Here's the thing about choosing IoT connectivity: there's no single "best" option. If anyone tells you LoRa is always the answer, or that LTE-M is the future and everything else is obsolete, they're probably selling something. I've been managing connectivity component budgets for industrial IoT devices for about six years now, and the right choice depends entirely on what you're building and where it's going.
I've compared costs across dozens of vendors—from module manufacturers to network operators to antenna suppliers. In this guide, I'll break it down by the three most common scenarios I see, and help you figure out which one matches your project. Note: pricing here is based on Q4 2024 quotes; the semiconductor market changes fast, so verify current rates before locking in your BOM.
Three Scenarios, Three Different Answers
Before we get into the weeds, let's map out the three situations where you might be reading this.
Scenario A: You're building a low-power sensor network for agriculture, smart buildings, or asset tracking. Devices are battery-powered, deployed in remote areas, and need to run for years with minimal maintenance.
Scenario B: You're making a device that needs reliable, real-time communication—like a medical monitor, a fleet tracker that reports location every 30 seconds, or an industrial controller. Uptime and latency matter.
Scenario C: You're somewhere in between. Maybe a smart city application where some sensors are in dense urban areas and others are in basements. Or a mixed fleet where some assets are mobile and some are stationary. You're considering a hybrid approach.
Your answer changes depending on which scenario you're in. Let's walk through each one.
Scenario A: Low-Power, Long-Range, Low-Data—Go LoRa
If your devices need to send small packets (like temperature readings, occupancy counts, or vibration alerts) a few times per day and run on a coin cell battery for 5-10 years, LoRa is almost always the most cost-effective choice. I've seen this play out repeatedly in the smart agriculture space.
Let's talk numbers. A Semtech LoRa transceiver (like the SX1276 or the newer SX1262) costs roughly $2-4 in volume (10k+ units). Compare that to an LTE-M/NB-IoT module, which typically runs $10-20 for comparable volumes. That's a 5x difference before you even get to the network side.
Network costs for LoRaWAN are also significantly lower. If you're deploying a private network, a single LoRaWAN gateway (like the Semtech SX1301-based ones) covers several kilometers and costs a few hundred dollars. Or you can use a public LoRaWAN provider for maybe $2-5 per device per year.
Cellular IoT (LTE-M/NB-IoT) typically runs $5-15 per device per year for low-data plans, but that's just the connectivity. The module cost, as I mentioned, is higher. And power consumption? BLE and Wi-Fi alternatives can't match LoRa's range, while Zigbee and Z-Wave have shorter range and different ecosystem requirements. Actually, let me clarify—I'm not saying LoRa beats cellular in every metric. For raw data throughput, cellular wins. But for this specific use case (low power, long range, small data), the TCO favors LoRa.
From the outside, it looks like you're just comparing chip prices. The reality is the total system cost difference is even bigger when you factor in battery replacement (or lack thereof), gateway density, and certification costs. With LoRa, you can often skip expensive carrier certification. With cellular modules, you're looking at FCC, PTCRB, and carrier-specific approvals that can add $20,000-50,000 per module variant. We didn't have a formal certification cost tracking process at first. Cost us when we realized our "cheap" cellular module required three separate carrier certifications that blew our timeline by four months (note to self: always add certification to the BOM cost early).
If you're going this route, you'll want to pair the LoRa transceiver with a properly tuned antenna. Semtech antennas (or third-party ones designed for LoRa bands) are critical for achieving that advertised range. A poorly matched antenna can cut your range by 30-50%, which then requires more gateways—and more gateways means more cost. That $1 saved on an antenna can cost you $200 in extra gateway hardware.
Scenario B: High-Reliability, Real-Time, High-Data—Consider Cellular (or Ethernet)
Now, if your device sends large files, streams video, or needs real-time command-and-control, LoRa isn't the right tool. It's optimized for small, infrequent packets. Trying to push video over LoRa is like trying to fill a swimming pool with a garden hose—technically possible if you have infinite patience, but not practical.
For this scenario, you're looking at cellular modules (LTE-M, NB-IoT, or full LTE) or—for stationary devices—Ethernet, if it's available. This is where the recent Sierra Wireless acquisition by Semtech becomes relevant. Before the acquisition, Semtech was strong in LoRa but didn't have a native cellular play. Now, through Sierra Wireless (which operates as part of Semtech), they offer a full suite of cellular modules and routers, including the Airlink line.
