The Helsinki-based AI infrastructure company expects deployments to exceed 100,000 GPUs through 2027 and is opening its first-availability window across three regions for the first time, on the back of NVIDIA Preferred Partner status and tie-ups with ARM, Supermicro and Compal.