The Stockholm-based startup achieved these results by deploying Quantum Foundation Models that are both device-aware and problem-aware. By utilizing NVIDIA CUDA-Q, cuQuantum, and cuTensorNet, the team optimized their workflows on the Leonardo Supercomputer, one of the world's most powerful systems. CEO Vish Ramakrishnan noted that the system successfully outperformed state-of-the-art AI models developed by teams at Google, Salesforce, and Amazon.

To ensure the validity of these gains, CTO Isaiah Hull oversaw a strict zero-shot evaluation protocol, intentionally excluding test series from the training set to eliminate data leakage. The company is now pursuing a dual market strategy, offering both cloud-based and on-premises deployments. By integrating NVIDIA NVQLink for low-latency connections between GPUs and quantum processors, FirstQFM aims to secure a foothold for enterprises looking to integrate quantum-enhanced forecasting into their existing operational infrastructure.