Almost every 5G network in the GCC launched on non-standalone. That was the correct decision at the time: NSA reuses the existing 4G core as an anchor, so the operator gets 5G radio speeds to handsets without touching the core network. Time to market was measured in months instead of years.
The question now is not whether standalone is better in the abstract. It is what specifically an operator cannot sell until it has an SA core, and whether those things are on the revenue plan.
What NSA cannot do
Non-standalone gives you throughput. It does not give you any of the following.
Network slicing. A slice is an end-to-end logical network with its own performance guarantees. It requires a 5G core with the service-based architecture and the network slice selection function. Without SA, “slicing” is QoS with better marketing. Every enterprise proposition that depends on a guaranteed, isolated, contractually-defined service — private network as a service, guaranteed capacity for a broadcast, isolated connectivity for a government department — needs this.
Ultra-reliable low-latency communication. URLLC latency targets cannot be met through a 4G core anchor. Industrial automation, remote operation and closed-loop control sit here. So does most of the credible enterprise 5G revenue in this region.
VoNR. Voice over New Radio requires the 5G core and an IMS integrated with it. On NSA, voice falls back to VoLTE on the 4G radio. That works, but it means the 5G radio is a data-only asset and the operator carries two voice paths indefinitely.
Massive machine-type communication at scale. The connection-density targets that make smart city and large-scale IoT propositions viable are an SA capability.
Edge computing with a usable anchor. MEC depends on the SA user plane function being placed where the traffic needs to break out. On NSA, the topology fights you.
The dual-mode core changes the arithmetic
The historic objection to SA was that it meant building a second core network and running two operations models in parallel. That objection is now largely obsolete.
A cloud-native dual-mode core combines the Evolved Packet Core and the 5G Core into one multi-access platform. The same infrastructure serves 4G subscribers, 5G NSA subscribers and 5G SA subscribers. AMF, SMF, SMSF, NRF, NSSF, CHF, PCF and UPF run alongside MME, SGW and PGW. The migration becomes a configuration and capacity exercise rather than a second network build.
The economics follow. On commodity x86 hardware at roughly 40 Gbps per socket and 80 Gbps per server, a single cabinet configuration can serve five million subscribers. For most operators in this region that is headroom, not a constraint — which means the conversation shifts from capital cost to organisational readiness.
What actually gates SA launch
In practice we see four blockers, and only one of them is the core itself.
- Voice. VoNR requires the IMS to be integrated with the 5G core, device support verified, and EPS fallback engineered for everywhere VoNR coverage does not reach. This is the workstream that most often sets the launch date.
- Transport and synchronisation. SA tightens latency budgets and, in TDD bands like n41 and n78, synchronisation errors present as interference rather than as an outage. Transport readiness is assessed late far too often.
- OSS/BSS. Charging, policy and provisioning need to understand slices and 5G session models. Slicing you cannot bill for is a demonstration, not a product.
- Device base. SA support in the installed handset base, and correct behaviour on fallback, has to be verified rather than assumed.
A reasonable sequence
For an operator in the UAE or the wider GCC running NSA today, the sequence that tends to hold is: deploy the dual-mode core alongside the existing EPC and migrate 4G traffic onto it first, proving the platform under real load with no new radio dependency. Then enable SA on a contained radio footprint with VoNR and EPS fallback fully engineered. Then extend, with slicing and MEC brought forward only where a signed enterprise requirement exists to justify them.
The mistake is to treat SA as a radio programme. It is a core, voice, transport and OSS programme with a radio configuration change at the end of it.
The decision, stated plainly
If your three-year revenue plan is consumer data, NSA is not holding you back and the SA business case will be thin. If it includes enterprise connectivity, private networks, guaranteed service levels, industrial applications or anything you intend to sell with a latency figure in the contract, then standalone is not an upgrade — it is the prerequisite, and the only real question is how early you start the voice workstream.