A two-wire, 10 Mbit/s Ethernet physical layer that carries power and data to field instruments, including into hazardous areas. The technology has passed plant-scale testing; for most owners the open question is when the economics of replacing working 4-20 mA wiring make sense.
2026 Edition · Industrial Control, Automation & Instrumentation
The ControlCurve Maturity Map for Industrial Control, Automation & Instrumentation
Fifty emerging technologies — from Ethernet-APL to agentic AI in the control loop — mapped across field instrumentation, field control devices, DCS/PLC/ESD/HMI control systems, and the industrial IT/OT/critical control information infrastructure that connects and protects them. A cross-section of 17 is AI/ML-driven and spans all four categories.
50Technologies tracked
4Categories covered
18Full, sourced analyses
Oct 2026Last revised
Full analyses: the technologies closest to practical use
In-depth pages on the entries with a near-term or current time to proven use. Each explains how the technology works, where it pays and where it does not, the evidence behind its placement with dated milestones, and the questions to ask before committing, with factual claims linked to public sources. Three entries were re-placed in this review: IO-Link moved up to Proven in Use, while TSN with OPC UA FX and private 5G moved back to Reality Check. The reasons are given on each page.
A single optical fibre turned into thousands of measurement points along a pipeline, well, cable or perimeter. Temperature sensing is long established; acoustic sensing for leak and intrusion detection is in production, but its performance depends heavily on installation and signal interpretation.
Self-organising 2.4 GHz mesh networks for process instruments, standardised since 2010 and now a routine engineering choice for monitoring points where running a cable would cost more than the measurement is worth.
A standardised point-to-point link that gives ordinary sensors and actuators digital parameters, identification and diagnostics. With around 71 million nodes installed it is mainstream in machine automation; its growth edges are process-adjacent equipment, wireless and safety.
Electric actuators with built-in positioning, diagnostics and communications taking over duties that pneumatic actuators have dominated for decades. Methane rules and the cost of plant air are pushing adoption; fail-safe design and upfront cost are holding it back.
Digital valve positioners that analyse their own data and tell maintenance what is wrong and what to do about it. The diagnostics are mature; the value now depends on whether the plant acts on them.
Legged and wheeled robots, and drones, that run routine inspection rounds in process plants, including in explosion-hazard zones. Certified hardware exists and named operators run it; turning the data into maintenance work is the hard part.
Automated partial-stroke and proof-test functions for safety valves, built into positioners, smart solenoids and safety systems. The technique is decades old; what is maturing is the record-keeping that lets a test count toward the safety case.
Control software that runs on standard servers or industrial PCs instead of proprietary controller hardware. The first open-architecture plant and server-based controllers from mainstream suppliers are now in production; the long-term support model is still unproven.
SCADA, HMI and historian functions hosted in a cloud region instead of on servers in the plant. Proven for monitoring distributed assets such as water networks and well pads; for closed-loop control of a process unit the local control system remains the place to be.
An architecture in which plant systems publish their data once, to a shared event-driven namespace, instead of being wired together pair by pair. The messaging standards are settled; the governance needed to make it work across an enterprise is not.
Machine learning and reinforcement learning applied to closed-loop process optimisation, alongside or above classical model predictive control. Specific commercial results exist; most of the evidence still comes from suppliers and their early customers.
Security operations that monitor and respond to threats across both business IT and the industrial control environment, with OT-specific visibility, playbooks and authority. Widely recommended and increasingly expected by regulators, but unevenly implemented.
Applying zero trust principles, verify every user, device and connection and grant only the access needed, to industrial control networks. CISA published dedicated OT guidance in April 2026; in practice most of what is sold as zero trust OT is good segmentation and remote-access control, which is worth having under any name.
Passive monitoring of industrial networks that learns normal communication patterns and flags deviations, combined with protocol decoding and threat signatures. A mature product category in which machine learning is one ingredient, not the whole story.
