PROFINET vs EtherCAT vs OPC UA FX with TSN: The 2026 Deterministic Ethernet Decision

Most machine builders asking “which industrial Ethernet should we standardise on?” are asking the wrong question. The three contenders are not three answers to one problem. EtherCAT and PROFINET are mature, certified, hardware-backed fieldbus families that solve cyclic I/O and motion today. OPC UA FX is a vendor-neutral information and connection model that rides on top of a converged network, with Time-Sensitive Networking (TSN) supplying the determinism. Treating them as like-for-like rivals produces bad purchasing decisions.

This matters in 2026 because the OPC Foundation’s Field eXchange (UAFX) work has moved from concept to release candidates, controllers from several vendors are in interoperability events, and TSN silicon is shipping in switches and device chips. The PROFINET vs EtherCAT vs OPC UA FX question is now a real architecture decision for new lines, not a research topic.

You will leave with a layer-by-layer mental model, a cycle-time and jitter comparison with sources and honest labels, a decision matrix by use case, and a phased migration path that does not bet the plant on a specification that is still maturing.

What this covers: the layer model that separates the three, how each achieves determinism, a comparison matrix, failure modes, and a pragmatic 2026 recommendation.

Context and Background

Deterministic Ethernet means that a frame sent every cycle arrives within a bounded, known latency and with bounded variation (jitter), regardless of other traffic. Standard switched Ethernet cannot promise this, because a full-size frame already being transmitted blocks everything behind it. Industrial protocols solved that problem in different ways over two decades, and the answers are why the three technologies behave so differently.

EtherCAT, managed by the EtherCAT Technology Group (ETG, founded in November 2003 according to the ETG site), avoids switching entirely. A single frame travels through all devices, and each device reads and writes its own data “on the fly” as the frame passes. The ETG states cycle times of 100 microseconds or below, distributed-clock jitter of 1 microsecond or less, and more than 90 percent frame utilisation. It is standardised in the IEC 61158 series, with Safety over EtherCAT in IEC 61784-3, and sits in the IEC 61784-2 profile family as Communication Profile Family 12.

PROFINET, maintained by PROFIBUS and PROFINET International (PI), uses ordinary switched Ethernet with a priority-tagged real-time class (RT) and, for synchronised motion, Isochronous Real-Time (IRT), which reserves time slots in each cycle using dedicated hardware. PROFINET is Communication Profile Family 3 in the same IEC family. If you want the protocol-level mechanics of IRT, our PROFINET IRT and TSN real-time industrial Ethernet guide goes deeper than this comparison can.

OPC UA FX is different in kind. It extends OPC UA, standardised as IEC 62541, with a connection manager, information models and profiles for field-level exchange. Its first release scope concentrates on controller-to-controller (C2C) exchange using OPC UA PubSub, with controller-to-device (C2D) following. The OPC Foundation’s Part 80 overview describes the release as mapping connections onto PubSub “including the usage of TSN”, and states that UAFX builds on the IEC/IEEE 60802 TSN Profile for Industrial Automation. Part 80 is the overview; Part 81 covers connecting devices and the information model; Part 84 covers profiles.

The status matters, so here it is plainly. The OPC Foundation’s March 2026 field-level communications update reports the UAFX 1.00 series release candidate V1.00.04, with Part 81 extensions to the ConnectionManager and ConnectionConfigurationSets. A Berlin interoperability event in February 2026 focused on next steps for TSN-based real-time communication, and a multi-vendor C2D demonstration was being organised for SPS Nuremberg in November 2026. In other words: promising, actively tested, not yet a drop-in replacement for a fieldbus on a high-speed servo axis. We tracked the revision differences in OPC UA FX v1.00.04 vs v1.00.03.

For authoritative detail, read the EtherCAT Technology Group technology page, the PI PROFINET over TSN page, and the OPC UA Part 80 specification.

The Core Argument: Three Different Layers, Not Three Rivals

The direct answer: EtherCAT is a complete, hardware-optimised fieldbus that wins on raw cycle time and cost per node in a single-vendor cell. PROFINET is the best-supported option when Siemens or PI-certified ecosystems dominate and you need both RT and IRT. OPC UA FX over TSN is the only one designed for multi-vendor, IT-converged networks, but its determinism comes from the network underneath, and its ecosystem is still forming.

Figure 1: A first-pass decision flow for PROFINET vs EtherCAT vs OPC UA FX. Hard synchronisation and single-vendor cells point to EtherCAT, mixed ecosystems to PROFINET or FX.

