LoRaWAN vs NB-IoT vs Sigfox: How to Budget an IoT Sensor Pilot

The short answer: one decision drives most of your pilot cost, and it is not the radio technology. It is whether you run a private network you own (LoRaWAN, where you buy a gateway and pay nothing per device) or rent a carrier network you do not own (NB-IoT and Sigfox, where you buy no gateway but pay a subscription per device every year). Get that call right and the rest is line items. Get it wrong and you either overpay capex for a 12-device pilot or sign up for subscriptions that compound for years. This post gives you the numbers, a worked budget for a 10-50 device pilot, and the Q&A we field before scoping an IoT engagement.

The one decision that drives 80% of your IoT pilot cost

Every LPWAN comparison wants to argue about range and battery life. Those matter, but they are not where your money goes. The cost-defining question is the network ownership model.

• Private network (LoRaWAN). You buy a gateway, mount it, and own the network. Adding devices after that costs only the price of the device. No per-device subscription, no carrier in the loop. Your spend is mostly capex: gateway, install, antennas. You can also use public LoRaWAN coverage (The Things Network, Helium) and skip the gateway entirely, trading control for someone else's coverage map.
• Carrier network (NB-IoT, Sigfox). There is no gateway to buy because the mobile operator already built the towers. You pay a subscription per device per year. Your spend is mostly opex, and it never stops. NB-IoT runs on licensed cellular spectrum operated by mobile carriers; Sigfox runs on a dedicated operator network with national coverage partners.

That single fork sets the shape of your budget. Private LoRaWAN front-loads cost into hardware you own. Carrier networks spread cost into a subscription that grows with your fleet and your years in production. Everything else, the module prices and battery numbers below, is rounding error next to which curve you signed up for.

LoRaWAN vs NB-IoT vs Sigfox: when does each win?

Each radio has a job it is good at. The figures below are typical 2026 ranges, not quotes; your volume, region, and carrier will move them. Module prices are corroborated by Link Labs and TEKTELIC; Sigfox and NB-IoT subscription and message-cap figures from the NB-IoT vs Sigfox cost comparison.

Module hardware itself is close: a LoRaWAN sensor module runs roughly $8-15, an NB-IoT module roughly $5-12. Sigfox modules are cheap too. The hardware is not the deciding cost. The network model is.

What does a 10-50 device pilot actually cost?

Here is a line-item budget for a 25-device pilot, the size most clients start with. Treat these as typical ranges with stated assumptions, not a quote.

Option A: private LoRaWAN

• Sensor modules: 25 x ~$8-15 = ~$200-375.
• Gateway: one indoor unit ~$100-300, or one outdoor/industrial unit ~$1,000-3,500 for a real deployment.
• Install: mounting, antenna, cabling, lightning protection for an outdoor site ~$1,000-2,000.
• Subscription: none. You own the network.
• Pilot total: roughly $1,300-6,000 mostly capex, then near-zero recurring.

Option B: carrier NB-IoT or Sigfox

• Sensor modules: 25 x ~$5-12 = ~$125-300 (NB-IoT); Sigfox similar.
• Gateway: none. The carrier owns the towers.
• Install: minimal, just mount the sensor.
• Subscription: NB-IoT ~$1-6/device/yr; Sigfox ~$1-14/device/yr. For 25 devices that is ~$25-350/yr, every year.
• Pilot total: roughly $150-650 in year one, then a recurring annual bill that scales with device count and years.

For a one-off 25-device pilot you are running for a few months, the carrier route is cheaper to start because you skip the gateway. The question is what happens when the pilot becomes a fleet.

Capex vs opex: where the TCO crosses over

This is the calculation that actually matters. Walk a worked example over five years.

• Private LoRaWAN: one outdoor gateway plus install, say $3,000 capex, plus $12 per device hardware. Adding devices is just hardware. At 100 devices over five years: $3,000 + (100 x $12) = ~$4,200, with no recurring network fee.
• Carrier (NB-IoT at $4/device/yr): $10 per device hardware plus $4 per device per year. At 100 devices over five years: (100 x $10) + (100 x $4 x 5) = $1,000 + $2,000 = ~$3,000.

In this example the carrier route still wins at 100 devices over five years, because the gateway capex is large relative to a small fleet. The crossover flips the moment your device density per gateway climbs. One gateway can serve hundreds to thousands of LoRaWAN nodes, so the gateway cost is fixed while the carrier subscription is per device and unbounded. Run the same math at 1,000 devices over five years: LoRaWAN is $3,000 + (1,000 x $12) = ~$15,000, while NB-IoT is (1,000 x $10) + (1,000 x $4 x 5) = $10,000 + $20,000 = ~$30,000. Private LoRaWAN is now half the cost. The more devices per gateway and the longer the horizon, the more the private network wins. The right answer depends on your real device count and lifetime, not a brochure. We build that model with clients before anyone buys hardware, the same way we scope fixed-price vs time-and-materials work.

Q&A: is Sigfox dead?

No, but it had a near-death moment. The original Sigfox company entered receivership in 2022 and was acquired by Unabiz, which now operates the network and has committed to it long-term. The technology still does one thing better than anyone: tiny messages, very long battery life, very low opex. The constraint is real though, roughly 140 uplink messages of 12 bytes per day. If your use case fits in that envelope (a daily meter reading, a presence flag, an alarm), Sigfox is cheap and durable. If you need throughput or two-way control, it is the wrong tool. Treat it as alive but narrow.

Q&A: can I switch networks later?

Only if you plan for it. The radio is baked into the module at the hardware level, so a Sigfox sensor cannot become an NB-IoT sensor with a firmware update. Switching networks means new hardware in the field. The pragmatic hedge is multi-radio modules (some support LoRaWAN plus cellular fallback) or designing the pilot so the application layer is network-agnostic. The expensive mistake is committing 5,000 single-radio devices to a network model before you have validated the TCO curve. Pilot small, decide, then scale, which is exactly why we benchmark before committing, the same discipline behind our shipped-vs-failed product work.

Q&A: what battery life should I actually expect?

It depends almost entirely on how often you transmit, not the headline radio spec. A sensor that sends one small message per day can run years on a single coin cell or AA pack across all three technologies. A sensor that transmits every few minutes drains in months. Sigfox and LoRaWAN have an edge on the lowest-power end because the radios are simpler; NB-IoT draws more during transmit because cellular protocols are heavier. Before you trust any vendor's "10-year battery" claim, ask at what transmit interval and message size they measured it. The honest answer is always conditional.

Q&A: indoor vs outdoor coverage, what changes?

Carrier networks (NB-IoT, Sigfox) live or die on the operator's coverage map. Run a coverage check for your exact site before you commit, especially indoors, in basements, or in rural areas where cellular is thin. NB-IoT was specifically designed for deep indoor penetration and usually wins underground. With private LoRaWAN you control coverage by placing your own gateway, which is the whole point: no carrier dependency, no dead-zone surprises. If your site has poor cellular coverage and you cannot move it, private LoRaWAN often stops being a cost decision and becomes the only decision. See our software development and IoT services for how we run a site survey before scoping hardware.

"Pick the network model first, not the radio. Capex you own beats opex you rent the moment your fleet outgrows a single gateway."