The headline 5G URLLC figure, a user-plane latency below 1 ms at 99.999% reliability, comes from 3GPP Release 15/16 radio design targets, not from a typical factory deployment [S1][S3]. The 1 ms number refers to the radio interface budget for a single subcarrier-slot transmission, not to a complete PROFINET cycle on a shop floor [S1].
Real plant data from 2020-2025 trial reports shows 5-10 ms one-way over-the-air latency even under tuned URLLC configurations, and 10-50 ms end-to-end once PROFINET RT, PROFIsafe, 5G core, and edge compute are stacked on top [S4][S5]. A reference target for remote process automation, separate from the radio figure, sits at 50 ms end-to-end with 99.9999% reliability and data rates up to 100 Mbps [S3].
What the 1 ms figure actually covers
3GPP Release 15 introduced URLLC as a 5G service class with a sub-1 ms user-plane latency target at the radio interface, paired with five-nines reliability for mission-critical traffic [S1][S3]. Transmissions over a fraction of a slot, down to a minimum of two OFDM symbols against the usual 14, were added to the 5G New Radio design specifically to chase that target [S3].
3GPP Release 16 added redundant transmission paths across the radio and core so that a single link failure does not cause a packet to miss its delay budget [S3]. These features are real, but they are radio-layer and core-layer features; they do not by themselves determine closed-loop performance in a discrete-automation cell, which is what most plant engineers care about. A useful pressure transmitter in a fast process loop still has to share the same wireless span with everything else.
What a real plant deployment measures
The Ericsson and Audi demonstration in Kista, January 2020, ran PROFINET RT and PROFIsafe over a private 5G URLLC network to control a robot cell building an airbag subassembly, with a laser-curtain safety stop triggered over the wireless link [S5]. That test is widely cited as proof that safety-rated motion control can run over 5G, but the documented deployment still required edge hosting of the PROFINET controller and tight integration with time-sensitive networking bridging [S5].
Independent surveys from 2022-2025 keep finding that 5G URLLC, even with Release 16 redundancy, does not meet the latency or reliability profile of every critical loop inside industrial automation, particularly where hard real-time safety functions are involved [S6][S9]. The reliable closed-loop envelope that 5G can sustain in practice is closer to the 5-10 ms one-way range than to the 1 ms radio target, and that gap is what drives the TSN bridging pattern now common in brownfield plants [S4][S6]. A shop floor that already uses industrial valve positioners on wired buses will not gain much by re-architecting them onto raw 5G.
Decision matrix: raw 5G vs 5G plus TSN vs wired
Three deployment patterns dominate industrial 5G work in 2026. Use these criteria to pick between them. [S2]
Raw 5G URLLC, radio only, no TSN bridging: lowest cost per radio, but the documented end-to-end envelope is 10-50 ms and reliability depends heavily on coexistence with eMBB traffic on the same cell [S1][S3][S6]. Best for telemetry, condition monitoring, and mobile asset tracking on a flow meter reading, not for safety stops.
5G URLLC with 5G-TSN bridging and on-prem edge: a programmable TSN switch such as Canoga Perkins SyncMetra bridges the cellular segment into a deterministic TSN island, keeping the wireless hop in the 5-10 ms range and bounding the end-to-end jitter to TSN class budgets [S4]. This is the pattern most new private 5G factories converge on, because it keeps PROFINET and EtherNet/IP toolchains intact [S4][S5].
Wired PROFINET, EtherCAT, or deterministic Ethernet: still the reference for hard real-time loops under 1 ms cycle time, especially safety-rated drives and motion control, where 5G URLLC alone is not yet considered sufficient [S6][S9]. Use this where the failure cost is a line stop, an equipment write-off, or a personnel injury.
Where the marketing claim diverges from the spec
Vendor material often quotes URLLC as "latency under 1 ms" without separating the radio hop from end-to-end application latency [S2][S4]. The 3GPP source material itself uses different reliability targets for different use cases, from 99.999% general URLLC up to 99.9999% for remote process automation, paired with relaxed latency budgets of 50 ms [S3]. Treating those numbers as interchangeable is a common procurement error.
Coexistence with enhanced mobile broadband (eMBB) and massive machine-type communication (mMTC) on the same 5G cell remains an open problem in 3GPP Releases 15-19, because scheduling gains for one service class can starve another on the same spectrum [S1]. For a brownfield plant, that means a private 5G slice dedicated to URLLC will behave differently from a shared slice that also carries video, AR/VR, or IT traffic [S1][S4]. An existing PLC rack on a wired backplane will not be made more deterministic by adding 5G to the cell.
Selection criteria for a plant engineer
First, classify the loop. Hard real-time safety functions, motion control under 1 ms cycle, and closed-loop position regulators belong on wired Ethernet or TSN-anchored wired backhaul; the 5G URLLC literature still treats these as out of scope for wireless alone [S6][S9].
Second, set the radio budget. If the design needs 99.999% reliability at the 5 ms one-way mark with PROFINET encapsulation, plan for 5G URLLC plus edge-hosted PROFINET plus TSN bridging, and budget the integration time accordingly [S4][S5]. If a vendor quotes 1 ms without specifying the protocol stack, treat that as the radio figure only, not the loop figure [S1][S3].
Third, account for coexistence. A mixed-traffic private 5G cell will show higher worst-case latency than a dedicated URLLC slice, so the same radio delivered as part of an enterprise-wide 5G LAN should not be specified against the same numbers [S1][S4]. For guidance on choosing sensors and other field hardware that survive that variability, see the 2D vs 3D vision system selection matrix for similar criteria-based engineering trade-offs.
Standards and citations worth tracking
3GPP Release 15 introduced the URLLC service class with the sub-1 ms radio target and Edge Computing as the mechanism for cutting end-to-end latency [S3]. Release 16 added redundant transmission across the radio and core to harden reliability within a delay budget [S3]. Releases 17-19 continue to refine coexistence between URLLC, eMBB, and mMTC on shared spectrum, a problem the academic survey literature flags as unsolved in production [S1][S6].
3GPP TS 22.261 service requirements and TS 23.501 QoS framework are the underlying specifications that define the 50 ms, 99.9999% remote-automation envelope cited above [S3]. IEEE 802.1Qcc and the IEC/IEEE 60802 TSN profile for industrial automation are the standards that 5G-TSN bridging inherits, and that engineers specifying 5G URLLC plus TSN will encounter directly [S4].
For projects that combine 5G with mobile robotics or AGV fleets, the published safe-AGV-over-5G work with SICK and Audi is the most concrete reference deployment available, and the article on fixed IR vs contact wireless sensors covers a similar safety-versus-flexibility trade-off in a different domain. PROFINET over 5G URLLC, as run in the Kista demo, remains the most-cited industrial benchmark for the pattern, and the 2025 academic survey confirms that latency and reliability targets in 5G URLLC remain case-dependent rather than universal [S5][S6].