5G standalone deployments are turning network slicing from an architecture discussion into a wholesale product-design question. For MNOs, the viable offer is not an abstract low-latency tier but a governed set of PCF policy controls, admission rules, user-plane locations and charging records that an MVNO can sell without creating unbounded exposure on the host network. The commercial test is whether each promised attribute can be measured, limited and settled.
Wholesale slicing is moving from concept to product definition
Open API exposure and 5G SA investment are making slice-based wholesale more plausible. They do not yet make a national slice a uniform product. Coverage, core topology and operational readiness vary by market and region, while many wholesale organisations still rely on provisioning and assurance processes designed for undifferentiated mobile data. The first offers should therefore comprise a limited catalogue of service constructs, each tied to defined resources, eligibility rules and an identifiable customer proposition.
Initial demand is more likely to come from enterprise-focused MVNOs, fixed-mobile providers, transport and logistics propositions, and platform operators with recognisable traffic classes than from mass-market consumer bases. These customers can specify where traffic originates, which applications matter and how degraded service affects their operations. That gives the host MNO a firmer basis for capacity planning and liability limits.
5G SA coverage is expanding, but geographic consistency remains materially below the level implied by many national retail coverage claims.
Enterprise MVNO demand is shifting from generic data access towards bounded performance profiles for specific sites, devices and traffic classes.
CAMARA and NEF exposure are creating more credible interfaces for policy requests, service assurance and application-aware controls.
Regulators are paying closer attention to wholesale equivalence, including whether differentiated network treatment can be accessed on transparent and non-discriminatory terms.
The schedule must distinguish among a dedicated or reserved resource construct, a prioritised shared slice and an application-specific QoS policy. They carry different congestion, assurance and liability profiles. A reserved construct consumes capacity even when utilisation is low. A prioritised shared slice changes treatment during contention but may not provide a hard capacity floor. An application policy can improve selected flows without creating an independently engineered slice.
Each construct also needs explicit boundaries for geographic availability, access technology coverage, roaming, device and SIM eligibility, and persistence across inter-PLMN mobility. Tier-2 MNO, Western Europe, ~18M subscribers reflects a common early model: limited metro-area availability, two preconfigured service profiles and distribution through enterprise MVNO channels before any broad national offer. That approach constrains operational variance while the operator gathers utilisation, assurance and support data.
Admission control and UPF placement set the actual SLA boundary
Admission control belongs in the commercial specification, not only in engineering runbooks. The MNO must state whether the MVNO receives a reserved capacity envelope, a capped concurrent-session allowance or policy-based priority during RAN contention. Those mechanisms are not interchangeable. A capacity envelope can support a committed load within a defined footprint, while priority treatment remains dependent on radio conditions and demand from other users.
PCF rules, AMF selection, NSSF behaviour and SMF steering require a coherent ownership model. The MVNO may request a profile through an agreed interface, but the host should retain final authority over resource admission, congestion controls and emergency policy changes. Contract language should identify which requests are automatic, which require approval and which can be suspended to protect network integrity.
UPF placement often determines latency more directly than the slice label presented on a tariff sheet. A regional UPF can support a bounded proposition if the service area and transport path are known. An edge UPF requires a defined site footprint, failover path, capacity threshold and backhaul responsibility. Where the MVNO supplies an enterprise overlay, SD-WAN underlay or private application endpoint, the host should measure performance only to the agreed N6 demarcation. End-user application response time introduces systems and routes outside the host’s control.
Assurance should separate radio access availability, PDU session establishment, packet delay, packet loss and throughput. A single headline percentage conceals the dependencies that determine liability. Radio availability may be healthy while session establishment fails because of policy or core selection. A session may establish successfully while packet delay deteriorates after traffic is moved to a different user-plane location. Each domain needs its own measurement point, sampling method and exclusion logic.
Roaming and multi-IMSI designs require a separate policy path. Greenfield MVNO, post-2023, multi-IMSI stack illustrates the issue: a domestic commitment generally does not survive IMSI switching, visited-network registration or breakout through a partner packet core. The retail service must identify when the protected profile applies, how the device is notified of a profile change and what fallback treatment replaces it.
Charging treatment will determine whether the model scales beyond bespoke contracts
A flat per-GB rate is inadequate when the host reserves capacity, applies premium QoS or steers traffic to a specific UPF. A repeatable model needs a fixed availability component, a committed usage allowance and clearly priced overage treatment. The fixed element pays for footprint and resource readiness. The allowance establishes the expected load. Overage pricing gives the host a mechanism to recover incremental cost without renegotiating the schedule each month.
The MNO’s CHF and OCS logic should distinguish slice- or DNN-associated traffic from general data usage while preserving auditable mediation records for wholesale settlement. Rating dimensions may include profile, location, time band, admitted capacity and user-plane destination. They should remain stable enough for the MVNO to reconcile invoices against session and usage records rather than accept an aggregate premium-data line item.
Auditable settlement also requires correlation among PCF decisions, CHF or OCS events and session outcomes. A policy approval should be traceable to the relevant PDU session, charging identifier and measured service profile. If admission is denied, the record should show whether the agreed envelope was exhausted, a policy dependency failed or the network was unavailable. Without that chain, both parties can observe the same usage volume while reaching different conclusions on chargeability and SLA performance.
Failed admission and degraded service need distinct charging treatment. A session denied because the MVNO has exhausted its contracted envelope is not equivalent to a general network failure. Nor is a successfully admitted session that misses its latency target equivalent to unserved demand. The contract should assign event classes, invoice treatment and service-credit eligibility to each outcome, with records visible to both parties through the agreed assurance process.
Retail terms must remain inside the wholesale boundary. An MVNO should not promise latency, throughput or geographic continuity beyond what the host measures and commits. The same alignment should cover maintenance windows, exclusions, service credits and force-majeure treatment. Otherwise, the MVNO assumes open-ended retail liability while the upstream contract provides only a narrower remedy.
Commercial governance should include quarterly capacity reviews, forecast submission windows, profile-change lead times and a defined process for adding UPF locations or expanding the footprint. MVNE servicing 12+ tenants in EMEA shows the scaling risk. If a host creates tenant-specific PCF rules without a standard catalogue, mediation model and change-control process, operational cost will rise faster than wholesale revenue. Exceptions should therefore carry explicit implementation charges, minimum terms and retirement conditions.
The first viable slice-wholesale offers will be narrow, geographically bounded and attached to identifiable enterprise traffic classes. That is a commercial discipline rather than a technical limitation. MNOs that treat admission, policy, UPF topology and charging as one commercial control plane will be better positioned to convert 5G SA investment into repeatable MVNO revenue, with liabilities that remain measurable and settlement records that can withstand monthly reconciliation.