System Modeling for Advanced AI Superconducting Hardware

Superconducting electronics (SCE) are rapidly emerging as a transformative platform for high-speed, energy-efficient computing and quantum information processing. Leveraging unique physical properties—such as near-zero dissipation and dispersion at frequencies exceeding 100 GHz, picosecond-scale switching, and sub-attojoule energy consumption—SCE offers a compelling alternative to conventional CMOS technologies. These attributes enable unprecedented energy efficiency, computational density, and interconnect bandwidth, making SCE ideal for next-generation high-performance computing (HPC), artificial intelligence (AI), and quantum systems. At imec, a NbTiN-based superconducting process has achieved a decade’s worth of CMOS-equivalent device scaling in just two years, demonstrating the feasibility of high-density SCE integration. Positioned between CMOS and quantum technologies, SCE can serve as a digital backend for quantum information systems, benefiting from advanced heterogeneous integration techniques to co-package quantum, SCE, and CMOS components. Furthermore, the inherent radiation hardness of superconducting devices makes them well-suited for aerospace and space-based applications. This presentation will highlight imec’s ongoing efforts in superconducting device design, fabrication, and system-level modeling. Special emphasis will be placed on work supported by the NSF Florida Semiconductor Engine, showcasing system-level performance of superconducting hardware in advanced AI and high-performance computing.