A new report analyzes the role of GPUs and the potential of the “GPU-as-a-service” model in transforming 5G and real-time inference
Artificial intelligence (AI) is no longer a future technology for mobile operators: it is becoming the engine that redefines edge computing in 5G networks. According to a newly published report by Mobile Experts Inc., the adoption of graphics processing units (GPUs) in network infrastructure is gaining traction among major operators worldwide, who are beginning to strongly explore business models such as “GPU-as-a-service” and “Inference-as-a-service.”
Betting on GPUs at the Heart of 5G
In the past six months, tech giants like NVIDIA have promoted concrete proposals to use their Grace Hopper processors and Grace Blackwell for critical network tasks, such as radio access network (RAN) processing in 5G. More than 15 operators have already begun investing in local data centers with GPU infrastructure, in a strategic move aimed at providing AI services and processing at the edge of the network.
The report is based on interviews with 12 of the top 20 global mobile operators. Its findings reveal that operators are seriously interested in becoming GPU-as-a-service providers, with a focus on monetizing applications that require low-latency AI inference close to the user.
Distributed Edge or Regional Data Centers?
The study also addresses a key question: Will operators install GPU infrastructure in every neighborhood to run 5G networks and localized AI workloads, or will they opt for regional or national data centers? There is no single answer, but the report offers a detailed comparison of operational (OPEX) and capital (CAPEX) costs, as well as expected benefits in terms of energy savings and service profitability.
To make decisions, operators must consider the type of application they want to support. Latency is a determining factor. The report classifies business opportunities by latency requirements, from critical tasks with 3 ms to those with more than 10 seconds, to identify where it is worth deploying edge infrastructure.
Key AI and AR-Driven Applications
Among the areas with the greatest potential, the report highlights specific use cases in the automotive, industrial, and consumer sectors that require real-time AI inference:
- Connected vehicles with autonomous driving or advanced assistance capabilities.
- Smart factories, where sensors and cameras require immediate analysis.
- Augmented reality (AR) for entertainment or remote assistance, with ultralow latency requirements.
These applications not only require a robust 5G connection, but also local processing to ensure immediate response times and prevent congestion on the backbone network.
A Profitable Future for Operators
The report predicts that the “Inference-as-a-service” model will be an emerging revenue source for mobile operators, especially if they manage to position themselves as AI service providers at the edge. The combination of 5G connectivity, edge computing, and GPUs represents a unique strategic opportunity to capture value in a new digital era where data and its instant processing will be key.
Accessing the Report
Subscribers to Mobile Experts’ Edge AI 2025 report will receive full access to the 23-page study, detailed financial analysis, forecasts by application type and latency, and direct access to the analysts who produced the report.
With AI redefining RAN capabilities, edge computing, and operators’ revenue structures, the report stands as an essential roadmap for technology leaders in the mobile sector.