Operator strategies for implementing GPU-as-a-service platforms and infrastructure

Some operators are hoping to launch GPU-as-a-service (GPUaaS) offerings to unlock a new revenue stream. However, these operators will need to introduce new infrastructure and platforms to support their offerings because enabling GPUaaS introduces new technical requirements compared with using on-premises GPUs for internal AI use cases.

Information included in this report

• An analysis of the requirements of the technology stack needed to support GPUaaS offerings
• Suggestions for how operators can implement the technology stack required to support GPUaaS offerings
• Insights into the viability of using RAN-based GPUs to underpin GPUaaS offerings