Summary

  • Orange’s published plan puts the final metropolitan 2G wave on 20 October, but makes a later 1–15 December exception for regulated white zones. Arcep’s broader timetable runs from late September to late December.
  • At June 2026, Arcep counted 2.01 million M2M SIMs in 2G/3G-only terminals across four operators. That is a national combined-technology count, not Orange’s remaining 2G customer base.

For a mobile subscriber, a network sunset can look like a date on a calendar. For an alarm panel, lift emergency phone, payment terminal or remote controller, it is a dependency hidden inside equipment that may have been installed years ago. Orange’s final metropolitan 2G wave makes that distinction operational: the signal can be scheduled off while the inventory of services that depend on it remains incomplete.

Orange’s current timetable lists a staged closure: Nice on 22 September 2026, 44 departments on 6 October and the remaining 43 on 20 October. It describes the metropolitan network as then fully stopped, except in regulated white zones, where Orange customers will lose 2G between 1 and 15 December. Arcep’s consumer guide, updated on 10 September, summarizes Orange’s mainland phase as running from late September through late December and says the calendar can change. Those descriptions need not mean the same endpoint is switched on one nationwide date: the operator’s local schedule and the regulatory exception are separate controls to check.

The scale of the endpoint problem is visible in Arcep’s latest published inventory, but the denominator matters. At the end of June, subscribers of Bouygues Telecom, Free Mobile, Orange and SFR had 2.01 million SIMs in machine-to-machine terminals compatible only with 2G or 3G, equal to 7.7% of the active SIM fleet in that device class. The observatory excludes SIMs built into vehicles by vehicle manufacturers. A separate 1.91 million SIMs sat in 2G/3G-only terminals for voice, SMS or mobile internet. These are France-wide counts across four operators, combine 2G-only with 2G/3G equipment, and do not reveal Orange’s share.

They show why a mobile handset count is an incomplete proxy, not how many Orange devices will fail.

Arcep lists alarms and remote surveillance, personal teleassistance, heating controls, connected intercoms and video-phones, and some medical devices among systems that may still use 2G or 3G. Orange adds payment terminals, lifts, meters, sensors, vehicle geolocation and industrial equipment to its examples. The device owner may not know a SIM is present, and an installer or service company may control the equipment while another provider holds the customer contract. A cellular shutdown therefore tests an installed service chain, not just handset replacement.

The replacement standard is also use-specific. Orange tells business customers to move 2G-only voice devices to 4G with VoLTE, while its guidance points IoT data and SMS modules toward 4G LTE-M. A device marked ‘4G’ is not, by itself, proof that a voice call, alarm signal or remote status message works on the new path. Arcep likewise says the 4G layer covers more than 99.9% of the metropolitan population and that more than 99.8% of 2G/3G radio sites are equipped with 4G; those network facts do not verify a particular module, SIM profile, indoor installation or service workflow.

For services bought from a retail operator that uses a host network, the notice path adds another handoff. Orange Wholesale says the retail operator is responsible for informing its own customers about the host network’s 2G/3G shutdown and what it means for them. That allocates communication responsibility, but a notice is not an endpoint inventory or a successful test. The service provider, equipment vendor, site owner and network host can each hold a different part of the evidence.

The useful readiness receipt is therefore more specific than ‘customer informed’. It joins each affected SIM to a device model and location, identifies whether the traffic is voice or data, records the selected replacement technology, and documents a real test of the service after migration. For a telealarm, that means an end-to-end alarm event is received; for payment or lift equipment, the relevant transaction or emergency call succeeds. Public sources do not disclose Orange’s device-level inventory or migration test results, so neither the national Arcep count nor the published shutdown date can stand in for that proof.

Sources