If you are picking an LPWAN network in 2026, it almost always comes down to two names. LoRaWAN has crossed 125 million connected devices globally. NB-IoT powers around 400 million cellular IoT endpoints, most of them inside China. Outside China, LoRaWAN still leads the LPWAN market with roughly 40 percent share. And yet, most product teams we talk to still pick the wrong one for their use case. The decision usually gets made on a single axis (cost, or coverage, or battery life) instead of the five that actually matter together. This guide walks through those five axes with real 2026 numbers and ends with a quick decision matrix you can apply in an afternoon. If you want a wider view of how these fit alongside MQTT, BLE, and cellular, our IoT communication protocols guide is a good companion read.
The two technologies in one paragraph each
LoRaWAN is an unlicensed sub-GHz LPWAN. In India it runs at 865 to 867 MHz, in Europe at 868 MHz, and in North America at 902 to 928 MHz. You own the gateways. You own the network server. You can deploy it on a single factory rooftop or across an entire city. The protocol is maintained by the LoRa Alliance, and the chip-layer modulation is controlled by Semtech. Typical range is 2 to 5 km in dense urban areas and 10 to 15 km in rural line-of-sight. Battery life on a single AA lithium cell is 5 to 10 years for a device sending 100-byte uplinks every 30 minutes.
NB-IoT is a licensed cellular LPWAN standardized by 3GPP as part of LTE Release 13, with Release 16 and 17 refinements now widely deployed. It runs in existing telco spectrum, typically 700 MHz to 2.1 GHz depending on the operator, and rides on the public mobile network. You do not own the radio. You pay an operator per SIM per month. Coverage follows wherever the operator has enabled NB-IoT. Typical deep-indoor link budget is about 164 dB, which gives strong penetration into basements and metal cabinets. Battery life on a 5Wh battery is 5 to 8 years.
Axis 1: Coverage and who controls it
This is the single most important question, and most teams skip it. With LoRaWAN you control coverage. If your gateway works, your device works. If you need coverage in a village in Vidarbha where no carrier has bothered to enable NB-IoT, you put a gateway on a water tank and you are done. The flip side is capital. Every gateway is yours to buy, install, and maintain.
With NB-IoT you rent coverage. In 2026, Jio and Airtel have enabled NB-IoT across most Indian metros and tier-1 cities, but rural and semi-rural coverage in India is still patchy. Before committing to NB-IoT, drive-test the actual deployment locations with an NB-IoT scanner. Do not trust the operator coverage map, which is typically LTE and not NB-IoT. Europe and China both have far more complete NB-IoT coverage than India.
A simple rule of thumb: if your devices move across unpredictable geographies (asset tracking, logistics, nomadic equipment), NB-IoT wins. If your devices are stationary in a known perimeter (factory, farm, housing society, CGD ring-fence), LoRaWAN almost always wins. This is exactly why most of our electric meter and gas meter retrofit projects end up on LoRaWAN.
Axis 2: Total cost of ownership over 7 years
Hardware and connectivity economics look very different on day one versus year seven. Here is how a 10,000-device deployment plays out over 7 years.
LoRaWAN node cost in volume is Rs 350 to Rs 800 depending on sensor complexity. A gateway is Rs 15,000 to Rs 50,000 and covers 1,000 to 10,000 nodes. The network server is either self-hosted (free software, small cloud bill) or bought from ChirpStack, The Things Industries, or Actility at Rs 10 to Rs 50 per device per year. Seven-year TCO lands around Rs 900 to Rs 1,800 per device all-in.
NB-IoT module cost in 2026 is Rs 600 to Rs 1,200, which is slightly higher than LoRaWAN because of the LTE stack. The SIM and data plan is where it bites. Indian operator pricing for low-volume NB-IoT is Rs 20 to Rs 60 per SIM per month, falling to around Rs 10 per SIM for large fleets with negotiated contracts. Seven-year connectivity alone is Rs 840 to Rs 5,040 per device. Total TCO works out to Rs 1,400 to Rs 6,200 per device.
The TCO gap is real. LoRaWAN wins by 30 to 70 percent in most fixed-deployment scenarios once you cross 500 devices. Below 500 devices or across wide geography, NB-IoT can win because you avoid the gateway capex altogether.
Axis 3: Latency and message frequency
LoRaWAN is duty-cycle limited. In India the 865-867 MHz band has a 1 percent duty cycle at most frequencies. Practically, this means a Class A device can uplink every 30 seconds to 2 minutes for small payloads, or every 10 to 30 minutes for larger ones. Downlink latency is 1 to 30 seconds depending on class. For meter reading, environmental sensing, soil moisture, asset tracking, and most industrial telemetry, this is plenty.
