How hard is it to outperform a classical simulator at running a quantum optimization algorithm?

Platforms for studying variational quantum-classical algorithms (VQAs) with superconducting qubit processors reaching beyond the limits of exascale emulation limits are on the horizon. In this talk, we review recent work on one pattern of VQA, the QAOA ansatz. First, we refine expected boundaries for scaling up noisy simulation with QAOA with tensor networks, limited by entanglement [1]. Still, initial states and final solutions with QAOA typically have low entanglement. We thus clarify the evolution of entanglement during the execution of the algorithm [2]. Next, we report QAOA runs on the recent Aspen-M 80Q platform at Rigetti [1].