If RISC-V is the one place Europe could focus, what is the missing piece that keeps it from being a realistic server platform today, the tools, the silicon, or the software ecosystem?
It is mostly the software ecosystem, with silicon second and tools third. The ISA itself is not the blocker, the server world wants boring things like stable firmware, boot chains, kernel support, hypervisors, PCIe device behavior, and a lot of unglamorous validation, and RISC-V is still patchy there outside a few vendor-specific stacks. On the hardware side, you can buy RISC-V parts today, but not at the sort of performance, memory bandwidth, and platform maturity that makes them a drop-in rival for Arm in a hyperscale rack.
Compared with Arm servers, RISC-V wins on openness and long-term control, which matters for sovereignty, but Arm still wins hard on software breadth and shipping platform quality. Europe could absolutely build parts of a RISC-V server story, but the missing piece is not one magic chip, it is a whole ecosystem that behaves like a server platform instead of a dev board family.