Dev Kit → Module → Bare Chip: When to Move On
There are three stages that virtually every hardware product moves through, from early concept all the way to optimized production.
Most founders either skip one of these stages entirely or get stuck in one way too long, and both mistakes will destroy your margins and blow your timeline.
I’m going to show you exactly when to move from a dev kit, to a module, and to a full custom design, and what it costs you if you get the timing wrong.
Okay, let’s start with Stage 1, development kits.
Stage #1 – Development Kits
A dev kit is a pre-built evaluation board, like an Arduino, Raspberry Pi, ESP32 dev board, or STM32 Nucleo, that lets you test hardware and firmware concepts quickly without designing a custom PCB.
This is where almost every product should start, and this is commonly called a Proof-of-Concept prototype.
You’re validating your concept, testing firmware logic, and trying to move fast without spending a fortune on custom circuit board design.
Dev kits are also great for demonstrations, whether you’re showing functionality to investors, early customers, or even just proving to yourself that the idea actually works.
The beauty of this stage is speed and low cost, because you can get a working prototype together in days or weeks instead of months.
But the main limitation is size and cost.
If your product has any meaningful size constraint, like a wearable, a compact sensor, or anything that needs to fit inside a specific enclosure, a dev kit probably won’t get you there.
You can make it functional, but you can’t make it look or feel like a real product, and you definitely can’t ship it to customers in that form for most products.
It’s too big, too exposed, and it’s not designed for manufacturing.
You can demo it, but you can’t put it on a shelf or ship it to a customer who expects a finished product.
But in some cases, if your product’s form factor allows it and the price point supports it, you can sometimes run a limited-release paid pilot using dev kit-based units with a small group of early adopters.
Even if you’re losing money on each unit, you’re getting honest feedback from people who are actually using the product in their environment, not just telling you it looks cool at a tradeshow.
That kind of feedback is worth far more than margin at this point, because it tells you what to fix, what to prioritize, and whether people will actually pay for what you’re building.
I always encourage you to take the path that gets something into the hands of paying customers as soon as possible and for as little money as possible.
Too many founders rush past the dev kit prototype once they have it working, thinking there’s no more value they can extract from it.
It’ll be in your best interest to get as much feedback on your POC prototype from potential customers as possible, so you can incorporate that feedback into the next prototype stage.
One more thing to think about while you’re still choosing your dev kit, and this is a mistake I see all the time.
A lot of people grab an Arduino for their dev kit prototype because it’s easy to get started with, and that makes sense for learning.
But Arduino doesn’t give you a clean upgrade path to production, because there’s no Arduino module or Arduino chip that’s a good fit for production.
So when it’s time to move to Stage 2, you’re basically starting over with a new platform, new programming environment, and new firmware.
A much better approach is to pick a platform from the start that has a clear path through all three stages.
If you go with an ESP32, for example, you start with an ESP32 dev kit in Stage 1, move to an ESP32 module in Stage 2, and then if you ever reach the volume to justify it, you can move to a bare ESP32 chip in Stage 3.
The same is true for platforms like the nRF52 or STM32, they all offer dev kits, modules, and bare chips so you can move through each stage without throwing away your firmware and starting from scratch.
So the signal that you’re ready to leave Stage 1 is when the dev kit is either too large for your target form factor, or you’ve validated your concept and need to start reducing unit cost and improving manufacturability.
Stage #2 – Modules
Now it’s time to move on to Stage 2, modules.
A module is a pre-certified, production-ready sub-assembly that typically combines a microcontroller or microprocessor with RF circuitry, memory, and power management, all in a compact, shielded package.
Think ESP32 modules, Nordic nRF modules, u-blox GPS modules, or the Raspberry Pi Compute Module.
This stage is the right move when your product is too small for a dev kit, when it includes any kind of wireless connectivity like Bluetooth, WiFi, or cellular, or when you’re using a microprocessor and don’t want to deal with complex DDR memory layout and high-speed signal integrity.
One of the biggest advantages of modules is certification.
They typically come pre-certified for FCC and CE on the RF portion, which saves you a huge amount of time and money compared to designing and certifying your own wireless circuit from scratch.
Certification testing for a custom RF design can easily cost tens of thousands of dollars.
If your design fails, you’re looking at a redesign, a retest, and months of delays on top of that.
Modules let you skip all of that on the wireless side.
Now one thing to keep in mind is that a pre-certified module doesn’t get you completely out of certification.
You’ll still need to pass testing under FCC Part 15B, which verifies that your overall device isn’t emitting unwanted electromagnetic interference from things like your PCB traces, power supply, or display.
That testing typically runs a few thousand dollars, which is a lot more manageable than the $15,000 or more you’d spend on Intentional Radiator testing if you were designing a custom wireless circuit.
However, keep in mind that cost isn’t the only factor.
Sometimes, even at lower volumes, you have to go fully custom because a module simply won’t fit in a tiny wearable or a highly space-constrained device.
Many founders want to jump straight from a dev kit to a full custom design because they think modules are too expensive per unit.
I get it, when you look at a module that costs $4 versus a bare chip that costs $2, it feels like you’re leaving money on the table.
But when you factor in the engineering time, regulatory certification costs, and the complexity of bringing RF and high-speed signals onto a custom board, modules are almost always the better economic choice until you’re at serious volume.
You’re not at rock-bottom unit economics yet, but you’re in the range where a real business works and you can generate healthy margins.
So for most hardware products, Stage 2 is where they live for a long time, often years, and that’s completely fine.
The signal that you’re ready to leave Stage 2 is when your volume is high enough that the module cost is a meaningful drag on your margin, and you have the engineering resources and capital to take on a full custom design.
Stage #3 – Full Custom Board Design
Stage 3 is where you ditch the modules entirely and design a fully custom PCB, sourcing the microcontroller or microprocessor chip directly and handling your own RF design if your product needs wireless.
This stage makes sense when you’re producing at volumes where the module cost is materially hurting your margins, which is typically tens of thousands of units and up, or when you need size or performance optimization that modules simply can’t deliver.
But going from a module-based design to a full custom board isn’t just a PCB redesign.
It means new regulatory certification, including FCC, CE, and potentially others, on your complete design, including any RF circuitry you’re now responsible for.
That process is expensive and time-consuming, and those costs absolutely have to be factored into your unit price calculation before you commit.
You also take on significant new design complexity.
If your product has wireless, you’re now responsible for antenna design, impedance matching, and antenna tuning, all of which require specialized expertise and test equipment that most small teams don’t have in-house.
So this is exactly why jumping to a full custom board design too early is one of the more expensive mistakes a hardware founder can make.
The unit cost savings don’t materialize until volume justifies it, and the upfront investment in certification and engineering can get very expensive.
But when the volume is there, this is where your profit margins really improve.
You’re sourcing components directly, removing the module premium, and optimizing your bill of materials aggressively.
At meaningful production volume, the margin difference between Stage 2 and Stage 3 can be what separates a business that scales profitably from one that stalls out.
I’ve seen founders reduce their unit cost considerably by making this transition at the right time.
But I’ve also seen founders spend too much money and time on a custom design they didn’t need yet, and that money would’ve been far better spent on marketing or inventory.
So, while this is the path to the highest margins, it’s also where the highest risk lives.
This is why you don’t do it until the market has already proven the product.
