The Tracker Is a Network Contract With a Case Around It

Every GPS tracker sold since the late 1990s has been a bet on a mobile network staying switched on. For most of the installed base in the United Kingdom that bet now has an end date, and it is published by the regulator rather than by the customer.

Ofcom’s guidance to IoT and third-party device suppliers sets out the position without ambiguity. The UK’s 3G networks are already switched off. The 2G networks follow on operator-specific timetables: EE begins switching off from May 2029, O2 from summer 2029, and VodafoneThree during 2030. A voluntary 2G switch-off charter was published by the government in March 2026. And a preview of how this works in practice arrived early — O2 withdrew 2G access for inbound roaming from 1 October 2025, which took out any 2G device relying on a foreign roaming SIM more than three years before the national sunset.

Ofcom’s operative sentence is the one worth reading twice: all 2G-only devices will need to be upgraded to at least 4G, and a provider of a service that relies on 2G is responsible for ensuring continuity after the network goes. The regulator has just converted a network decision into a supplier liability with a date attached.

What a sunset does to a hardware market

Asset tracking is one of the larger populations of 2G and 3G modems in commercial use, alongside telecare alarms, security and fire alarms, ATMs and payment terminals — the same radio modules in different boxes. A sunset forces a replacement cycle that has nothing to do with the customer’s preferences or the vendor’s roadmap. It is a deadline.

Deadlines do something specific to a hardware market. They compress demand into a window, they push buyers to the cheapest compliant device rather than the best one, and they arrive with a deflationary kick: LTE Cat-M and NB-IoT modules cost less than the 3G units they replace and consume less data, so the connectivity line on the customer invoice falls at the same moment the device line spikes.

The result is a market where unit shipments can rise sharply while revenue per device falls. Anyone measuring this category by boxes shipped is measuring the deadline, not the business.

The second clock is documentary

The clock that actually determines whether a tracking business is worth owning is not the radio. It is eFTI — the EU regulation on electronic freight transport information, Regulation (EU) 2020/1056.

Its implementation has been slow, deliberate and is now scheduled. The first implementing and delegated acts entered into force on 9 January 2025, specifying how the IT systems should be designed and which data is exchanged. From January 2026 certified eFTI platforms and service providers can begin preparing for operations and member-state authorities may start accepting data held on them. Full application is 9 July 2027, when authorities in every member state must accept information shared electronically by operators through certified eFTI platforms.

The architecture matters more than the date. Business data sits on certified platforms; it is shared with an authority only on an explicit inspection request, through a unique access link in a machine-readable format such as a QR code; and operators can grant selective access to business partners. That is a public, certified, standardised exchange for consignment information.

It is not a competitor to telematics. It is a reclassification of what telematics is for. A tracking platform’s commercial value has rested on the private evidence it produces — journey, hours of service, fuel, temperature, arrival. If consignment evidence becomes a certified public standard, the defensible layer moves from capturing the data to certifying it, and certification is a regulatory status rather than a product feature. A status cannot be sold as a differentiator, and it cannot be defended by adding another dashboard.

The forecast, read properly

Pheonix Research’s fleet management forecast runs from USD 25.4 billion in 2025 to USD 58.7 billion in 2033 at a stated ~11.2 per cent. The endpoints compound at 11.04 per cent — a 16 basis point gap between the published rate and the arithmetic, small enough to be a rounding choice in the original model but worth recording, because it is the kind of discrepancy that accumulates when a number is requoted without being recomputed.

The more useful observation is internal to the report. Its component structure is software platforms, telematics and GPS hardware, and value-added services. Around 11 per cent compound growth across a decade is achievable for platforms and services. It is not achievable for a radio module being replaced by a cheaper radio module under a regulatory deadline.

Which means the 11 per cent has to be earned in the layer that does something with the data: cost allocation, driver behaviour, maintenance authorisation, and increasingly the compliance workflow that surrounds eFTI and cross-border tax evidence. That is the layer with recurring revenue and switching costs. The hardware is a refresh cycle wearing a growth multiple.

What to measure instead of units shipped

Three numbers distinguish a platform business from a re-equipping one. The share of revenue that is recurring rather than device-attached. The renewal rate of connectivity contracts, since a sunset forces renegotiation and reveals whether the customer was buying the vendor or the SIM. And revenue per active device across the refresh, because a generation of hardware that costs less to run should show up as a better service margin than the one it replaced, or it should not be called a platform.

We do not quote a global market size for GPS tracking systems here. The perimeter merges four different purchases — the device, the connectivity, the platform and the managed service — bought from different suppliers at different margins, and a single total would tell a supplier nothing about which of those four they are actually competing in.

The useful framing is simpler. The box is not the product. The network contract is the product, the deadline is the sales cycle, and after July 2027 the data it produces is a public standard. Everything else is a case.