Figure 1 reduces the decision to three questions: how hard is your synchronisation requirement, how many vendors share the cell, and whether the network must carry IT traffic alongside control. The flow is deliberately coarse. Its job is to stop teams from comparing a bus with a framework.

Where each technology lives in the stack

Think in three layers. The bottom is the link and scheduling layer: how bits and time are managed on the wire. The middle is the real-time protocol: how cyclic process data is framed, addressed and supervised. The top is the semantic layer: what the data means and how engineering tools discover it.

EtherCAT collapses all three layers into one vertically integrated design. The ring-style frame processing, the distributed clocks and the process-image mapping are all specified together, and the device-side logic lives in a dedicated EtherCAT SubDevice Controller (ESC) chip or an FPGA core. That integration is why it is fast, and also why it is rigid.

PROFINET splits the layers. The link layer is standard IEEE 802.3 with 802.1Q priority tagging. The real-time layer is PROFINET RT or IRT. The semantic layer is the GSDML device description and the PROFINET engineering model. PI has been integrating TSN: the PROFINET specification V2.4 with integrated TSN functionality has been published, per PI, as three document parts. PI describes TSN as complementary, addressing IRT-style clock-synchronous applications and the RT area, while RT and IRT remain in the specification.

OPC UA FX splits the layers most aggressively. It does not define the link layer at all. It assumes a TSN-capable network built from IEEE 802.1 standards and the IEC/IEEE 60802 profile. It defines the connection model and information model on top, and it reuses OPC UA PubSub, security and information modelling. That is why it can claim vendor neutrality, and why its determinism is only as good as the network you build.

The thesis the comparison tables miss

Most comparisons rank the three by cycle time. That is the least useful axis for a 2026 decision. The sharper framing is who owns the timing guarantee. In EtherCAT, the protocol owns it, enforced by silicon in every device. In PROFINET IRT, the protocol and its hardware own it. In OPC UA FX over TSN, the network owns it, through configured gate schedules and clock sync, and the application merely rides on top.

That has a profound operational consequence. When the protocol owns timing, a certified device either meets the cycle or it does not, and commissioning is largely device-by-device. When the network owns timing, a single misconfigured bridge, a wrong gate schedule, or a clock-sync fault can degrade every stream sharing that path. The skill set shifts from fieldbus tuning to network engineering. Teams that cannot staff for that shift should weigh it heavily, whatever the cycle-time table says.

Deeper Analysis: How Each One Actually Achieves Determinism

EtherCAT: one frame, processed in hardware

An EtherCAT main device sends a frame addressed to the whole segment. Each SubDevice’s ESC reads the datagrams addressed to it and inserts its own input data into the same frame while the frame is still moving downstream, with only a few nanoseconds of forwarding delay per device. The last device in a line turns the frame around, and the main device receives the completed frame once per cycle.

Figure 2: EtherCAT passes one frame through every device. PROFINET and OPC UA FX use a switched network where TSN bridges schedule traffic.

Three properties follow directly from that design. First, there is no per-device frame, so utilisation is very high; the ETG quotes more than 90 percent. Second, latency is the sum of per-device forwarding delay plus cable propagation, which is predictable by construction. Third, the main device does not need special hardware, since it can use a standard Ethernet port and a software stack, though very short cycles need a real-time operating system and careful driver work.

Synchronisation uses distributed clocks. One SubDevice, typically the first clock-capable one, acts as the reference clock. Beckhoff’s documentation states that clocks are synchronised with a precision below 100 ns regardless of distance between devices, with a 1 ns time unit and a zero point of 1 January 2000. The ETG headline figure is jitter of 1 microsecond or less and synchronisation “much less than 1 µs”. Those are best-case vendor-published figures, and your measured jitter will depend on the main device’s operating system, not on the bus.

The cost of the design is topology and coupling. Every device is in the logical path of the frame. Modern ESCs support cable redundancy and hot-connect groups, and the ETG quotes cable-break detection in under 15 microseconds, but the architecture is fundamentally a segment under one main device. Crossing between two EtherCAT segments, or between a segment and a different vendor’s controller, requires a bridging mechanism such as a gateway or an application-layer protocol. That is exactly the gap OPC UA FX targets.

