Spain-based Multiverse Computing, which shrinks LLMs to reduce energy and compute costs, raised a $570M Series C at a $1.7B valuation
The Spanish startup says its quantum-inspired compression technology can shrink LLMs by up to 95% — Multiverse Computing, whose technology …
Context & Ripple Effects
Multiverse’s latest round follows a rapid financing progression from its €25M Series A for tensor-network LLM compression to a €189M Series B as it expanded its compression business. The new valuation puts a much larger capital base behind the same proposition: reducing the compute and energy burden of large models.
The financing also broadly validates the trajectory signaled by earlier reports of a roughly €500M raise and €100M ARR. It matters because model compression is being funded not as a research adjunct, but as a commercial layer in AI deployment economics.
First-order effects
- Multiverse gains $570M to scale its LLM-compression technology, with a $1.7B valuation giving it a stronger currency for hiring, product development and customer acquisition.
- Its existing and prospective customers have a better-capitalized supplier focused on reducing the compute and energy requirements associated with deploying LLMs.
Second-order effects
- Model-optimization vendors and infrastructure providers face added pressure to demonstrate measurable cost and performance benefits, rather than relying on access to ever-larger compute pools.
- The round directs more investor attention toward software that makes deployed models cheaper to run, complementing spending on model training and underlying hardware.
Third-order effects
- If commercial adoption sustains, AI infrastructure could allocate more value to the efficiency layer—compression, optimization and deployment tooling—alongside model builders and compute suppliers.
- The scale of the round suggests that compute-cost reduction may become a distinct venture category; whether it does will depend on independently repeatable savings across customer workloads.
The trend: AI investment is broadening from securing more compute to financing technologies that reduce the compute required for useful model deployment.