I've handled IoT connectivity procurement for about six years. In that time I've made my share of mistakes—20-odd significant ones, totaling well over $15,000 in wasted budget. The costliest single error was comparing Semtech and Crown Castle as if they were two ways of achieving the same thing. They're not. That wrong framing cost me $3,200—maybe $3,400, I'd have to check the final PO—and a three-week delay that cost us a client's patience.
My opinion: the "Semtech vs Crown Castle" question is a category error. It's not that one company is better than the other. They operate at different layers of connectivity, and conflating them will push you toward bad decisions. I know because it pushed me into exactly that.
How I made the $3,200 mistake
In March 2024, we were building a pilot deployment of 40 environmental sensors across three industrial buildings. The client's operations team wanted to track temperature, humidity, and vibration telemetry on pumps and HVAC units. They'd been pitched a cellular solution using macro tower infrastructure—the kind Crown Castle owns tens of thousands of. The pitch sounded reasonable: "established infrastructure, monthly leasing, predictable pricing." From the outside, macro towers look like the safe answer. The reality is that the coverage map ends where your sensors begin.
Ours were going inside parking garages, a basement maintenance room, and a mechanical space with a 7.1-channel acoustic monitoring array. The LTE signal in the basement was one bar if we were lucky. The array buffered constantly—which is to say, it made the exact kind of janky performance that erodes a client's confidence in your engineering judgment.
We spent $2,400 on LTE modules and data plans, then the extras arrived: $200 in antenna license fees, $400 in carrier certification engineering, $600 on a signal booster we bought in a mild panic. That's $3,200. No, $3,600—I'm mixing it up with the certification line item. Call it a four-figure lesson that could have been solved with a single LoRaWAN gateway for a quarter of the cost.
What I missed about Semtech
Here's the part where I should have known better. When Semtech acquired Sierra Wireless in August 2022—a roughly $1.2 billion deal per Semtech's announcement at the time—it stopped being "the LoRa chip company" in any meaningful sense. The combined portfolio spans the transceiver, the gateway, the cellular module, and the industrial router. The Sierra Wireless XR60 5G/LTE router is a good example: it lets a factory run private LoRaWAN for sensors while keeping a cellular failover link for critical data streams. That's not a competitor to Crown Castle's towers. It's a complement to them.
But my mental model of Semtech was stuck years in the past. I treated it like a component vendor and ignored the systems-level story. That blindness is exactly what led me to frame the question as "Semtech or Crown Castle?" in the first place.
The antenna reality check
People assume bigger infrastructure means better coverage. In industrial IoT, that assumption is often backwards.
A single LoRaWAN gateway with properly placed Semtech antennas can cover a 2–3 km outdoor radius and, crucially, penetrate concrete parking structures that swallow cellular signals entirely. The LoRa Alliance has published multiple deployment case studies since 2023 that show this pattern. Meanwhile, a colocated antenna on a macro tower does exactly one thing well: it serves phones and mobile broadband over long distances. That's valuable, but it's not the same job.
The pricing asymmetry is hard to ignore:
- Crown Castle colocation lease: $1,200–$2,000/month depending on market and load (industry-reported rates as of August 2024; verify current pricing).
- LoRaWAN gateway with Semtech SX1302 concentrator: $300–$1,500 one-time, depending on channel configuration.
- LTE data for 40 sensors: roughly $35/device/month—$1,400/month in perpetuity.
- Private LoRaWAN for 40 sensors: no monthly fee with your own gateway; a managed network server runs about $1–$2/device/month.
Three things stand out from this comparison: the monthly multiplier, the hardware difference, and—critically—the coverage physics.
The monthly multiplier is the one that gets ignored. A colocation lease is an operating expense that never leaves your budget. A gateway is a capital expense that shows up once. For a company running dozens of industrial facilities, that difference compounds quickly. We now run a simple test on any quote: How long until the recurring fees exceed the one-time cost? If the answer is less than 18 months, the cheaper-looking option is usually the expensive one.
Objections I hear
"Semtech is much smaller than Crown Castle financially. Does that mean it's riskier?"
Crown Castle's balance sheet is clearly bigger—around $8 billion in annual revenue by late 2024, versus a fraction of that for Semtech. Size matters if you're buying at enormous scale. For a pilot deployment, the relevant question is whether the product family has traction. LoRaWAN devices number in the hundreds of millions globally, per LoRa Alliance counts. If that's a startup risk, then most industrial IoT projects since 2019 made the same bet.
"Crown Castle publishes rates. Isn't that more transparent?"
Published rates for the tower lease, yes. But then come backhaul, power, installation, and lease escalation clauses. I've learned to ask "what's NOT included" before I ask "what's the price"—that habit alone would have saved my $3,200. The vendor who lists all fees up front, even if the total looks higher, usually costs less in the end. That principle applies both when you're leasing tower space and when you're comparing chip vendors' reference designs.
"Phones need cellular. What does a LoRa chip have to do with that?"
Nothing, and that's exactly the point. A macro tower exists to serve phones, and it's great at it. The sensors in our pilot were not phones. They sent 12-byte payloads four times a day. Putting them on a network designed for voice and video mobile traffic is like leasing a highway lane to deliver a letter. The right tool for that job was a narrowband private network—and that's not a knock on Crown Castle. It's just a different job.
What I'd tell my 2024 self
The first thing I'd change is the question itself. Stop asking "Semtech or Crown Castle?" and start asking "macro coverage, private coverage, or both?" They're not substitutes; they're different tools in the same kit. In our current architecture, we use cellular for the routers and LoRa for the sensors, and Crown Castle's network handles the wide-area backhaul between sites. Semtech provides the chips and modules that make the private side work.
I only believe this framing because I ignored it once and paid for the lesson. Now my procurement checklist starts with three lines: specs confirmed, timeline agreed, hidden fees identified—in that order. It's caught 47 potential issues over the past 18 months, if I'm remembering the count right.
The real lesson from my $3,200 mistake isn't that one company beats the other. It's that comparisons only work when you're comparing the right things.