PROFINET: classes of service on switched Ethernet

PROFINET uses standard Ethernet frames with an EtherType assigned to PROFINET real-time data. In RT, cyclic frames get priority through 802.1Q tagging and are forwarded by ordinary store-and-forward switches. Determinism is statistical: the guarantee depends on switch queueing, link load and the absence of competing high-priority traffic. For most discrete and process automation, with cycle times in the low millisecond range, that is entirely adequate.

IRT changes the mechanism. Each cycle is divided into a reserved phase for time-critical frames and an open phase for everything else. IRT-capable switch ASICs, built into each device, forward reserved-phase frames according to a pre-computed schedule, with the schedule computed by the engineering tool from the topology. This is why IRT requires topology to be declared up front and why IRT-capable hardware is needed in every device along a synchronised path.

PI documents typical IRT performance in the sub-millisecond range for isochronous motion, but exact minimum cycle times and jitter depend on the device generation and send-clock settings. I could not verify a single authoritative minimum-cycle figure for the current specification on the sources I consulted, so treat any specific number you see quoted, including in vendor brochures, as a typical value to be confirmed against the device manual for the exact hardware you plan to buy.

PROFINET over TSN replaces the IRT-specific scheduling mechanism with IEEE-standard ones. The intent, per PI, is to let PROFINET share a network with other TSN-based protocols while still covering the application area of IRT and RT. In practice, vendor support is arriving in stages, and your engineering tool, controller firmware and devices must all support the same specification revision.

OPC UA FX over TSN: determinism from the network

OPC UA FX does not carry timing guarantees in its own framing. It uses OPC UA PubSub, with the UADP message encoding, to publish datasets, and relies on TSN to deliver them within a bounded time. That puts the determinism contract in a set of IEEE 802.1 standards.

Figure 3: In an OPC UA FX over TSN network, gPTP time keeps bridges aligned, Qbv gates open a protected window, and the cyclic frame crosses the bridge in bounded time.

The pieces, in the order they matter, are these:

• IEEE 802.1AS (generalised Precision Time Protocol, gPTP) gives every bridge and end station a shared notion of time. Without it, gate schedules drift apart and guarantees fail silently.
• IEEE 802.1Qbv (time-aware shaper) controls output queue gates with a repeating schedule so a protected queue gets an exclusive window each cycle.
• IEEE 802.1Qbu with IEEE 802.3br (frame preemption) lets an express frame interrupt a long best-effort frame, shrinking the guard band that Qbv would otherwise need.
• IEEE 802.1Qci (per-stream filtering and policing) protects the schedule from a misbehaving talker.
• IEEE 802.1CB (frame replication and elimination for reliability) sends duplicate frames along disjoint paths for seamless redundancy.
• IEEE 802.1Qcc (stream reservation and configuration) defines how streams and schedules are configured, with centralised models that match how industrial engineering tools already work.

The IEC/IEEE 60802 profile selects and constrains these features for industrial automation. Our own IEC/IEEE 60802 industrial TSN profile analysis covers what the profile requires and where its publication status stands; check the IEC and IEEE catalogues for the current state before you commit a design to it.

For the underlying network design, see our TSN industrial reference architecture and the OPC UA TSN deterministic IIoT reference architecture.

A worked timing budget

A cyclic exchange has to fit inside its period. Take an illustrative case, with the numbers labelled as a teaching example rather than a measurement. A 1 ms cycle carries 64 bytes of cyclic data per stream. A 100 Mbit/s link needs roughly 8 to 10 microseconds to serialise a frame of that size once preamble, headers, FCS and interframe gap are included. A maximum-size best-effort frame of about 1,518 bytes needs about 123 microseconds on the same link, plus overhead.

Without preemption, a TSN gate must close well before the protected window to be sure a best-effort frame cannot still be transmitting when the gate opens. That guard band costs roughly one maximum frame time per hop, which is why Qbv without Qbu wastes bandwidth on slow links. With preemption, the guard band shrinks to roughly the preemption fragment granularity. On gigabit links the numbers shrink by a factor of ten, which is one reason 1 Gbit/s TSN backbones are the practical default for new designs.

The lesson generalises. Determinism on a switched network is an accounting exercise: serialisation time, per-hop forwarding latency, guard bands and clock error must all be summed against the cycle. EtherCAT hides that accounting inside the protocol. TSN makes you do it, and your engineering tool or network configuration entity should do it for you.

Security, safety and the semantic layer

Security is where OPC UA FX has a structural lead. OPC UA defines application authentication with X.509 certificates, signing and encryption of messages, and a PubSub security model with key distribution through a Security Key Service. The OPC Foundation’s March 2026 update specifically calls out certificate-based secure communication for the planned C2C engineering workflow demonstration. PROFINET addresses security through PI’s security classes in recent specification versions, while classic EtherCAT has no built-in message authentication, so it is typically protected by physical segmentation and network design.

