There is a constraint at the heart of 5G monetization that no amount of engineering can overcome: the speed of light. Light travels through fibre at around 200,000 kilometres per second. That sounds fast, until you need a round-trip network response in under 10 milliseconds. Send a packet from a base station in Dublin to a regional data centre and back and you have already burned most of that budget before a single line of application code has run. For general broadband services, that is fine. For use cases that demand ultra-low latency such as computer vision and live video analytics to support remote healthcare, emergency services, and public safety, it is not.
The answer is not faster fibre. It is putting compute in the right place, and this is where 5G AI slicing could solve this physics problem and open new AI based revenues for CSPs.
A 5G AI Slice combines a dedicated, SLA-backed 5G network slice with GPU compute capacity at the network edge, allowing AI inferencing to run close to where data and experiences are generated rather than routing workloads to a distant cloud facility. This provides reduced latency, a better customer experience and opens the door for CSPs to play a bigger role in provision and delivery of AI services.
CSPs have historically sat outside the AI value chain, providing the connectivity over which AI services run while the value was captured by cloud providers, platform companies, and application developers. The 5G AI Slice changes that position. By bundling compute with connectivity CSPs can become the infrastructure layer on which enterprise AI applications depend, with the kind of proximity-based performance guarantees that hyperscale cloud cannot offer from a centralised location.
The commercial implication is that CSPs are no longer selling only connectivity. They are selling a managed platform for running AI workloads, rated by compute cycles, tokens, API calls, or outcomes rather than gigabytes. That is a different type of product, a different pricing model, and a different kind of customer conversation.
At TM Forum’s DTW event in June Qvantel, Nokia, BMC Helix, Infosys, and a group of CSP champions will show how CSPs can create, sell, monetize and assure 5G AI slices in TM Forum Moonshot Catalyst "Autonomy Accelerated". The catalyst centres on evolving autonomous network operations, and AI driven BSS for selling and assuring advanced offerings like 5G AI slice. In the service scenario focus is a football tournament where VIP ticket holders use AI glasses connected to a 5G AI Slice with edge GPU compute at close proximity. Supported by a 5G AI slice there are many options that the AI glasses could deliver to provide a VIP fan experience. These include real-time player data overlays, multi-camera feed selection, AI commentary, and personalised highlight reels for sharing on social media, that are processed at the edge within minutes of the final whistle. None of that is achievable from a centralised cloud within the latency constraints involved.
What makes the catalyst useful beyond the use case is what it validates on the operational side. The process of selling a 5G AI Slice is not straightforward. A sales representative needs to know the offer will work before committing to an SLA. The catalyst demonstrates AI-assisted intent probing, where the BSS submits a feasibility probe via the TMF921 API, Nokia's orchestration layer resolves it against network and edge compute capacity, and Infosys agents check traffic forecasts and maintenance windows. The result is a confirmed feasibility report before the offer goes to the customer. That is a meaningfully different sales process from selling a data plan.
The assurance side is equally important. Once the slice is live, a network event (such as a fibre cut) triggers automatic root cause analysis and an agent-to-agent workflow between BMC Helix's service management tooling and Nokia's autonomous network. A human approves the corrective action, the change is applied, and the intent state is restored and reported back to the BSS. The SLA holds without a manual operations team escalation.
None of this generates revenue without commercial systems that can support it. Monetizing 5G AI Slices means billing for GPU compute cycles, tokens, API calls and outcomes alongside network SLA parameters, none of which legacy BSS was built to handle. Modern, agile BSS lets commercial teams configure and launch SLA-driven offers in hours rather than months, handles revenue sharing across ecosystem partners, and closes the monetization loop between network and compute performance in real time.
The network edge is becoming a platform for AI compute, and CSPs are uniquely positioned to own it. Their distributed infrastructure, proximity to end-users, and high levels of enterprise trust create a competitive advantage that centralised cloud providers cannot easily replicate. The 5G AI Slice is the commercial product that puts that advantage to work.
CSPs that combine edge GPU compute, guaranteed 5G connectivity, and agile BSS into coherent B2B services will move well beyond the connectivity business. The ones that treat BSS modernisation as a back-office problem will find the commercial value flowing elsewhere.
The physics have not changed. But you do have a choice about where you put your compute and how you monetize it.
To find out more about this moonshot catalyst project click here.
Jukka Heiska
CMO, Qvantel