According to an announcement by Classiq CEO & Co-Founder Nir Minerbi, Classiq and Oracle have successfully demonstrated an end-to-end quantum software engineering workflow for a quantum portfolio optimization application.
The proof of concept involved using natural language to prompt Classiq’s quantum AI agent as the starting point for generating the application, and then using Oracle’s GPU infrastructure to simulate a 36-qubit circuit.
”With only light parameter tuning, the workflow produced a feasible portfolio allocation with a strong modeled objective value and a clear path for further refinement,” said the announcement.
In total, the full simulation took approximately five hours, a 4.63% objective gap relative to classical computing. However, the main purpose of the demonstration was not the problem of quantum portfolio optimization itself. Rather, it was meant to address an existing barrier in the development of quantum applications: quantum software developers need an efficient way to build applications and the compute capacity to test them at scale.
The application aimed to solve the discrete version of a Markowitz-style portfolio optimization problem, which very broadly works to maximise rewards or minimise risk. The application includes running the hybrid quantum-classical Quantum Approximate Optimization Algorithm (QAOA), which was also demonstrated by JPMorganChase and Argonne National Laboratory in 2023.
According to the announcement, Classiq’s end-to-end workflow was as follows:
“Start from a simple AI prompt, generate a structured Jupyter notebook, synthesize a nontrivial quantum circuit, and execute the simulation on NVIDIA GPUs via Oracle Cloud Infrastructure (OCI). With only light parameter tuning, the workflow produced a feasible portfolio allocation with a strong modeled objective value and a clear path for further refinement.”
The Jupyter notebook, including model formulation, hybrid QAOA workflow, and analysis, was produced in under 15 minutes and then refined by Classiq engineers before execution.
The application modeled 12 assets, each with eight possible allocation levels, requiring three qubits per asset and resulting in a 36‑qubit circuit. This goes beyond Classiq’s actual platform capability, which is meant to support quantum simulations with a maximum of 29 qubits in its standard flow.To extend the simulation to 36-qubits, Classiq routed its circuits to an NVIDIA DGX A100 node hosted on OCI. The simulation ran across the node’s eight A100 GPUs and took about five hours.
According to the teams, the proof‑of‑concept highlights how AI‑assisted quantum programming and scalable classical simulation can accelerate development workflows as quantum hardware continues to mature. The approach is intended to support organizations exploring quantum optimization and other near‑term applications that require rapid iteration and high‑performance simulation.
For more details, read Nir Minerbi’s LinkedIn post here.