For functional safety, EtherCAT has Safety over EtherCAT (FSoE), rated SIL 3 capable per the ETG, and PROFINET has PROFIsafe. Both are black-channel protocols that tolerate an untrusted transport. UAFX Part 80 states that both safety and standard process data exchange are supported in the connection model, but the safety mapping is far newer and certified product choice is much narrower. If your machine has a safety function on the critical path, the maturity of the safety stack and its certification body, not the fastest cycle, will usually decide.

On the semantic layer, OPC UA brings its information-model strength: companion specifications for machine types, robotics, and others, plus the Asset Administration Shell mapping trend. PROFINET relies on GSDML for device description and, increasingly, OPC UA companion specifications for upward integration. EtherCAT uses the EtherCAT Slave Information (ESI) XML and CANopen over EtherCAT (CoE) object dictionaries, with the ETG Technical Working Groups publishing device profiles. If your goal is feeding a digital twin or unified namespace, our OPC UA FX field exchange reference architecture shows where the model fits.

The 2026 Comparison Matrix

The table below summarises what can be sourced. Where a figure comes from the ETG or OPC Foundation it is attributed; where I could not confirm a number, the cell says so rather than guessing.

Dimension	EtherCAT	PROFINET (RT / IRT / over TSN)	OPC UA FX over TSN
Governing body	ETG	PI	OPC Foundation
IEC family	IEC 61158, 61784-2 CPF 12, safety in 61784-3	IEC 61158 and 61784-2 CPF 3	IEC 62541 for OPC UA, UAFX parts 80 to 84
Determinism owner	Protocol and ESC silicon	Protocol and IRT ASIC, or TSN network	TSN network configuration
Quoted cycle time	100 microseconds or below (ETG)	Sub-millisecond for IRT, exact minimum not verified	Depends on network and device, no single published minimum verified
Quoted jitter	1 microsecond or less (ETG)	Not verified on sources consulted	Set by gPTP accuracy and gate schedule
Topology	Line, tree, star, daisy-chain	Switched, with IRT needing declared topology	Any TSN-capable switched topology
Mixed-vendor controllers	Gateways needed	Within PROFINET only	Core design goal
Convergence with IT traffic	Limited, separate segment	Good for RT, constrained for IRT	Core design goal
Built-in security model	Not message-authenticated by default	PI security classes in recent specs	OPC UA X.509 and PubSub security
Safety	FSoE, SIL 3 capable	PROFIsafe	Safety data exchange in model, newer
Ecosystem maturity	Very high	Very high	Emerging, release candidates in 2026

Read the cycle-time rows with care. The EtherCAT numbers are the protocol’s published capability; the practical cycle in your machine is limited by the main device’s CPU, the drive’s own control loop, and the process image size. A servo drive that closes its position loop at 62.5 microseconds gains nothing from an exchange rate faster than it can use. The same caution applies to PROFINET and FX: the headline cycle assumes ideal hardware.

Decision matrix by use case

Use case	Best fit today	Why	Watch out for
Multi-axis packaging or CNC with tight sync	EtherCAT, or PROFINET IRT	Mature drives, sub-microsecond DC sync	Vendor lock-in on the drive family
Siemens-centric discrete line	PROFINET	Engineering toolchain and device catalogue	IRT needs declared topology and capable hardware
Process plant with many remote I/O islands	PROFINET RT	Millisecond cycles are ample, strong diagnostics	Over-specifying IRT where RT suffices
Two controllers from different vendors exchanging data	OPC UA FX C2C	Vendor-neutral, secure by design	Release-candidate maturity, test interoperability
Greenfield plant planning a converged OT and IT backbone	OPC UA FX over TSN, with a fieldbus in cells	Single network, standard IEEE 802.1 tooling	Needs real TSN network engineering skill
Small machine, cost-sensitive, single vendor	EtherCAT	Low cost per node, simple topology	Remote segments and cross-vendor links
Retrofit with existing fieldbus	Keep, add OPC UA upward	Lowest risk	Do not rip and replace for protocol fashion

The most important row is the last one. Replacing a working fieldbus to adopt a newer standard is rarely justified by cycle time. It is justified by a business need such as multi-vendor controller exchange or network consolidation.