Pricing for cellular modules varies wildly. A basic LTE-M/NB-IoT module for volume orders might be $8-12. A full LTE module with GNSS and voice support can hit $25-40. But here's the thing—the TCO isn't just about the module. You also need:
- A data plan ($5-25/month per device depending on data allowance)
- Certification costs (mentioned above—easily $20k-50k per variant)
- Higher power consumption (larger battery or more frequent charging)
- Firmware complexity (cellular stacks are more complex than LoRaWAN)
I want to say the total cost for a cellular-connected device over a 5-year lifetime is roughly 3-5x more than an equivalent LoRa device, but don't quote me on that—it depends heavily on data usage and deployment scale. For a high-value device where reliability is paramount (like a medical alert system or a critical infrastructure monitor), that premium is absolutely worth it. Client feedback scores improved by 23% when we switched from a lower-cost but less reliable connectivity option to a premium cellular solution for a remote healthcare monitoring project. The $50 difference per device translated to noticeably better client retention.
People assume cellular IoT is always more expensive upfront. What they don't see is the long-term operational simplicity for certain use cases: no gateway management, no network planning, just a SIM card and you're online. For a small deployment (<50 devices), cellular can actually be cheaper because you skip the gateway hardware.
And if you're dealing with stationary devices in an office or factory environment? Don't overlook the Airlink routers from Semtech. They're designed for enterprise-grade cellular connectivity with features like VPN support, dual SIM failover, and remote management. When I audited our 2023 spending, I found that our "DIY" connectivity solution using off-the-shelf modems had cost us 15% more in downtime and maintenance than a pre-certified router would have.
Scenario C: The Hybrid Approach—When You Need Both
This is the scenario I see most often in practice, and it's the one where most people get stuck. You have a mixed fleet: some sensors in the field (Scenario A), some in dense urban environments (Scenario B-ish), and some on the move. Or you're building a device that needs to be future-proofed for different deployments.
The temptation is to pick one technology and standardize. The smarter approach—and the one that's becoming more common in 2024-2025—is a hybrid module that supports both LoRa and cellular. Semtech's portfolio now covers both sides: LoRa transceivers for the long-range, low-power sensors, and Sierra Wireless cellular modules for the high-bandwidth, real-time applications. Some manufacturers offer modules that include both LoRa and LTE on a single board, which simplifies design and certification.
Is it more expensive than a single-technology solution? Yes, upfront. The BOM cost for a dual-mode module might be $15-25 vs. $3-4 for LoRa-only or $10-15 for cellular-only. But it gives you flexibility. You can deploy the same device in a rural farm using LoRa and in a city using LTE, depending on what the customer needs. From a procurement perspective, standardizing on fewer SKUs can actually save you money in the long run through volume discounts and simpler logistics. It's tempting to think you can just standardize on one technology and save money. But the "always pick one" advice ignores the reality that your customers have different needs.
If I remember correctly, about 60% of the industrial IoT projects I've reviewed in the past two years ended up with some form of hybrid architecture—either dual-mode modules or separate devices for different tiers of service.
How to Decide Which Scenario You're In
Here's a quick checklist I use when evaluating new IoT projects. Ask yourself these questions:
- How much data does each device send per day? If it's measured in bytes (not kilobytes), you're likely in Scenario A. If it's megabytes or more, you're in Scenario B.
- How often does the device need to communicate? Once per hour or less? Scenario A. Every few minutes or in real-time? Scenario B.
- What's the expected battery life? If you need years on a coin cell, LoRa (Scenario A) is your only realistic option. If you can recharge annually or have mains power, cellular (Scenario B) becomes viable.
- How many devices are you deploying? Under 50 units? Cellular may be cheaper due to avoided gateway costs. Over 500? LoRa's lower module cost and free (or cheap) network access tilts the math back.
- Where are your devices located? If they're all in one area you control, private LoRaWAN makes sense. If they're spread across multiple cities or countries, public cellular is simpler.
I built a simple cost calculator after getting burned on hidden fees twice—once for a rushed certification, once for a gateway that turned out to be incompatible with our chosen module. It's basically a spreadsheet that compares TCO over 3 years for LoRa, cellular, and hybrid options. I'd share it, but honestly, every deployment is different enough that a generic calculator can be misleading. Better to build your own based on your specific BOM and deployment plan. This was accurate as of Q4 2024. The IoT module market changes fast, so verify current pricing and availability before finalizing your design.
And one more thing—if you're still using the Semtech LoRa chips (SX1276, SX1262, etc.), that's a solid foundation. They're mature, well-documented, and have a huge ecosystem. Just don't assume they're the answer for every problem. That's the mistake I made in 2022 when I tried to use LoRa for a high-frequency fleet tracking application. It worked, barely, but the device's battery life was 2 months instead of the promised 2 years. Learn from my error: match the technology to the use case, not the other way around.
