Industrial Ethernet accounted for 79% of newly installed nodes worldwide in the 2026 study, up from 76% in 2025 [S1][S2].
The HMS Networks 2026 analysis covers newly installed nodes rather than the complete installed base, and its leading Ethernet protocols were PROFINET at 30%, EtherNet/IP at 25%, and EtherCAT at 20% [S1][S2].
What the 79% New-Node Share Means
Industrial Ethernet accounted for 79% of newly installed nodes worldwide in the 2026 study, compared with 76% in 2025 [S1][S2]. Industrial Ethernet's share of newly installed nodes increased by 3 percentage points, from 76% in 2025 to 79% in the 2026 study [S1].
This statistic must not be read as the share of all operational controllers, instruments, or switches. It describes new node installations, so a large installed base of PROFIBUS DP, Modbus RTU, DeviceNet, or other fieldbus equipment can remain fully relevant to expansion and replacement decisions [S4]. HMS had tracked the series for 12 consecutive years by its 2026 analysis [S1].
Protocol-Level Comparison for New Projects
The three leading Ethernet protocols continued to lead in the 2026 analysis [S2].
PROFINET is the broad default where European machine builders and mixed PLC, I/O, and drive requirements dominate. EtherNet/IP is the stronger North American fit for Rockwell-oriented PLC and drive architectures. EtherCAT is the motion-focused option when synchronized drives, low-latency cyclic control, and EtherCAT devices dominate the bill of materials [S1][S4]. Modbus TCP fits simple supervisory or Modbus-compatible integration, although its deterministic performance is less suited to demanding motion control [S4].
Where Fieldbus Remains the Better Choice
Fieldbus is appropriate when the expansion boundary contains controllers, transmitters, remote I/O, drives, or sensors that already support PROFIBUS DP, Modbus RTU, DeviceNet, or CANopen and replacing them creates unnecessary cost or downtime. Modbus RTU is also appropriate for long RS-485 links, simple master-slave arrangements, low-cost remote devices, and installations where 24 VDC power and data can share suitable fieldbus wiring [S4].
Low-cost AS-i can carry data and 24 VDC power over unshielded two-wire cable, making it useful for dense low-level sensor and actuator networks [S4]. Do not select a fieldbus merely to avoid Ethernet switches: the device ecosystem, electrical distance, diagnostic requirements, and replacement inventory must justify the choice.
Migration Options by Project Condition
A brownfield expansion should preserve the existing fieldbus trunk where compatible devices remain, then use gateways only at genuine protocol boundaries. The decision should be based on the number of unsupported devices, engineering-tool licenses, spare-part availability, and required diagnostic coverage rather than on the 79% market statistic alone. [S4]
A greenfield machine with standard PLC, I/O, servo, HMI, and OPC UA or enterprise integration is a strong Industrial Ethernet candidate, with protocol selection driven by the controller family and device ecosystem. For deterministic motion, EtherCAT is the specific high-performance choice; for broad mixed automation, PROFINET or EtherNet/IP will usually have greater practical alignment. The industrial router and industrial switch architecture must then match the selected protocol rather than being chosen as a generic Ethernet layer.
For a plant with separate legacy and modern sections, a staged migration limits operational disruption: retain qualified fieldbus equipment, isolate traffic at the boundary, and replace fieldbus devices as their operating condition and supportability justify it. A fieldbus gateway can bridge selected traffic, but it adds configuration, diagnostic and cybersecurity responsibilities that do not exist in a native Ethernet cell.
Wireless, Process Automation and Boundaries
The 79% figure is primarily a factory automation result, not permission to replace every process fieldbus. SPE, single-pair Ethernet, and Advanced Physical Layer activity is relevant to process and sensor applications, while established fieldbus can remain justified where equipment, distance, power delivery, or installed-base constraints dominate [S1][S4].
Industrial Ethernet protocols cannot be treated as interchangeable. PROFINET, EtherNet/IP, EtherCAT, Modbus TCP, and CC-Link IE use different device objects, commissioning tools, discovery behavior, and performance mechanisms. Select the protocol family from the controller and device requirement, then verify cycle time, update rate, redundancy behavior, topology, connector, and environmental ratings for the actual cell.
Practical Specification Checklist
Specify the network by application, not by headline share. For every proposed segment, record the controller model, device list, protocol variant, cyclic update requirement, maximum cable distance, power method, redundancy method, diagnostic method, environmental exposure, and engineering-tool license [S1][S4].
Use PROFINET at 30% for projects aligned with its 30% Ethernet share, EtherNet/IP at 25% for Rockwell-centered North American estates, and EtherCAT at 20% for motion-intensive cells where sub-100-microsecond cycle performance is supported by the selected architecture [S1][S4]. Reserve Modbus TCP for simple Ethernet-connected Modbus equipment and select Modbus RTU only where its serial electrical and master-slave characteristics fit the physical layer.
The HMS study expects average annual growth of newly installed nodes to be approximately 7.7% over the following five years, with fieldbus migration contributing to expansion [S1]. Track two project-level signals rather than treating that forecast as an automatic specification rule: the percentage of new devices lacking the incumbent protocol and the annual cost of maintaining proprietary gateways, spares, and fieldbus expertise.
Use the 79% share to set the default for greenfield Ethernet selection, then use device compatibility, motion requirements, physical-layer constraints, lifecycle cost, and migration effort to justify every fieldbus exception [S1][S2][S4].