For most of its existence, the IoT device has had one job: collect data and send it somewhere else to be understood. That model is under pressure from two directions at once, and the response from the industry is becoming hard to ignore. Edge AI IoT devices–ones that process and act on data locally rather than routing it to the cloud–are moving from pilot programmes into mainstream product portfolios in 2026. The timing is not accidental.
Cloud-dependent IoT has a cost problem that is getting worse. The global memory shortage, driven by AI data centres consuming an unprecedented share of DRAM and NAND production, has pushed component prices to levels that are reshaping device economics across the board.
IDC has described the reallocation of silicon wafer capacity toward high-bandwidth memory for AI infrastructure as structural, not cyclical, with effects expected to persist well into 2027. For IoT OEMs, that means building products that make more calls to cloud infrastructure is becoming more expensive at exactly the wrong time.
A device that can reason locally, reduce cloud dependency, and operate on a leaner memory footprint is no longer a premium proposition. It is a cost management strategy. There is a second pressure that is less about cost and more about what IoT products can credibly charge for.
As more enterprise buyers expect recurring value from connected devices, OEMs have been moving toward subscription-based models where ongoing intelligence justifies the fee. A sensor that sends raw data is a hardware commodity. A device that detects anomalies, flags maintenance needs, or makes operational decisions locally is a different product category with different pricing power.
Edge AI is what makes that transition possible at scale.
The market is voting with its product roadmaps
IoT Analytics called 2026 the inflexion point for this shift in its semiconductor predictions, noting that OEMs would move from early pilots to broad portfolio refreshes marketed as edge AI-enabled devices.
That prediction is now showing up in what companies are actually shipping. MediaTek debuted its Genio platform for smart retail at NRF 2026 in January, built around on-device generative AI for point-of-sale and inventory systems with no cloud requirement.
At Embedded World this week in Nuremberg, SECO unveiled a new system-on-module based on MediaTek’s value-tier Genio 360 processor–specifically positioned for cost-sensitive embedded applications where local AI inference needs to be affordable, not just possible.
The global edge AI market was valued at US$24.91 billion in 2025 and is projected to reach US$118.69 billion by 2033, growing at a CAGR of 21.7%, according to Grand View Research.
Perhaps the clearest signal of where the market is heading came in February, when Texas Instruments announced its acquisition of Silicon Labs, whose Series 3 IoT platform delivers a tenfold improvement in processing performance over its predecessor and is designed specifically for intelligent edge devices, including wireless gateways, cameras, and wearables.
TI’s intent, according to industry analysts at Futurum, is to manufacture these chips at scale on its own 300mm wafers to bring down per-unit cost. When a company of TI’s scale acquires an edge AI IoT platform and immediately focuses on making it cheaper to produce, it is not placing a long-term bet. It is responding to demand it can already see.
The complexity that comes with it
The shift is real, but it does not arrive without friction. Moving intelligence onto the device solves the cloud dependency problem and creates a different one: how to deploy, update, and monitor AI models running across large, heterogeneous fleets of devices in the field, many of which have limited connectivity and no physical access.
Edge Impulse, which exhibited at IoT Tech Expo in London in February, has built its platform around exactly this challenge, enabling AI inference across device types without bespoke integration for each hardware variant. It is a meaningful problem, and the maturity of the software ecosystem around it is still catching up to the hardware.
Not every IoT application needs on-device inference, and the case for edge AI is stronger in some verticals than others. Industrial predictive maintenance, smart retail, and healthcare monitoring are well ahead of smart metering or basic environmental sensing. But the direction of travel across the industry is clear, and the economic forces accelerating it are not going away.
The memory crisis has compressed what might have been a gradual transition into something more urgent. For enterprise buyers and IoT product teams alike, the question is shifting from whether edge AI belongs in the roadmap to how quickly a credible version of it can be in production.
The IoT device that just sends data to the cloud is starting to look like the dumb terminal of this decade. The one that thinks for itself is already in production.
See also: Designing industrial IoT around measurable ROI
Want to learn more about the IoT from industry leaders? Check out IoT Tech Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including AI & Big Data Expo and the Cyber Security Expo. Click here for more information.
IoT News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