What the existing comparisons already say

We compared the classic motion options in EtherCAT vs PROFINET vs SERCOS for motion control, and a shorter overview sits in PROFINET vs EtherCAT vs OPC UA FX with TSN in 2026. This article goes further on the question the earlier ones touch only briefly: who owns the timing guarantee, and what that means for the team that has to run the plant.

C2C versus C2D: the distinction that decides adoption speed

OPC UA FX defines two interaction models, and they have different adoption curves. Controller-to-controller (C2C) is about PLCs and edge controllers exchanging data with each other, which today is done with a patchwork of proprietary links, gateways and OPC UA client-server. Part 80 states that the current release concentrates on C2C. The OPC Foundation’s March 2026 update describes a live C2C engineering workflow demonstration emphasising certificate-based secure communication on real PLC hardware.

Controller-to-device (C2D) is the harder one. It means a controller driving motion devices, remote I/O and field instruments with the cyclic discipline of a fieldbus, and Part 80 identifies it as addressed in later releases, with a multi-vendor demonstration planned for SPS Nuremberg in November 2026. We cover the distinction in OPC UA FX C2D vs C2C. My view, which is opinion rather than fact: C2C will be productive in real plants first because the timing demands are looser, while C2D will take longer to displace fieldbuses in motion because it must match both their performance and their certification depth.

Trade-offs, Gotchas, and What Goes Wrong

Clock failure is a silent failure in TSN. If the grandmaster is lost or a bridge loses sync, gate schedules misalign and cyclic frames can miss windows. Unlike a broken EtherCAT cable, which is detected within microseconds per the ETG, a drifting TSN clock may degrade performance before it raises an obvious alarm. Monitor gPTP offset and path delay on every bridge as a first-class health metric.

Gate schedules are brittle to change. Adding a stream to a Qbv network means recomputing schedules across every bridge on the path. Centralised configuration under 802.1Qcc is designed for this, but tool support varies, and brownfield additions can force a recalculation that touches running production. Reserve schedule headroom at design time.

Mixing a fieldbus gateway into a deterministic path erases the guarantee. A gateway that translates between EtherCAT and OPC UA FX adds its own buffering and jitter. If the downstream device needs synchronised outputs, the gateway’s timing becomes the weakest link. Measure end to end, not per segment.

Vendor claims about “TSN support” are not equivalent. A switch may support 802.1AS and Qbv but not Qbu, Qci or 802.1CB. A device chip may support talker functions but not the configuration model your engineering tool uses. Request the feature list against the specific standards above and the IEC/IEEE 60802 profile, and insist on interoperability evidence, not a logo.

Certification maturity is uneven. EtherCAT and PROFINET have long-established conformance test regimes. UAFX is in release-candidate stage at the time of writing, and the interoperability events cited above exist precisely because the ecosystem is still converging. Specifications can change between candidates, so pin the exact revision in your purchasing documents.

Anti-pattern: choosing on the benchmark slide. A cycle-time bar chart in a vendor deck measures a best-case configuration. Ask for the topology, node count, payload size, main-device hardware and measured jitter histogram. If a vendor cannot supply them, the number is marketing.

Anti-pattern: skipping the people question. TSN shifts responsibility from automation engineers to network engineers. If those teams have never shared a project, budget for training and a joint commissioning process before the first line goes live.

Migration: A Phased Path That Limits Risk

The safest 2026 posture is coexistence. Keep deterministic fieldbus segments inside machines, where they already work, and introduce OPC UA FX and TSN at the layer where multi-vendor and IT convergence create value.

Figure 4: A four-phase migration path. Fieldbus stays in the cell while OPC UA FX C2C and a TSN backbone are introduced first, and C2D comes last.

Phase 1, keep the cell. Do nothing to the working fieldbus. Document each cell’s cycle time, jitter, device count and vendor lock-in so you can measure any later change.

Phase 2, link controllers with OPC UA FX C2C. Replace proprietary controller-to-controller links, or ad hoc OPC UA client-server pairs, with FX connections where both controllers support it. This is the lowest-risk step: it adds the new technology at the edge of the cell, not inside the motion loop.

Phase 3, build the TSN backbone. Install TSN-capable bridges with gPTP at the plant level, begin with one or two protected streams, and instrument clock offset. Run the existing traffic in parallel until you trust the measurements. The TSN vs 5G URLLC comparison is useful if wireless segments are in scope.

