S01.11 · Technology & Digital Infrastructure
Consumer electronics, industrial equipment and connected IoT devices spanning roughly $1T in consumer spend plus 21B+ connected units.
Hardware, Devices & IoT spans consumer electronics, industrial equipment and connected devices. There is no single defensible market number, because the pieces are measured on different bases: consumer electronics at roughly $1.0-1.1T for 2025 (Precedence Research), the connected-device base in units — 21.1B devices, up 14% in 2025 (IoT Analytics). Keeping them separate matters more than blending them, because the line that now divides winners from commodity hardware runs through silicon: devices that can run AI inference locally, and devices that cannot.
Consumer electronics has grown ~8-10% historically; connected-device units are growing 14% year over year; forward growth across the segment sits in the mid-to-high single digits, with AI-capable devices shifting mix toward higher-value units even where unit growth is modest — a price and mix story, not a volume one. Demand is global across the US, China and Europe; production concentrates in China (Foxconn, Pegatron) and is diversifying into Vietnam and India as manufacturers hedge the concentration.
Apple and Samsung dominate consumer devices; industrial IoT stays fragmented across thousands of niche integrators — one concentrated layer and one fragmented one inside the same category, which is why no single comp set covers it. Gross margins run 20-40%, well below software, on manufacturing capital intensity. Revenue is mostly transactional — a unit sale — with a services and subscription attach growing on top, and the attach is where the margin story lives. Upstream: semiconductor components and contract electronics manufacturers. Downstream: OEM brands, retail and distribution channels, industrial systems integrators. Regulatory gating centers on safety certification and, increasingly, tariffs — a live variable for a production base this concentrated in China.
Device makers are pushing into embedded software, services and telemetry-based data monetization. The services attach is the real economics: it layers recurring margin over hardware pricing that is flat to declining. Supply-chain diversification is protection against tariff exposure, nothing more.
Agentic AI automates device-level diagnostics and predictive maintenance, compressing the field-service cost line that has followed hardware after it ships. The larger effect is on differentiation itself. As on-device AI capability standardizes, the feature sets manufacturers used to compete on converge, and competition moves up the stack to the software and services layered on increasingly similar hardware — the classic commoditization sequence, arriving on schedule.
The new boundary inside the segment is edge inference silicon: connectivity-only devices on one side, devices that run inference locally on the other. The segment's own horizon for mass agentic adoption is two to five years, and the gate is the cost curve of edge AI silicon, not demand — the demand is already there.