SoftBank Corp’s Infrinia Team, led by President & CEO Junichi Miyakawa, has developed Infrinia AI Cloud OS, a dedicated software stack for AI data centers. The stack integrates resource scheduling, hardware acceleration management and data pipelines to optimize throughput for large-scale AI training workloads. In tests on a 1,024-GPU cluster, Infrinia AI Cloud OS reduced data transfer latency by 30% and increased utilization rates from 65% to over 85%. SoftBank plans to roll out the system across its domestic data centers in the second half of fiscal 2026, with commercial licensing to external operators beginning in early 2027. Investors will watch for potential revenue contributions from licensing fees and an expanded services business as enterprise demand for turnkey AI infrastructure continues to grow.