Machine-readable inventories of the software components inside control-system products, used to find which assets are affected when a vulnerability is disclosed. The EU Cyber Resilience Act turns SBOMs from good practice into a legal requirement for products sold in the EU from December 2027.
Site-owned cellular networks that promise high-capacity, secure wireless coverage across plants, ports and mines. The technology works, but in the process industries deployments have grown more slowly than forecast, and most serve mobility and monitoring rather than control.
Low-earth-orbit satellite broadband bringing tens-of-milliseconds latency to well pads, pipelines, mines and substations beyond terrestrial networks. It has changed what is practical at remote sites; it is a shared, best-effort service, and OT designs need to treat it that way.
The AI & machine learning cross-section
AI isn't siloed into one box on this chart — it's a thread running through all four categories, from edge-AI transmitters in the field to agentic process control and OT-aware anomaly detection. Use the AI & ML cross-section toggle above the chart, or the directory below, to see only the AI-driven entries.
17
of 50 are AI/ML-driven
How this chart is built
Horizontal position — editorial judgment
Which of the five stages a technology sits in (and how far across it) is an assessed call, weighed from vendor announcements, standards-body milestones, analyst commentary and observed deployment counts. This is the part of the chart that reflects our research.
Time horizon — editorial judgment
Each technology also carries a stated "now / near / mid / far" time-to-proven-use estimate, shown in its detail panel. Like stage placement, it is a judgement we state explicitly so it can be checked against later evidence; the full analyses set out the evidence behind each estimate.
Evidence level — E1 to E5
Separately from the stage, every entry has an evidence level: E1 concept, E2 field pilot, E3 first production use, E4 multi-site references, E5 routine specification. It answers a different question from the stage. A technology can be heavily over-promised and still have real production sites (E3), or quietly proven with little attention. The levels are explained in the methodology.
Curve height — illustrative only
The line is a reference expectation curve (attention rises, falls back, then recovers as results arrive), drawn once as a visual reference and mathematically smoothed through a fixed set of shape points. A marker's vertical position comes from sampling that same drawn curve at its horizontal (stage) position — it is not a separate score, and is not meaningful to compare between technologies.
In short: trust the left-to-right position and the stated time horizon as our assessment. Don't read anything into how high or low a point sits vertically — that's the shape of the reference curve, not a ranking. Full sourcing is in the Methodology.
Where each technology sits today
Filter by category, or use search to jump straight to a technology.
The full technology directory
Every technology on the curve, grouped by category, with a short outlook and estimated time-to-proven-use. Entries tagged "Full analysis" link to an in-depth, sourced page; the rest link to short entries that will be expanded in later editions.
Field Instrumentation
Transmitters, analyzers, sensors and the physical/digital layers that get process and asset data out of the field.
Two-wire, intrinsically-safe Ethernet that runs all the way to the transmitter — finally bringing native Ethernet, and the bandwidth it carries, into hazardous-area field instrumentation.
Field transmitters with an onboard inference chip that classifies faults, drift and process anomalies locally, before the signal ever reaches the control system.
A single fibre-optic cable turned into thousands of virtual sensors, measuring temperature, strain, vibration or acoustic signature continuously along a pipeline, well or cable run.
Machine-learning models that infer a hard-to-measure variable — composition, quality, emissions — from cheaper, faster process signals already being collected.
Builds on mature NE107-style device diagnostics by adding continuous, AI-scored health trending across an entire instrument fleet rather than one device at a time.
Self-forming, self-healing wireless mesh networks for process instrumentation, now a mainstream alternative to wiring for monitoring-class measurement points.
A simple, low-cost point-to-point protocol that turns basic sensors and switches into parameterisable, diagnosable smart devices without a full fieldbus stack.
Sensors powered by ambient vibration, thermal gradient or RF energy at the measurement point, eliminating battery replacement as a maintenance task entirely.
In-line NIR/Raman/FTIR analysers with the chemometric model embedded on-device, giving real-time composition data without a separate host-PC modelling step.
Instruments that read their intended duty from a digital-twin/engineering template and configure their own ranging, tag and diagnostics on power-up.