Phase 4, move devices to C2D where certified products exist. Only when multi-vendor C2D products have passed interoperability and certification for your device class should FX start replacing a fieldbus segment. For synchronised motion, validate jitter with your actual drives under production load before any cutover.

Each phase should have an explicit rollback. Because the phases touch different layers, a failed Phase 3 does not undo the value of Phase 2.

Practical Recommendations

For a new single-vendor machine where cost and motion performance dominate, EtherCAT remains the lowest-risk, best-documented choice. Its numbers are published, its ecosystem is wide, and nothing in the 2026 FX release candidates changes that for the inner motion loop.

For a plant standardised on a PROFINET ecosystem, stay with PROFINET, use RT wherever millisecond cycles suffice, reserve IRT for the genuinely synchronised axes, and ask your controller vendor for a written roadmap on PROFINET over TSN support and the specification revision it targets.

For a greenfield or multi-vendor site, design the network as a TSN backbone from day one, adopt OPC UA FX C2C as soon as your controller vendors ship certified implementations, and keep fieldbus inside cells until C2D matures. Procure TSN switches against the specific IEEE 802.1 feature list, not a generic claim.

Use this checklist before you sign a purchase order:

• Does the vendor name the exact specification revision, such as UAFX 1.00.04 or PROFINET V2.4?
• Is there independent interoperability evidence, such as participation in an OPC Foundation or PI event?
• Which 802.1 standards does each switch implement, and which TSN profile does it follow?
• What is the measured jitter histogram under load, with your topology and payload?
• Who owns the gPTP grandmaster, and what is its holdover behaviour?
• How are schedules configured and changed, and who is trained to do it?
• Where do safety functions run, and what is the certified safety stack?

Frequently Asked Questions

Is OPC UA FX going to replace PROFINET and EtherCAT?

Not in the near term. OPC UA FX defines an open connection and information model over TSN, but its first release concentrates on controller-to-controller exchange, with controller-to-device products still being demonstrated in 2026. PROFINET and EtherCAT have large installed bases, mature certification and hardware-enforced timing. The likelier outcome is coexistence, with FX linking controllers and plants while fieldbuses continue inside machines.

Is EtherCAT or PROFINET faster?

EtherCAT publishes cycle times of 100 microseconds or below with jitter of 1 microsecond or less, per the ETG. PROFINET IRT also reaches sub-millisecond cycles, but I could not verify one authoritative minimum figure for current hardware. Practical speed depends on the main device, drives and payload size, so compare measured results on your own configuration, not headline numbers.

Does PROFINET work over TSN today?

PI states that the PROFINET specification V2.4 with integrated TSN functionality has been published, with the work described as complete. Product availability is another matter: controllers, devices and engineering tools must all support the same revision. Ask your vendors which specification version and which TSN features they implement, and request interoperability evidence.

What does TSN actually add to Ethernet?

TSN is a set of IEEE 802.1 and 802.3 standards. The key ones are 802.1AS for time synchronisation, 802.1Qbv for scheduled gates, 802.1Qbu with 802.3br for preemption, 802.1Qci for stream policing and 802.1CB for redundancy. Together they let converged Ethernet carry control traffic with bounded latency alongside best-effort IT traffic. They do not make an unconfigured network deterministic.

Can I mix EtherCAT and OPC UA FX in one plant?

Yes, and for most brownfield plants this is the sensible approach. Keep EtherCAT inside motion cells and use OPC UA FX between controllers or toward the plant backbone. The boundary is a controller or gateway, so measure the timing across it. Do not expect a gateway to preserve cycle-synchronous behaviour across protocols without verification.

Is OPC UA FX ready for production in 2026?

Partly. The OPC Foundation reports release candidate V1.00.04 and active interoperability testing, including a Berlin event in February 2026. Controller-to-controller use is the nearest to practical deployment. For safety-critical or high-speed synchronised motion, wait for certified multi-vendor products and run your own validation before relying on it.

Further Reading

• PROFINET vs EtherCAT vs OPC UA FX with TSN in 2026, our earlier overview of the same three technologies.
• EtherCAT vs PROFINET vs SERCOS for motion control, a motion-focused comparison.
• OPC UA TSN deterministic IIoT reference architecture, the network design in detail.
• PROFINET IRT and TSN real-time industrial Ethernet guide, the IRT mechanism and its TSN path.
• EtherCAT Technology Group, technology overview, the source of the EtherCAT figures quoted here.
• OPC Foundation, UAFX Part 80 overview and concepts, the primary specification text.

By Riju — about