Announcements, releases and milestones from across the platform.

A 2026 survey finds 68% stuck in pilots and proofs of concept, with legacy integration and data silos named the largest barrier by nearly half of respondents.

On real production time series a widely used unsupervised detector caught 4 of 22 anomalous processes at best, while a temporal convolutional autoencoder reached 0.99 — a result about architecture, not model size.

The IIS3DWB10IS reaches a 35 µg/√Hz noise floor above 10 kHz with 200 g dynamic range, and runs inference inside the sensor at 40 MIPS.

The RM255C-GL and RG255C-NA cover the global M.2 and North American LGA form factors, with an extended temperature option and fallback to LTE where RedCap networks are not yet live.

The reduced-capability profile has moved past trials: operators in 27 countries are investing, fourteen have launched commercially, and the testing list spans another dozen markets.

Two new radio units — one for the European 3.8 to 4.2 GHz local-industrial band and a US CBRS unit — take the system to 15 countries, with American availability in summer 2026.

Portwell has built Core Ultra 200S Series processors into industrial edge platforms aimed at AI inference, part of a broader move to put usable model capacity in cabinet-mounted hardware.

The case being made is convergence rather than speed: one physical network carrying control traffic, video and analytics without the first being disturbed by the others.

Switches from Cisco, Belden, Moxa and Siemens are in the field and OPC UA PubSub over TSN has reached production pilots, but scheduling is still configured through each vendor's own engineering tool.

Analysts describe a shift from training scale to inference placement, on the argument that the data that most needs a fast decision is generated at factories, retail sites and remote installations.

The ProLinx Edge Gateway runs compact inference models, digital twin connectivity agents and analytics containers alongside protocol translation in a single device rated for harsh environments.

Analysts now track scheduling and orchestration software for mobile robots separately from the robots, a sign that the layer has become a purchasing decision rather than a bundled feature.
Tell us what you're building and we'll help you scope the first deployment.