Oracle and Classiq Integrate Quantum AI Agents with OCI for 36-Qubit Portfolio Optimization HPC Simulation

Oracle and Classiq Integrate Quantum AI Agents with OCI for 36-Qubit Portfolio Optimization HPC Simulation Oracle Corporation and quantum software engineering platform developer Classiq have completed a successful high-performance computing (HPC) proof of concept connecting natural-language artificial intelligence generation with massive classical simulation clusters. The end-to-end software workflow demonstrates how an AI agent can synthesize complex, enterprise-ready quantum code from an abstract user prompt and automatically compile it into an executable circuit. To validate the synthesized program at scale, the teams routed the workload to an NVIDIA DGX A100 supercomputing node hosted on Oracle Cloud Infrastructure (OCI), successfully executing a demanding 36-qubit simulation that surpasses the bounds of standard local development environments. Technical Architecture & Specifications / Operational Implementation The technical blueprint highlights the massive computational resources required to simulate high-depth quantum circuits on classical silicon. The standard compiler flow on the Classiq platform is typically capped at a maximum of 29 qubits. Scaling this baseline to a 36-qubit state vector maps an exponential expansion of data parameters, generating approximately 68.7 billion complex amplitudes. Assuming single-precision storage parameters, holding the raw state-vector variables in stasis requires roughly 512 GiB of GPU device memory before factoring in simulator engine overheads. To accommodate this data volume, Classiq exported the circuit into an independent OpenQASM script via an external execution-adaptation layer. The file was transmitted via SSH to an OCI host directory, which triggered an isolated NVIDIA cuQuantum Appliance container running inside a Docker virtual environment. The calculation pipeline utilized mpirun script calls to distribute matrix contractions symmetrically across the node’s eight NVIDIA A100 GPUs, utilizing a periodic polling loop over SSH to feed classical parameters back to the localized Jupyter compiler interface. Algorithmic Mechanics & Financial Portfolio Allocation The target application modeled a discrete, multi-variable iteration of a standard Markowitz-style mean-variance portfolio optimization problem. The problem set was generated in under 15 minutes by passing a natural-language behavioral prompt to Classiq’s expert AI agent, which mapped a 12-asset portfolio across distinct corporate sectors under strict budget boundaries. The parameter constraints included: Asset Allocation: Each asset could receive an integer assignment ranging from 0 to 7 chunks (requiring 3 qubits per asset to encode the eight possible discrete states binary-wise). Budget Constraint: The combined chunk allocations across all 12 assets were hard-constrained to sum to a fixed value of exactly 15 chunks. This parameter boundary defines a vast computational search space encompassing 68.7 billion raw allocation combinations before applying the total budget requirement. To parse this space, the system implemented the Quantum Approximate Optimization Algorithm (QAOA), a hybrid variational method. The Classiq synthesis engine mapped out a 36-qubit circuit utilizing three distinct QAOA layers, producing a gate-level depth of 730. The entire hybrid noiseless simulation completed in approximately five hours on the OCI hardware, running 16,384 shots per sampling call across 15 outer iterations governed by a classical COBYLA optimizer. When compared against a classical benchmark compiled from 200,000 randomized feasible states, the quantum-sampled result yielded an optimization profile with only a 4.63% objective value gap, confirming the functional viability of the AI-automated software loop. Strategic Positioning & Ecosystem Integration The integration shifts the paradigm of near-term quantum software development by demonstrating an enterprise-grade transition from abstract concept to large-scale deployment. By combining Classiq’s automated optimization compiler with the high-throughput parallel compute resources of OCI, the workflow eliminates the need for developers to program manually at the restrictive physical gate level. This abstraction layer enables rapid algorithm prototyping, model debugging, and workload validation for high-value verticals like financial risk management, aerospace engineering, and logistics planning. This strategic combination allows enterprise R&D groups to scale their quantum activities on high-performance classical clusters today while maintaining full code portability as physical fault-tolerant processors continue to emerge. You can review the official corporate technical overview detailing the hybrid software integration via the Classiq insights portal here. May 27, 2026 Mohamed Abdel-Kareem2026-05-27T01:53:56-07:00 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.