At last year’s Mobile World Congress, one of the major themes was the ‘accelerator wars’, as chip providers raced to launch products that would address the biggest challenge for running a virtualized RAN (vRAN) on general-purpose processors (GPP). This is a key goal of the Open RAN community, but GPP has struggled to match the performance of dedicated silicon for processor-intensive tasks in Layer 1 of the RAN, such as beamforming and forward error correction. With AI-hype in full bloom, industry leader Intel is now on the defensive in vRAN against GPU-centric products from AI specialist Nvidia, among others. A year on from MWC23, and that challenge of running a vRAN on GPPs has been largely addressed, even if many…