There were several partnerships announced at MWC 2022, related to the biggest challenge for vRAN and Open RAN, achieving a system-on-chip for the distributed unit (DU) – which handles near-real time Layer 1 and 2 functions – that can deliver the same performance as a specialized processor. Ericsson, which was branding its ‘Ericsson Silicon’ more clearly than ever in Barcelona, argues that there will inevitably be trade-offs from using merchant architectures for these highly specialized processes. But Marvell, Qualcomm, Nvidia and Intel, plus a host of start-ups, beg to differ. Marvell and Qualcomm have both unveiled accelerators for DU functions in recent months, planning their ‘inline acceleration’ approach is superior to the ‘look-aside’ technique favored by Intel. But Intel has…