightweight models capable of running on devices as varied as smartphones, drones, or autonomous cars, even without an Internet connection.
Read also: ChatGPT in 2025: How Many Users in France and Worldwide?
A Funding Round to Conquer Europe… and Beyond!
Announced on June 12, 2025, the €189 million funding round marks a turning point for Multiverse Computing. The company now has the resources to accelerate the commercialization of CompactifAI, backed by a consortium of prestigious investors, including:
- Bullhound Capital,
- HP Ventures,
- Toshiba,
- CDP VC,
- as well as the European Commission.
This capital injection aims to democratize access to AI, particularly for European SMEs, often excluded from expensive technologies reserved for industrial giants.
With a team of 180 employees, Multiverse Computing already collaborates with major players like Meta (Llama), Mistral, or DeepSeek, adapting their models for clients as diverse as Bosch or Crédit Agricole.
The goal? To enable companies of all sizes, as well as sensitive sectors like healthcare or defense, to deploy high-performance and sovereign AI, whether on clouds like OVH or locally.
Quantum Origins, for a Hybrid Future?
While Multiverse Computing shines today in AI, its roots lie in quantum computing.
Initially specialized in developing software for quantum computers, in partnership with leaders like IBM, the company discovered that its algorithms, based on quantum tensors, excelled at compressing LLMs.
This strategic pivot toward generative AI doesn’t mean abandoning quantum: the startup continues to collaborate with players like Pasqal and anticipates a future convergence between these two technologies. “Compact AI addresses a current urgency, but quantum remains at the heart of our long-term vision,” says Lizaso Olmos.
Concrete Applications for Immediate Impact
By making AI models more compact, CompactifAI opens up unprecedented possibilities. Compressed algorithms can be integrated into devices requiring local execution, such as autonomous cars, military drones, or medical equipment. This technology also extends the lifespan of older processors, reducing costs for businesses while addressing data sovereignty issues, particularly crucial in Europe.
The energy, healthcare, and defense sectors are particularly attracted to this approach, which combines performance, confidentiality, and sustainability.
By reducing the energy consumption of models, Multiverse Computing is part of a responsible AI approach, responding to growing concerns about the environmental impact of digital technologies.
Read on this topic: Explosion in Data Center Electricity Consumption: Is AI Too Energy-Hungry?
A Vision for Inclusive and Sustainable AI
Multiverse Computing isn’t just following trends: it’s redefining AI priorities. By offering compact, high-performance, and accessible models, the Spanish startup proves that innovation doesn’t necessarily require increasingly massive solutions. As the AI race accelerates, CompactifAI offers a smart alternative, making artificial intelligence more inclusive and environmentally friendly.
With its European ambitions and its focus on the quantum future, Multiverse Computing could well become a key player in a new era for AI. One thing is certain: when it comes to artificial intelligence, size isn’t everything, and this Spanish gem is here to prove it!
Source: Lesechos