Chip-scale mechanical and optical sensing elements bringing multi-parameter measurement (pressure, gas, particulate) into form factors small enough for previously un-instrumented locations.
Field Control Devices
Final control elements, actuators, positioners, remote I/O and the robots and wearables now working alongside them.
I/O termination that virtualises the signal-processing and protocol-conversion layer in software running on standard servers, rather than in dedicated I/O card firmware.
Electric actuators with integrated positioning, diagnostics and communications displacing pneumatic actuation on valves, dampers and switching devices.
Positioner-embedded analytics that move beyond NE107-style status flags to predicted remaining useful life and a recommended maintenance action.
Crawlers, quadrupeds and drones performing routine visual, thermal and gas-detection inspection rounds that would otherwise require a technician in the field.
Camera- and vision-model-based systems that automatically detect corrosion, leaks, mechanical wear or missed safety steps from continuous video or periodic imagery, replacing purely manual visual inspection rounds.
Smart-glasses and tablet-based augmented reality that lets a remote specialist see what a field technician sees and annotate the live view.
Cryptographic device identity, secure boot and signed firmware built into field controllers and smart final-control elements themselves, not bolted on at the network layer.
Automated execution and logging of safety final-element proof tests (partial-stroke testing, full-stroke testing) without manual technician intervention.
Powered or passive exoskeletons and biometric wearables supporting field technicians during manual valve operation, lifting and confined-space work.
Field-level networks that automatically detect a failed link or node and reroute traffic without engineering intervention, extending mesh-networking concepts to wired Ethernet-APL segments.
Field-mounted I/O and junction boxes that terminate Ethernet-APL directly, pushing the marshalling and protocol-conversion layer out of the control room and into the field.
Motor control centres and motor-driven final elements with embedded machine-learning models trending vibration, current signature and temperature for early fault detection.
Control Systems — DCS · PLC · ESD · HMI
The DCS, PLC, safety and HMI/SCADA layer — where control logic executes, and where AI is arriving fastest.
Control logic execution decoupled from proprietary controller hardware and run instead as software on standard, virtualised compute — scaled, patched and failed-over like any other enterprise application.
AI agents that don't just recommend a setpoint change but are given standing authority to execute a bounded set of control actions autonomously.
SCADA and HMI platforms delivered and licensed as a cloud service, with the visualization and historian layer hosted off-premises rather than on a local server.
Generative-AI assistants embedded in engineering tools that draft PLC/DCS logic, translate P&IDs into control narratives, or answer natural-language questions about an existing configuration.
A single, event-driven data model that every system — DCS, MES, historian, analytics — publishes to and subscribes from, replacing point-to-point integrations between each pair of systems.
Time-Sensitive Networking standards combined with OPC UA FX give standard Ethernet the deterministic, low-jitter behaviour that motion and safety control traffic requires.
Machine-learning-based advanced process control that continuously retrains against live plant data, rather than the fixed linear models of first-generation APC.
Graphical, template-driven configuration environments intended to let process and automation engineers configure standard control strategies without hand-writing IEC 61131-3 logic.
An operator interface that answers plain-language questions ('why is the reactor trending high?') by querying live process data, alarms and procedures, rather than requiring navigation through faceplates and trend screens.
AI assistants that search across historian data, alarm history, maintenance records and equipment manuals to propose a likely root cause for an abnormal situation — distinct from a general conversational HMI in that they're built specifically for diagnostic reasoning, not open-ended Q&A.
Virtual and augmented-reality environments built on a plant's digital twin, used to train operators on startup, shutdown and abnormal-situation procedures without touching the live process.
Analytics that continuously reassess a safety instrumented system's real-time risk reduction — factoring in bypasses, degraded redundancy and process conditions — rather than relying solely on the static SIL calculated at design time.
Industrial IT / OT / Critical Control Information Infrastructure
The networks, security architectures and information infrastructure that connect and protect the control environment.
