When the first NR-capable roaming agreements were negotiated, they were layered onto existing IOT frameworks that had been designed for an LTE world. The assumption was that 5G roaming would initially be low volume, that LTE fallback would absorb most traffic, and that a modest NR IOT premium could be agreed bilaterally as volumes grew. Two years into meaningful 5G roaming deployments, the assumptions have aged poorly.
The LTE IOT architecture and its limits
LTE IOT structures were built around a relatively predictable relationship between data volume and cost. The cost to serve a megabyte of LTE data on a mature network is well understood, spectrum is amortised, and pricing models evolved over a decade to reflect this. IOT negotiation became a commercial exercise in which the technical cost basis was largely stable.
NR changes the cost basis in two directions simultaneously. The NR RAN is more expensive to operate per site than LTE, particularly in sub-6 GHz SA configurations where coverage expectations are higher. But NR also delivers significantly higher per-session throughput, which means a single roaming session can generate far more traffic — and far more cost — than an equivalent LTE session. IOTs set as a flat per-MB rate do not account for the throughput multiplier; those set as a per-session rate underestimate data volume.
What settlement data reveals
Operators running per-RAT cost attribution against their TAP records are finding that the effective margin on NR sessions ranges from thin to negative on bilateral agreements where the IOT was set before 2024. The precise figure varies by partner geography, traffic mix, and network configuration — but the directional signal is consistent across most data sets we have reviewed.
The partners most at risk are those with high inbound NR roaming from markets where devices are advanced (Scandinavia, South Korea, Japan) and where the bilateral IOT was last negotiated under LTE-era assumptions. These combinations generate the highest per-session cost against the oldest pricing.
Renegotiation leverage and evidence requirements
Bilateral IOT renegotiation is a commercial process, but the operators securing better outcomes are those presenting the conversation as a technical one first. Detailed per-RAT cost attribution, traffic-type breakdown, and session-length distribution data reframe the negotiation from "we want higher rates" to "here is the actual cost structure and where current pricing diverges from it".
Operators without this granularity are finding that counterparts — who may also be operating on margin pressure — are resistant to rate changes without supporting evidence. The analytics infrastructure required to generate that evidence is the same infrastructure that identifies which partner relationships are most urgent to renegotiate. The commercial and analytical problems are the same problem.
