TII Integrates Quantum Cloud with NVIDIA CUDA-Q, Scales Simulations to 500,000 Qubits

Abu Dhabi, UAE – March 17, 2026 – The Technology Innovation Institute (TII), the applied research pillar of Abu Dhabi’s Advanced Technology Research Council (ATRC), has announced the integration of its Quantum Computing Cloud Platform with the NVIDIA CUDA-Q platform for hybrid quantum-classical computing. This integration aims to broaden access to TII’s quantum hardware and simulators, utilizing the NVIDIA CUDA-Q programming interface.

Bridging Quantum and Classical Computing

The integration allows researchers and developers worldwide to submit quantum jobs directly to TII’s physical Quantum Processing Units (QPUs) and simulators, accessible at https://q-cloud.tii.ae. By combining TII’s cloud infrastructure with NVIDIA’s hybrid programming model, the collaboration seeks to lower technical barriers and facilitate high-performance experimentation in quantum computing workflows.

“Write-Once, Run-Anywhere” Functionality

The platform offers a “write-once, run-anywhere” experience through two distinct pathways:

• Native Integration: Utilizing TII’s specialized Python client to deploy quantum circuits and algorithms directly to TII’s cloud infrastructure.
• Standardized CUDA-Q Interfaces: Leveraging standard Python or C++ CUDA-Q code to target TII’s cloud-based QPUs as a seamless backend.

Expanding UAE’s Quantum Capabilities

“Our goal is to build quantum computing on our in-house QPUs both accessible and performant for the global research community,” said Dr. Leandro Aolita, Chief Researcher of TII’s Quantum Research Centre. “By enabling CUDA-Q users to submit jobs directly to our cloud platform, we are not just providing a service; we are integrating the UAE’s sovereign quantum-technology capabilities into the global fabric of hybrid high-performance computing (HPC).”

Large-Scale Quantum Annealing Simulations

In a related development, TII, in collaboration with NVIDIA, has demonstrated the simulation of adiabatic quantum annealing (QA) algorithms for problem instances involving up to 500,000 qubits. The implementation utilized tensor-network contraction based on belief propagation and custom compilation with cuTENSOR to parallelize inference algorithms on GPU-accelerated infrastructure. The largest circuits simulated contained approximately 1.5 × 10⁹ two-qubit entangling gates.

Benchmarking and Future Applications

Benchmarking against the MQLib repository indicated that the simulator achieved solution quality surpassing the evaluated heuristic solvers for 500,000-qubit Quadratic Unconstrained Binary Optimization (QUBO) problems. This emulator is accessible via an experimental cloud platform, supporting task submission through a web interface or a Python-based programmatic client. The project aims to enable academic and industrial partners to explore quantum-inspired approaches for complex, real-world optimization challenges in fields such as materials science, cryptography, and optimization.

Further details on the CUDA-Q cloud integration can be found in the official TII announcement. Additional information on the 500,000-qubit annealing simulations is available here.