How AI Changes Edge Computing for Mobile Operators

Mobile Experts provides a strategic view of AI in the mobile network

CAMPBELL, Calif., April 1, 2025 /PRNewswire/ -- Mobile Experts released a fresh report today, settling some contentious questions in the mobile industry about the role of GPUs in the 5G network.

For the past six months, NVIDIA has proposed and demonstrated the use of highly capable GPUs such as Grace Hopper and Grace Blackwell processors to run the computing workloads such as Radio Access Network (RAN) processing in the 5G network. Meanwhile, more than 15 different mobile operators have been investing in GPUs and data centers to offer "GPU-as-a-service" to their customers.

This new report examines the costs and the energy savings associated with this approach, in order to illustrate the strategic direction of mobile operators. Will the operators really invest in GPU infrastructure in every neighborhood, to run both 5G networks and local AI workloads in the neighborhood? Or will regional or national data centers be a better choice?

This investigation led to interviews with 12 of the top 20 operators worldwide, to understand the business issues and technical roadblocks to offering "AI-as-a-service" or "Inference-as-a-service" offerings. The conclusions are somewhat surprising: The operators are keen to become "GPU-as-a-service" providers, and are actively investigating the localized applications that will drive revenue.

The Mobile Experts report provides a breakdown of expected "Inference-as-a-service" revenue, to illustrate how much of the revenue will require low latency in both network and computing areas. The forecast tracks applications with latencies from 3 ms to more than 10 seconds, to understand where the high-dollar opportunities will be found and where the investment in GPU infrastructure is likely to take place.

This report also highlights a few key applications, including specific Artificial Intelligence (AI) and Augmented Reality (AR) applications in automotive, industrial, and consumer markets that will drive a need for real-time AI inferencing.   

To see the Table of Contents for this report, click here:

Subscribers to Edge AI 2025 will receive:

• Full access to the 23-page Edge AI report;
• Clear CAPEX and OPEX analysis of the costs and savings associated with 5G network migration to a GPU platform;
• A breakdown of GPU-as-a-service revenue, segmented by the latency requirement; and
• Access to the analysts behind the reports.

To learn more about Mobile Experts research, click here.

About Mobile Experts Inc.:

Mobile Experts techno-market analysis is used by more than 150 suppliers in the mobile/wireless industry, as well as multiple mobile operators, hyperscalers, and managed service providers.     MEXP was the only analyst firm to correctly predict the failure of femtocells and the use of small cells for capacity in 2011, and was the only firm to predict the rise of 5G FWA as early as 2015.    

Mobile Experts combines a deep-dive technical approach with hard-nosed economic analysis to predict the long-term future of the mobile industry. Recent publications focus on Industrial Private Cellular, Virtual RAN, AI impact on RAN Capacity, Small Cells, Edge Computing, and more.

Contact:
Rachel Winningham
[email protected]
+1 (408) 374-0690
www.mobile-experts.net

SOURCE Mobile Experts Inc.