Why the category line on the datasheet matters
When a proposal says a device "supports 4G", it has told you almost nothing. LTE was designed with device categories — profiles that trade speed against cost, power and complexity — so that a CCTV router and a water-meter radio can share one network without sharing one bill of materials. The category determines the module price (a spread of several times between Cat-4 and Cat-1bis), the antenna count, the power draw that sets your solar and battery sizing, and whether the device works on South African networks at all. Two devices with identical spec-sheet marketing can differ in exactly the ways a remote site discovers six months in.
The family, one table
| Cat-4 | Cat-1 | Cat-1bis | Cat-M1 (LTE-M) | NB-IoT (Cat-NB) | |
|---|---|---|---|---|---|
| Peak speed (DL/UL) | 150 / 50 Mbps | 10 / 5 Mbps | 10 / 5 Mbps | ~0.3–1 Mbps | Tens of kbps |
| Bandwidth used | 20 MHz | 20 MHz | 20 MHz | 1.4 MHz | 200 kHz |
| Antennas | 2 | 2 | 1 — smaller, cheaper devices | 1 | 1 |
| Deep-sleep features (PSM/eDRX) | No | No | No | Yes | Yes |
| Mobility / handover | Full | Full | Full | Yes | Poor — fixed devices only |
| Module cost tier | Highest | Low | Lowest of the LTE set | Low | Lowest |
| Needs special carrier deployment? | No — any LTE network | No — any LTE network | No — any LTE network | Yes | Yes |
| South Africa reality (2026) | Works everywhere | Works everywhere | Works everywhere | No committed deployment | One carrier, verify per site |
The South African twist: the "IoT categories" are the ones you mostly can't use
Globally, Cat-M1 and NB-IoT were designed as the low-power IoT endgame — deep sleep modes, extended coverage, minimal modules. But both need the mobile operator to deploy and maintain specific network features, and that is where South Africa diverges from the international datasheet: LTE-M has no committed local deployment, and NB-IoT remains a single-carrier network to be verified site by site. Meanwhile Cat-1 and Cat-1bis need nothing special at all — they are ordinary LTE devices, attached to networks that already cover the country, on modules that now undercut what Cat-M1 hardware costs in practice. This is why international reference designs arrive here specifying Cat-M1 and have to be re-specified: the elegant global answer is the wrong local one. (The same buying rule applies downward: since new 2G/3G-only device activations were barred from the end of 2024, the legacy fallback is closed too — details in our network comparison.)
The gap Cat-M1 was meant to fill — battery-powered sensors that sleep for years — does not go unfilled in South Africa. It is exactly the role LoRaWAN plays in our deployments: coverage you own, no carrier feature dependency, and battery life the cellular categories only approach in their deep-sleep modes.