Multiverse Computing Raises $570M Series C at $1.7B Valuation to Shrink AI Models with Quantum Physics

Who's backing the bet

The Series C is co-led by three institutional investors: Forgepoint Capital International, BNP Paribas Solar Impulse Venture Fund (BNPP SIVF), and Bullhound Capital . Other confirmed participants include Qatar Development Bank and HP Inc., alongside a broader consortium that includes Santander Alternative Investments, Tikehau Capital, Orange Ventures, Scania Invest, NAventures, Zouk Capital, SETT, the EIC Fund, the Basque Government's Hazten Scale-Up Fund, and Kutxa Fundazioa .

How CompactifAI works

CompactifAI works by "tensorizing" the self-attention and multi-layer perceptron layers of transformer-based models using a specific type of tensor network . This effectively truncates correlations within the model, with the degree of compression controlled by the bond dimension of the tensor network. A brief period of distributed retraining then allows the compressed model to recover most of its original accuracy .

In the January 2024 paper, the team demonstrated that CompactifAI combined with quantization can reduce the memory size of Meta's LLaMA-2 7B model by 93%, cut its parameter count by 70%, accelerate training by 50%, and speed up inference by 25% — all with an accuracy drop of just 2–3% . The EU's EuroHPC Joint Undertaking has separately confirmed that CompactifAI alone compresses LLaMA-2 7B to only 30% of its original size while recovering over 90% of accuracy after retraining .

The company has since released compressed versions of newer models, including 80% compressed versions of Llama 3.1-8B and Llama 3.3-70B, which it says achieve 60% fewer parameters, 84% greater energy efficiency, 40% faster inference, and a 50% cost reduction without sacrificing accuracy . In February 2026, Multiverse released HyperNova 60B, a 50% compressed version of OpenAI's gpt-oss-120B, on Hugging Face .

From quantum algorithms to AI compression

Multiverse Computing was originally founded as a quantum algorithms company and later pivoted to AI model compression under founder and CEO Enrique Lizaso Olmos . The company applied its core expertise in quantum physics — specifically tensor network mathematics — to build CompactifAI, recognizing that the same mathematical tools used to approximate complex quantum systems could be repurposed to efficiently compress neural network weights .

The Series B that set the stage

The Series C represents a dramatic acceleration from the company's prior round. In June 2025, Multiverse Computing raised a $215 million (€189 million) Series B, led by Bullhound Capital and with participation from HP Tech Ventures, Forgepoint Capital International, SETT, CDP Venture Capital, Santander Climate VC, Toshiba, and others . That round valued the company at over $500 million, representing a 5× jump from its Series A valuation of roughly $108 million .

The $1.7 billion pre-money valuation for the Series C thus represents another 5× step-up in roughly 13 months .

Growth trajectory and market opportunity

The company reported that its annualized revenue grew more than 10× since June 2025 . One report, cited in a Tech Funding News LinkedIn post, specifies 96× year-over-year Q1 2026 sales growth . While the exact figure varies by source, the trajectory is clearly rapid, and the company's valuation implies an upper-bound 17× valuation-to-annualized-revenue ratio according to one analyst .

A figure of $106 billion as the total addressable market for AI inference was mentioned in some Series B coverage, though it was not directly confirmed in the primary sources retrieved for this analysis .

What the funding will be used for

Multiverse Computing says the Series C capital will fund expansion of its efficient-model library, continued research and development, sovereign AI infrastructure and software, and buildout of operations in East and Southeast Asia, the Middle East, and Canada . The company is targeting enterprise deployments across manufacturing, finance, energy, aerospace, cybersecurity, defense, and healthcare .