Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Modular Design to Scale Faster
Transcript
- Lucas: If these conversations are useful for what you're building or running, you should know we keep these episodes deliberately ad-free. We don't run sponsors, we don't read mid-rolls. The only reason we can do that is because listeners who find value here sometimes choose to support the show directly. Luna: Yeah, it's a small thing that makes a big difference for us. Lets us focus entirely on what's useful for founders and operators, not on selling something between segments. Lucas: Exactly. So if that model appeals to you, the link is buy me a coffee dot com slash fexingo. No pressure, no perks, just keeping this independent. And with that, let's get into today's angle. Lucas: We've talked a lot on this show about specific manufacturing techniques — injection molding, ultrasonic welding, conformal coating. But there's a higher-level design decision that often determines whether a startup scales smoothly or rebuilds from scratch every generation. Luna: You're talking about modular versus integrated design. Lucas: Exactly. And it's one of those decisions that looks trivial on paper but has massive downstream consequences. So let's anchor on a specific case: August Home, the smart-lock company. Luna: I actually installed one of their locks a few years ago. The module that holds the motor and electronics slides into the back of the deadbolt assembly, right? Lucas: Right. That's the key insight. August separated the electronics module from the mechanical core. When they released the second generation, they didn't redesign the deadbolt — they just upgraded the electronics module. Same mechanical parts, same tooling, same injection molds. Luna: So they saved on tooling costs twice. The mold for the mechanical part can cost fifty to a hundred thousand dollars. Avoiding a second set of molds is a huge win. Lucas: Exactly. And that's the modular payoff. But it's not just about tooling. It's about inventory risk. If you have a single integrated product, every component is unique to that device. If sales miss, you're stuck with unsold inventory of custom parts. With a modular design, you can carry a common base module and swap the application-specific module based on demand. Luna: That's why investors are starting to ask about modularity. I've heard a few hardware VCs mention that a modular architecture can reduce the capital needed to scale by thirty to forty percent. Lucas: It's true. And the classic counterexample is the original Roomba. iRobot's early Roombas were heavily integrated — the main board, sensors, motors, and battery were all in one custom shape. When they wanted to add a new feature, they often needed a new motherboard layout, sometimes a whole new chassis. Luna: Right, and that meant every model generation required a full requalification cycle — thermal testing, drop testing, FCC certification again. Lucas: Exactly. Compare that to a company like Framework, the laptop startup. They modularized the entire laptop — expansion cards, battery, screen, mainboard. You can upgrade the mainboard without replacing the chassis, keyboard, or battery. That's extreme modularity, and it's expensive upfront, but it lowers the cost of each subsequent generation. Luna: But there's a trade-off. Modular designs are usually larger, heavier, and less power-efficient than integrated designs. If you're building a wearable or a drone, every gram matters. Lucas: That's a real tension. And it's why the best approach is often semi-modular. You modularize at the subsystem level — the power subsystem, the compute subsystem, the sensor array — but integrate within each subsystem. That way you get some reuse without the overhead of full modularity. Luna: So the decision really depends on how many generations you expect to ship. If you're planning a single product and then moving on, integrated might be faster and cheaper. But if you're building a platform, modularity pays back after generation two or three. Lucas: And the cost of modularity has come down dramatically in the last five years. The main reason is standardized connectors. usb c, M.2 slots, FPC connectors — they're cheap, reliable, and widely available. In 2020, adding a modular connector might have cost fifty cents per unit. Now it's closer to ten cents. Luna: Plus the ecosystem of off-the-shelf modules. You can buy a compute module from Raspberry Pi or Qualcomm, a cellular module from u-blox, a sensor module from Bosch. You don't have to design every board from scratch. Lucas: That's the second-order effect. Modular design lets you leverage the supply chain's existing modules, which means you don't have to certify each module individually. The module manufacturer already handled FCC and CE. You only certify your own interconnect board. Luna: That can save months in certification time. And for a startup, time is the scarcest resource. Lucas: Let's bring in a contrasting case from a different scale: Tesla's battery pack. The Model S pack is modular — it's made of sixteen modules, each containing hundreds of 18650 cells. When Tesla wanted to improve the pack for the Model X, they kept the same module form factor but used a newer cell chemistry. They didn't redesign the entire pack architecture. Luna: But Tesla moved away from that with the 4680 cells. The structural battery pack in the Cybertruck is integrated — the cells are part of the chassis structure. Lucas: Right, and that's a deliberate trade-off. They optimized for energy density and structural efficiency, accepting higher design complexity. But Tesla has the engineering team and the volume to absorb that risk. A startup probably shouldn't. Luna: So what's the practical takeaway for a founder listening today? How do you decide? Lucas: Start by mapping your product's likely evolution over three generations. Identify the subsystems that are most likely to change — the wireless chipset, the sensor payload, the user interface. Those are candidates for modularization. The subsystems that are stable — the mechanical frame, the power supply — those can stay integrated. Luna: And be honest about your expected volume. If you're planning to ship ten thousand units, the upfront engineering cost of modularity — designing connectors, writing interface specs, testing interoperability — might not pay back. But if you're aiming for a hundred thousand units across multiple generations, it almost certainly will. Lucas: Exactly. And there's a cultural aspect too. Modular design forces you to define clear interfaces early. That discipline helps with team coordination, with contract manufacturing, with field service. Even if you never build a second generation, the process of thinking modularly makes your first product better. Luna: So the real lesson might be: design the interfaces, not just the product. Lucas: That's a perfect way to put it. Thanks, Luna. That's all for this episode of Hardware Startups with Fexingo.