Home / Technology / AI Models Shrink, Performance Stays Big
AI Models Shrink, Performance Stays Big
25 Feb
Summary
- Spanish startup Multiverse Computing develops compressed AI models.
- Their HyperNova 60B model is half the size with similar performance.
- The company is rumored to be raising €500 million at a high valuation.
Multiverse Computing, a Spanish startup, is developing compressed AI models to make advanced technology more deployable. Their CompactifAI technology, inspired by quantum computing, has been applied to OpenAI models, resulting in the HyperNova 60B. This model is approximately half the size of its predecessor but maintains high performance with lower latency and memory use.
The company aims to offer more compressed models in 2026. Multiverse Computing has also expanded globally, with offices in the US, Canada, and Europe, and counts major enterprise clients like Iberdrola and Bosch. They are reportedly in discussions for a €500 million funding round, which could value the company at over €1.5 billion. This potential valuation highlights the growing demand for AI solutions, particularly those offering regional or sovereign capabilities.
Multiverse Computing's work supports the development of AI solutions that are both powerful and cost-effective for businesses. Their focus on compression technology addresses a key bottleneck in AI deployment, making advanced capabilities accessible to a wider range of organizations. The company's growth trajectory and potential funding round underscore its significant impact on the European AI landscape.