A single security operations function with visibility and defined authority across both IT and OT estates, replacing the historical split between a corporate SOC and a plant that security 'doesn't reach'.
Continuous verification of every device, user and connection inside the OT network, rather than trusting anything already inside the traditional Purdue-model perimeter.
A composable security architecture where distributed OT security tools (detection, identity, policy) interoperate through common analytics and policy layers, instead of each product operating in isolation.
Passive network monitoring that baselines normal OT traffic and process behaviour, then uses machine learning to flag deviations that signature-based tools would miss.
AI agents that triage, investigate and in some cases respond to security alerts autonomously, extending the AI-SOC pattern rapidly emerging in enterprise IT security into the OT security operations centre.
A machine-learning approach where models train across multiple plants' data without that data ever leaving each site, then share only the improved model — letting an operator with many similar sites benefit from fleet-wide learning without centralising sensitive process data.
A machine-readable Software Bill of Materials for every control-system component, listing every third-party and open-source library it contains, so a newly disclosed vulnerability can be matched to affected assets in hours rather than weeks.
Cryptographic algorithms designed to resist attack from a future cryptographically-relevant quantum computer, replacing the RSA/ECC-based schemes protecting today's OT communications and remote access.
Dedicated, site-owned 5G networks providing high-bandwidth, low-latency, securable wireless connectivity across a plant, port or mine site, independent of public carrier infrastructure.
Low-earth-orbit satellite broadband providing low-latency connectivity to remote wellheads, pipelines, mines and substations that terrestrial networks never reached economically.
The successor wireless generation to 5G, targeted at sub-millisecond latency and native sensing/positioning capability — early standards work is only just underway.
A live, continuously-synchronised virtual model not just of the process, but of the control network itself — every controller, switch, firewall rule and data flow — for security testing, capacity planning and incident response rehearsal.
Software that continuously assesses an OT environment against IEC 62443 zones-and-conduits requirements, NIS2 and equivalent regulation, rather than relying on a periodic manual audit.
Distributed-ledger records intended to give an immutable, tamper-evident history of a control-system component's provenance, firmware updates and maintenance history.
Ruggedised, industrially-rated compute platforms running AI inference physically close to the process — in the control room or on the plant floor — rather than in a distant cloud region.
No technologies match your filters. Try clearing the search or selecting more categories.
Frequently asked questions
What is the ControlCurve maturity map, and why apply one to industrial control and instrumentation?
It is a map of how far each technology has travelled from idea to everyday use, across five stages: Emerging Idea, Over-Promise, Reality Check, Industry Acceptance and Proven in Use. Each entry also carries an evidence level from E1 (concept) to E5 (routine specification). Industrial plants run for decades and safety-related systems change slowly, so engineering, operations and technology leaders need to separate genuinely maturing capability from supplier-driven attention before committing capital or headcount.
How reliable are technology maturity predictions for industrial automation?
No maturity model predicts the future with certainty. This analysis is built from public vendor announcements, standards-body activity (IEC, ISA, OPC Foundation, FieldComm Group and others), trade and analyst reporting, and observed deployment counts as of the October 2026 edition, and it is reviewed periodically as new evidence emerges. Treat each placement as an informed, sourced judgment call — and weight it alongside your own site-specific constraints, risk appetite and asset age.
How is the position of each technology on the curve actually determined?
Two things are edited judgments and one is purely illustrative. The horizontal stage (Emerging Idea through Proven in Use) and the stated time-to-proven-use horizon are both editorial assessments, weighed from vendor announcements, standards-body milestones, analyst commentary and observed deployment counts. The curve's height at any given point is a reference shape drawn once for visual orientation — it is not a per-technology score, and two technologies at the same stage aren't being ranked against each other by how high their marker sits. See how this chart is built for the full breakdown.
Where does AI fit into this maturity map?
