Infleqtion Achieves 12-Logical-Qubit Milestone for Q4Bio Biomarker Discovery

Infleqtion Achieves 12-Logical-Qubit Milestone for Q4Bio Biomarker Discovery Infleqtion reported technical results from its Q4Bio project, involving the executing biomarker discovery algorithms using 12 logical qubits on its Sqale neutral-atom quantum computer. Supported by a $2M Phase III contract from Wellcome Leap and conducted in collaboration with academic partners at UChicago and MIT, the project integrates a hybrid quantum-classical workflow to identify high-impact feature sets from complex multimodal cancer data. By capturing higher-order correlations that are computationally inaccessible to classical systems, the platform aims to advance precision diagnostics. The milestone, showcased at NVIDIA GTC 2026, represents one of the most complex logical-qubit applications to date, achieving a 0.04% relative error compared to the optimal mathematical solution for the underlying discovery problem. The technical framework utilizes Instantaneous Quantum Polynomial (IQP) models, a class of quantum neural networks optimized for high-performance training on GPUs and efficient sampling on QPUs. Infleqtion leveraged the IQPOpt library and JAX to train these models on NVIDIA A100 nodes at the NERSC Perlmutter supercomputer, processing instances designed to scale to dozens of logical qubits. This “separate-hardware” approach—using GPUs for the heavy linear algebra of training and the Sqale QPU for exponential inference speedups—allows the system to “choose” multi-qubit correlations through a learned hypergraph structure rather than relying on simple pairwise couplings. To validate these models before physical execution, Infleqtion utilized the NVIDIA CUDA-Q platform for high-fidelity simulations. This enabled the team to establish “noiseless ceilings” and perform “noisy validation” using device-calibrated noise models, achieving simulation speedups of over 50x compared to CPU-only backends. This frictionless code path ensures that the same kernels can be stress-tested in a virtual environment on NVIDIA GH200 Grace Hopper superchips before being deployed to the neutral-atom hardware, significantly de-risking the experimentation phase for large-scale biomarker feature sets. A critical component of the Sqale platform’s fault-tolerant roadmap is the integration of NVIDIA NVQLink, which facilitates a real-time control loop between the QPU and classical accelerators. With reported round-trip latencies as low as 3.96 μs, NVQLink allows the system to transfer measurement data directly to GPU memory for immediate syndrome decoding and parameter updates. This microsecond-scale feedback is essential for the active error correction required to maintain logical qubits, allowing Infleqtion’s Pulse Processing Units (PPUs) to execute adaptive branching and corrective operations within the coherence window of the atoms. Looking ahead, Infleqtion is demonstrating the Sqale + NVQLink integration at GTC 2026 (Booth #345), highlighting its transition toward production-grade quantum-GPU supercomputing. The successful 12-logical-qubit run serves as a functional proof-of-concept for the Q4Bio program, which is entering its final phase. The company plans to scale this architecture to support over 30 logical qubits later in 2026, targeting increasingly sophisticated genomic and proteomic data analysis tasks that require the native error-suppression capabilities of fault-tolerant quantum hardware. For further technical details on the IQP model training and the NVQLink performance data, consult the official Infleqtion research blog here. March 18, 2026 Mohamed Abdel-Kareem2026-03-18T09:03:26-07:00 Leave A Comment Cancel replyComment Type in the text displayed above Δ This site uses Akismet to reduce spam. Learn how your comment data is processed.