NB-IoT has no duty cycle restriction. You can send data every few seconds if you want, though battery life tanks fast. Downlink latency in good coverage is 1 to 10 seconds. NB-IoT also supports larger payloads natively (up to 1600 bytes per message) versus LoRaWAN (51 to 242 bytes depending on data rate).
If your application needs sub-second command and control, neither technology is the right pick. Use LTE-M, 5G RedCap, or WiFi instead. If your app needs 5 to 30 second command latency (pay-as-you-go gas shutoff, or a vehicle immobiliser for two-wheelers), NB-IoT is the safer choice.
Axis 4: Power consumption in real conditions
Datasheets are marketing documents. Real power draw depends on link quality. A LoRaWAN device in good coverage (SF7) and an NB-IoT device in good coverage (CE Level 0) both draw negligible current averaged over a 30-minute interval, and both will last 5 to 10 years on the right battery.
The divergence shows up in poor coverage. A LoRaWAN device forced to SF12 because it is deep in a basement will drain its battery 10 to 30x faster than one at SF7. An NB-IoT device in extended coverage (CE Level 2) repeats transmissions 128 times and can drain a battery 50 to 100x faster than in good coverage. Both technologies punish poor coverage, but NB-IoT punishes it harder because every failed attempt still includes LTE signalling overhead.
For battery-powered devices in unpredictable RF environments, LoRaWAN is more forgiving. For line-powered devices where battery does not matter, NB-IoT wins on everything else.
Axis 5: Regulatory, security, and data sovereignty
NB-IoT gives you the cellular operator as a single throat to choke. DOT and TRAI regulate the spectrum. Your data flows through operator core networks. This is good for audit and compliance, since many enterprise buyers now require telco-grade SLAs. It is bad for sovereignty. An operator dispute, outage, or price hike hits your entire fleet at once.
LoRaWAN gives you a private network. You control the encryption keys (AES-128 at both the application and network layer), you choose where the network server lives, and you can run the whole stack air-gapped if needed. For defense, critical infrastructure, and SCADA-adjacent deployments, this matters a lot. Indian public sector buyers in power, water, and oil and gas are increasingly preferring private LoRaWAN for exactly this reason. If security is your driver, our IoT security guide for fleets and factories goes deeper on the threat model.
Decision matrix
Here is the quick narrow-down you can run in 5 minutes.
Pick LoRaWAN if: devices are fixed-location, the deployment is large (1000+ units in a defined area), devices are battery-powered, rural or semi-rural coverage is needed, data sovereignty matters, you have capex budget for gateways, and a 30-second uplink cadence is acceptable.
Pick NB-IoT if: devices are mobile or nomadic, deployments are small (under 500 units), devices are line-powered or uplink infrequently, devices are spread across multiple cities or countries, an operator SLA is required, you have no appetite to own gateway infrastructure, and deep-indoor urban coverage is critical.
Pick both (hybrid): large device fleets with a fixed-site core and a roaming edge. A city gas distribution company is a textbook example. LoRaWAN for society-level meters, NB-IoT for commercial and industrial PNG customers whose meters are spread across the city.
What to watch in 2026 and 2027
LoRaWAN 1.1 and the emerging 2.4 GHz plus sub-GHz hybrid profile solve the data rate ceiling. Expect 1 Mbps LoRa deployments in campus environments by late 2026. Gateway capacity is also scaling from a few thousand devices to tens of thousands per gateway in next-gen chipsets, which drops per-device cost further.
NB-IoT faces a slow squeeze from LTE-M (better for mobility and voice) and 5G RedCap (better for mid-bandwidth use cases). Most Indian operators now push LTE-M as the default for asset tracking and reserve NB-IoT for static metering. Keep an eye on 3GPP Release 18 NTN (non-terrestrial network) support, which brings NB-IoT over satellite. That single change could reshape the rural coverage calculus within two years.
If you are in the middle of a deployment decision, the right move is almost always a 3-month dual-protocol pilot. Deploy 50 LoRaWAN nodes and 50 NB-IoT nodes in representative locations. Measure real battery drain, real coverage, real operator data bills. Numbers on a vendor deck will lie to you. Your own data will not. The Akran IQ platform ingests both protocols through the same ingestion pipeline, which makes this kind of pilot cheap to run and easy to compare apples to apples. And if you are weighing whether to build the platform yourself or buy it, our piece on build vs buy IoT platforms lays out the tradeoffs.