AI and machine learning aren't treated as a single, separate category here, because in practice they cut across all four: edge-AI transmitters and soft sensors in field instrumentation, computer-vision inspection in field control devices, agentic process control and diagnostic copilots in control systems, and AI-driven OT anomaly detection in the IT/OT/CCII section. 17 of the 50 tracked technologies are AI/ML-driven as of this edition — each carries a dashed accent ring on the chart and an "AI/ML" tag, and the whole set can be isolated with the AI & ML cross-section toggle above the chart.
What is the difference between OT and IT in an industrial cybersecurity context?
IT (information technology) covers the systems that manage data — email, ERP, corporate networks. OT (operational technology) covers the systems that directly monitor and control physical processes and equipment — DCS, PLC, SCADA and safety instrumented systems. IT/OT convergence describes the growing integration of these two historically separate domains: it delivers real efficiency and visibility benefits, but it also extends the cyberattack surface into physical operations, which is why it anchors its own section of this Hype Cycle.
Which of these technologies should a plant prioritise first?
There is no universal answer, but a useful rule of thumb is to prioritise items at the Industry Acceptance stage with a near-term time horizon, where reference deployments already exist and risk is comparatively low, ahead of committing significant capital to anything still at the Emerging Idea stage. Field-level connectivity upgrades (Ethernet-APL) and an OT-aware cybersecurity architecture are the two areas most industrial sites cannot indefinitely defer.
Where did the idea for the curve come from?
The pattern of early enthusiasm, disappointment and gradual adoption has been described by many analysts, most famously in the hype-cycle concept popularised by Gartner, Inc. ControlCurve uses its own stages and evidence scale, built for industrial automation, and is independent of and not affiliated with Gartner.
Methodology & sources
Scope
Technologies with a plausible, material impact on process- and discrete-industry control, automation and instrumentation over the next 1–15 years, grouped into field instrumentation, field control devices, DCS/PLC/ESD/HMI control systems, and industrial IT/OT/critical control information infrastructure.
Placement
Each technology's stage and time-to-proven-use reflects publicly available evidence: vendor product announcements, standards-body milestones, analyst commentary, trade press and documented deployment counts, weighed by the editor. Where sources conflict or a figure comes only from a supplier, the full analyses say so.
Review cycle
Revisited at least twice a year. Placements move as evidence changes — a technology can advance, stall, or occasionally reverse; the "last updated" date at the top of this page is authoritative.
Reading the curve
Stage, time horizon and evidence level are our editorial assessments. The curve's vertical height is a reference shape drawn once for visual orientation only — it is not a per-technology score. See how this chart is built.
The AI & machine learning cross-section
AI/ML is deliberately not a fifth category. It's tagged as a cross-cutting theme on any entry whose core mechanism is AI- or ML-driven, so it stays filed under its natural category (field instrumentation, field control devices, control systems, or IT/OT/CCII) while remaining separately explorable. As of this edition, 17 of the 50 tracked technologies carry the AI/ML tag, spanning all four categories — visible on the chart as a dashed accent ring and in the directory as an "AI/ML" tag, and isolable with the toggle above the chart.
Sources cited in the full analyses, by publisher
- 5G-ACIA: Integration of 5G with Time-Sensitive Networking for industrial communications
- ABB: BASF selects ABB for greenfield projects after a successful Ethernet-APL trial
- ALL4: U.S. EPA finalizes oil and gas compliance deadline extension
- ANSI Webstore: IEC 62591:2016 Industrial networks – Wireless communication network and communication profiles – WirelessHART
- ANYbotics: ANYmal X: BASF Ludwigshafen deployment; ANYmal X
- AP Sensing: Power cable monitoring solution to French electricity transmission system operator
- ARC Advisory Group: Yokogawa's autonomous control AI adopted for use at an ENEOS Materials chemical plant; Will intelligent modulating electric actuators eventually dominate the control valve actuation market?; From spec to site: where Ethernet-APL adoption stands in 2026; 3GPP finalized Release 16 for 5G New Radio technology; Sparkplug now an official ISO/IEC standard
- AVEVA: AVEVA successfully validates its SCADA and MES software on AWS infrastructure
- Business Wire: Mitsubishi Electric completes full acquisition of Nozomi Networks
- CISA: IRGC-affiliated cyber actors exploit PLCs in multiple sectors (AA23-335A); 2025 Minimum Elements for a Software Bill of Materials (draft)
- Control Engineering: New insights: 100-controller ExxonMobil Open Process Automation
- Control Global: What you should know to adequately test automated valves; IO-Link Wireless designated as IEC 61139-3 standard; Plug-and-play punches in; Innovations sparkle at Emerson Exchange 2025; WirelessHART vs. ISA100.11a: what's the difference?
