If your plant stood up a private 5G pilot in 2023 or 2024, you’re staring at a decision right now, whether anyone’s put it on a project charter or not. The CBRS-based network in your welding cell or the shared-spectrum deployment in your EU facility is coming up on its contract renewal, and somebody with a budget is asking the obvious question: do we expand this, or did we just spend a few years proving Wi-Fi would’ve been fine?
That question gets asked as if it’s about coverage and throughput. It isn’t, mostly. The real bottleneck nobody budgeted for in 2023 is still sitting there unsolved: most private 5G deployments never got wired into the MES or Unified Namespace data layer. The network got faster. The data model didn’t get better. And that gap is exactly what determines whether expansion is worth it.
What the pilots actually proved
Give private 5G credit for what it’s good at. Deterministic latency and dedicated uplink bandwidth for AMR and AGV fleets is a real, well-understood advantage over shared unlicensed spectrum, especially as fleet size grows past a handful of vehicles and you start seeing contention on a shared Wi-Fi channel. Network slicing lets you separate safety-critical AGV traffic from bulk sensor backhaul on the same physical radio infrastructure, at least in principle. SIM-based device authentication is cleaner than managing Wi-Fi credentials across a fleet that’s constantly being reconfigured. These are legitimate, technically sound reasons plants piloted 5G in the first place, and none of that reasoning was wrong.
What most pilots didn’t do — because it wasn’t in scope, or nobody owned it, or the systems integrator selling the radio gear wasn’t the same integrator who understood ISA-95 data contracts — was connect the 5G core’s device and slice management to anything the MES, historian, or UNS actually consumes. The private 5G core knows which AMR is on which slice, what its signal quality looks like, and how much bandwidth it’s burning. The MES knows nothing about that, because nobody built the bridge. You end up with a genuinely fast, genuinely reliable network carrying the same loosely-structured, inconsistently-modeled data it was carrying before, just with fewer dropped packets.
The bolted-on connectivity problem, concretely
Here’s what this looks like on the floor. An AMR fleet expansion means more vehicles reporting position, battery state, task status, and fault codes. If that telemetry was never mapped into a consistent namespace — ISA-95-aligned equipment hierarchy, MQTT Sparkplug B payloads with real metric definitions, OPC UA information models with actual semantics attached — then every new AMR vendor, every new sensor type, every new use case gets integrated as its own one-off point-to-point connection. The 5G core doesn’t fix that. It just moves the same badly-structured data faster and more reliably.
This matters more as fleets scale, not less. A five-AMR pilot with a spreadsheet-style integration is annoying but survivable. A fifty-AMR fleet with the same ad hoc data handling turns into a genuine operational risk: traffic management decisions, WIP location, and dispatch logic all depend on data nobody modeled consistently, and now it’s load-bearing for production, not a demo.
The actual decision framework
Before renewing or expanding, work through these in order — they matter more than the RF site survey.
- Do you have a real data contract, or a pile of connections? If every device type — AMR, AGV, wireless sensor gateway — has its own bespoke integration into MES or your historian, expanding the network just multiplies that integration debt. Fix the data model before you fix the radio.
- Is the 5G core’s management plane talking to anything? Slice status, device health, signal quality — if none of that surfaces anywhere your MES or CMMS can act on it, you’re paying for network intelligence you’re not using operationally.
- What’s your actual device density and mobility profile? Wi-Fi 6E, with OFDMA and wider channels, genuinely handles moderate AMR fleets and fixed sensor backhaul well in a lot of plants. Private 5G earns its keep when you have real mobility at scale, wide-area coverage across a large facility or campus, or a hard requirement for guaranteed latency on safety-relevant traffic. If you don’t have that profile, you may be paying for capability you don’t need.
- Who owns this two years from now? Private 5G cores need ongoing management — spectrum coordination, SIM provisioning, slice configuration — that usually falls outside both classic OT and classic IT. If that ownership was never resolved during the pilot, expansion just scales the same organizational gap.
- Does expansion actually reduce integration work, or add to it? More AMRs, more AGVs, more wireless sensor points on a network with no data contract means more one-off integrations, not fewer. Expansion should be paired with — not substituted for — building a real UNS or MES data layer underneath it.
What “ready now” actually looks like
Deterministic connectivity for growing AMR/AGV fleets: ready, and increasingly the more defensible choice as fleet size and mobility increase. Network slicing as a genuinely useful operational tool: technically ready, but only if someone builds the interface between slice management and your MES or SCADA layer — most plants haven’t. Private 5G as a plug-and-play replacement for structured data integration: not ready, and it was never going to be, because that was never what the radio layer does.
What to actually do
If you’re at the renewal point, don’t let the conversation start with coverage maps and spectrum licensing. Start with an honest inventory of what data model — if any — sits underneath your current pilot. If the answer is “none, really, it’s point-to-point connections we built as we went,” fix that first, on whatever network you’re currently running, before you sign a multi-year expansion. A well-integrated Wi-Fi 6E deployment with a real Unified Namespace underneath it will outperform an expanded private 5G footprint with no data contract, on every metric that actually matters to production — traceability, dispatch logic, OEE calculation, maintenance triggers.
Private 5G is a legitimate answer to a specific set of problems: large-area mobility, guaranteed latency, fleet scale that strains contention-based wireless. It is not an answer to the harder problem most plants are actually carrying, which is that nobody defined what a “device” or a “metric” means consistently across the floor. Solve that, and the radio choice gets a lot easier to make honestly. Skip it, and you’ll be back here in another two or three years, renewing a faster network that still can’t tell your MES anything it can use.
This article was written with the assistance of artificial intelligence. While we aim for accuracy, the information may be incomplete, out of date, or incorrect, and should be independently verified before you rely on it for any decision. It is provided for general information only and does not constitute professional advice.
