-
Step 1: Define the operating environment before you look at specs
-
Step 2: Quote the carrier, not just the SBC
-
Step 3: Ask about the software lifecycle
-
Step 4: Calculate connectivity and installation costs
-
Step 5: Plan for failure, return shipping, and downtime
-
The TCO checklist
-
Common mistakes I still see
I've been a procurement manager at a 45-person industrial automation company for about six years. I manage an annual budget of roughly $1.4 million for components, edge compute hardware, and IoT infrastructure. It took me a few hundred purchase orders and one very expensive lesson to understand this: the number on the invoice is not the cost.
This checklist is for anyone selecting an industrial SBC, an AI edge device, or an IoT gateway and wants to avoid the hidden costs that show up months later. I use the same logic whether I'm buying a $30 sensor, a $1,800 carrier board, or even a replacement furnace control board. The device might be different, but the thinking is the same.
Step 1: Define the operating environment before you look at specs
The first mistake is usually comparing CPUs and memory before anyone has defined where the device will actually sit. A board running in a clean, temperature-controlled server room is a very different product from the same board mounted inside a manufacturing cell, a vehicle, or a clinic wall.
For industrial SBCs, I ask my engineers for three things: ambient temperature range, shock and vibration limits, and power input quality. If the environment is harsh, a cheaper board often needs additional cooling, conformal coating, or a stronger enclosure. Those add-ons can turn a $200 price advantage into a $600 project cost.
This is also why I like to see test data, not marketing claims. Ask for IEC 60068-2-1 and IEC 60068-2-14 reports for temperature, and an IEC 62368-1 certification for safety. These are not paperwork exercises. They're evidence that a board will survive the real world.
Step 2: Quote the carrier, not just the SBC
An industrial SBC is rarely the whole system. In most edge deployments, you need a circuit board carrier to break out the I/O, power connections, and communication ports. The carrier often decides what the SBC can actually do, and it's a line item people forget to price.
From the outside, it looks like the SBC is the brain and everything else is just wiring. The reality is the carrier determines which subsystems you can access and how the board gets mounted. I've seen a $350 carrier completely change the economics of a project.
On one of our recent gateways, the compute module was $695. The carrier that gave us dual Ethernet and an M.2 slot was another $340. I almost approved it without the carrier because it 'didn't seem necessary.' Then our engineer priced out discrete connectors, custom cabling, and an enclosure cutout layout: $520. The carrier was the cheaper and more reliable option.
So when you compare quotes, make the vendor list the SBC, the circuit board carrier, the power supply, the antenna connectors, and the enclosure as separate line items. If something is not on the quote, the cost doesn't go away. It just moves to your in-house team.
Step 3: Ask about the software lifecycle
Hardware is the easy part. The software lifecycle is where budgets start to leak.
For healthcare IoT solutions, for example, you often need documented operating system versions, security patches, validation reports, and audit trails. A board that looks cheaper on paper may have a shorter support window or require a separate agreement for firmware updates. That annual maintenance fee changes the total cost of ownership.
It's tempting to think that if a board meets the CPU specs, the software will just be another install. That advice ignores the integration work. In my experience, the bigger cost risk is not the board, it's the time your engineers spend making an operating system image, maintaining drivers, and backporting security fixes. If a vendor offers a 10-year software maintenance period, that's worth a real premium.
I also check whether the board vendor provides a public hardware change notification policy. For a long-term deployment, you want to know when a component is discontinued before it happens, not after.
Step 4: Calculate connectivity and installation costs
Edge device pricing usually focuses on the hardware itself, but the ongoing connectivity and installation costs are part of the TCO, especially for distributed deployments.
For IoT in transport and logistics, devices are mounted on vehicles, trailers, or shipping containers. That means cellular modems, GPS antennas, remote diagnostics, and possibly a data plan per device. The device might cost $400, but the per-unit cellular plan, the gateway infrastructure, and the remote device management platform can double the annual cost.
People assume the cheapest hardware is the fastest to deploy. What they don't see is the configuration effort. How many devices need to be provisioned? Who updates the firmware over the air? What happens when a driver changes trucks? Each of those questions has a cost:
- Per-device data fees
- Remote management software licenses
- Mounting hardware and labor
- Training for field technicians
- Security certificates and identity management
None of these show up on the SBC quote. But they are real line items in the lifecycle budget.
Step 5: Plan for failure, return shipping, and downtime
This is the step most people skip, and it's the one that has burned me more than once.
Last year, we had a furnace controller go down in one of our office buildings. The replacement carrier furnace board itself was about $90. But by the time we paid for expedited shipping, the service technician's time, and the two days of reduced operation, the real cost was closer to $260. The board cost was a third of the actual expense.
The same logic applies to AI edge devices. If a unit fails, who pays for return shipping? Is there a spare device in inventory? What's the downtime cost per hour? A vendor with a slightly higher price but a local advance-replacement program can be far cheaper in practice.
When comparing quotes, I ask about RMA turnaround time, warranty coverage for shipping, and whether the vendor offers a volume spare pool. A unit that takes four weeks to replace creates a very different total cost than one that gets swapped in 48 hours.
The TCO checklist
Before you sign a purchase order for any industrial SBC, AI edge device, or IoT controller, run through this:
- Does the hardware meet the temperature, vibration, and power requirements of the installation site?
- Have you priced the circuit board carrier, enclosure, connectors, and cabling?
- What is the software support period, and are security updates included?
- What are the ongoing connectivity, management, and provisioning costs?
- What happens when a unit fails? Replacement time, shipping, and downtime?
Common mistakes I still see
The first is treating the SBC as 'the product.' It's not. The product is the entire edge system, including the carrier, the software, the compliance paperwork, and the support process.
The second is comparing only the first-year cost. A healthcare IoT solution with a three-year deployment needs a three-year cost model. A transport and logistics rollout with 500 devices needs a five-year model. Don't let an attractive one-year price hide the renewal fees in year two.
The third is ignoring the cost of in-house engineering time. If a board requires days of integration work, it's not a cheap board.
Don't hold me to this exact ratio, but in my experience, the hardware invoice is usually less than a third of the real cost of an IoT edge deployment. Once you start tracking your own TCO, you'll probably land somewhere similar.
So next time someone sends you a quote with an attractive unit price, ask yourself: what's the rest of the picture?