- Dragos: Dragos 2026 Year in Review: new OT threats and ransomware
- Eclipse Foundation: The Eclipse Foundation announces Sparkplug as an international standard for a plug-and-play Industrial IoT
- eLynx Technologies: Setting up SCADA with Starlink: a guide for oil and gas operators
- Emerson: Emerson enhances control system software to increase operational agility
- ExecutiveGov: CISA, federal partners release zero trust guide for operational technology
- Help Net Security: CyberAv3ngers hit Unitronics PLCs at multiple US-based water facilities
- HiveMQ: IIoT milestone as Sparkplug becomes international standard
- Hughes: LEO satellites and IoT are transforming oilfields: smarter, safer
- Hydrocarbon Processing: Imubit launches closed loop AI optimization solution powered by reinforcement learning
- Inductive Automation: Ignition Cloud Edition now available in AWS Marketplace
- Inspenet: ExR-2: the autonomous inspection robot for the petrochemical industry
- IO-Link Community / PI: IO-Link continues growth in automation
- ISA: ISA-TR96.05.01-2017, Partial stroke testing of automated valves
- ISA Kuwait Section newsletter (Emerson reprint): Use of digital technology to test shutdown valves for SIS applications
- ISPreview: SpaceX show latency improvements on Starlink broadband network
- Journal of Petroleum Technology: A practical distributed acoustic sensing approach for gas-pipeline-leak detection
- Materialise: Shell robotic inspection for the global energy industry (ExR-1 case study)
- Mend.io: EU Cyber Resilience Act compliance guide
- NAMUR: NE 107 has been revised
- NIST: NIST publishes SP 800-82 Revision 3; SP 800-207, Zero Trust Architecture
- PI (PROFIBUS & PROFINET International): IO-Link Wireless is now an international IEC standard
- Pipeline Technology Conference: DFOS: what type of fiber optic sensing is most appropriate for leak detection on my pipeline?
- Plant Services: NAMUR NE 107 recommendations come to the United States
- Processing Magazine: Status update: ISA100 & WirelessHART
- RCR Wireless News: Three more private 5G deployments; From a standing start in Texas: the story of Dow Chemical's multi-site private LTE push
- SANS Institute: SANS report warns of rising gap between fast detection and slow recovery in critical infrastructure
- SC Media: CISA releases draft changes to SBOM minimum requirements for comment
- Siemens: Factory automation: virtual PLC at Audi
- TechTarget: EU Cyber Resilience Act reporting: what CISOs need to know
- Valve Magazine: EPA's final rule on methane and greenhouse gas emissions for actuators; New valve controller from Emerson offers embedded edge computing; FieldComm Group to host webinar detailing WirelessHART deployment guidelines
- Virtualization Review: Federal zero trust guide puts OT security under IT scrutiny
- Water Online: Ethernet-APL proves its value: expanded testing shows broad industry adoption
- Wikipedia: Advanced Physical Layer (overview of 10BASE-T1L, segment lengths and 2-WISE)
- World Oil: No fiber, no problem: what commercial satellite connectivity actually looks like at a remote job site
- Yokogawa: In a world first, Yokogawa's autonomous control AI is officially adopted for use at an ENEOS Materials chemical plant
Get the quarterly briefing
A short note when we re-stage a technology, add a new entry, or a standard/regulation moves — four times a year, nothing else. Email us and we will add you to the list; reply at any time to be removed.