The past year has seen significant progress in developing accelerators to improve the performance of virtualized RAN (vRAN), offloading the most demanding tasks from the central processor to an optimized chip. However, this has raised challenges to the ideals of Open RAN, since accelerators from different semiconductor providers are incompatible with one another, and can limit the multivendor interoperability between different RAN vendors’ base stations. The news that Google Cloud, Intel, Qualcomm, Samsung and Arm have – among others – formed the Unified Acceleration (UXL) Foundation to deliver a standard programming model for all their accelerator designs is, therefore, important and potentially welcome for Open RAN, as well as for other demanding cloud workloads such as many AI use cases.…