Telcos pivot to GPU-as-a-Service model to monetize the network edge
Radian Arc's David Cook proposes that telecommunications companies can generate significant revenue by transforming their network edge Points of Presence (POPs) into GPU-as-a-Service offerings. This model supports high-performance, low-latency services like cloud gaming and AI inference, leveraging existing infrastructure to create new income streams. By monetizing GPU capacity at the edge, telcos can evolve from mere connectivity providers to AI platforms.
Key Takeaways
- Telcos can utilize distributed edge POPs to provide GPU bandwidth on demand, billed per hour, per token, or per workload.
- Immersion cooling is recommended for unstaffed POP environments due to its lower maintenance requirements and operational simplicity.
- The model favors smaller, purpose-built AI models optimized for specific applications rather than massive frontier-scale foundation models.
- Initial revenue from cloud gaming provides the foundation for expanding into video analytics, public safety, and sovereign healthcare compute.
Why It Matters
This shift addresses the core challenge of monetizing 5G infrastructure beyond consumer data plans. By positioning compute at the edge, telcos bypass the latency and backhaul costs of centralized hyperscale clouds, offering a superior environment for real-time AI agents and interactive video services. For the broader ecosystem, this creates a 'distributed cloud' that directly competes with traditional data centers for latency-sensitive B2B workloads. Watch for the adoption of token-based billing as the standard metric for telco edge services, signaling a departure from traditional bandwidth-centric business models.
Additional Context
The push toward edge-based AI monetization has accelerated globally throughout early 2026. Per w.media (June 2026), SK Telecom and NVIDIA recently announced a gigawatt-scale 'AI Cloud' initiative, utilizing the NVIDIA DSX reference architecture to manufacture tokens at the network edge for industrial and sovereign AI services. This follows a broader trend where mobile operators are looking to salvage ROI from heavy 5G investments by serving as the primary infrastructure for the 'token economy.' Regional deployments are already testing the commercial viability of these edge GPU architectures. In February 2026, Radian Arc (now part of Submer Group) signed agreements with GTPL Broadband in India and VNPT in Vietnam to deploy GPU-as-a-Service capabilities directly within carrier networks. These partnerships specifically target cloud gaming as the introductory use case to establish high-utilization rates before layering on complex enterprise AI inference tasks. Competitive pressure is also mounting from traditional networking giants. Per Light Reading (June 2026), Cisco executives recently projected that AI inference traffic could represent 25% of total network traffic by 2035, prompting collaborations like the AI Grid with AT&T and NVIDIA. These industry moves collectively suggest that the role of the telco is shifting from a 'dumb pipe' to a sovereign AI service platform, with local data control and low latency serving as the primary competitive advantages over centralized hyperscalers.
Read full article at telecoms.com
Get this in your inbox → Subscribe
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source