Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Handle Component Obsolescence
Transcript
- Lucas: You design a product, you select a microcontroller or a specialized ADC, you build your entire firmware and PCB layout around that chip, and then six months into production the manufacturer sends a PCN — product change notification — saying the part is going end of life. That is a nightmare scenario for any hardware startup. Luna: It's one of those things you hear about but don't really plan for until it happens. I've talked to founders who lost months of development time because of a single discontinued component. Lucas: If these conversations help you anticipate problems like that, that's the whole reason we do this show. We keep it ad-free and focused on what's useful for builders — and that's possible because a small group of listeners chip in through Buy Me a Coffee dot com slash fexingo. It's a simple way to support the kind of nitty-gritty details we cover here. Luna: Yeah, it's genuinely helpful to have a resource that doesn't try to sell you something every five minutes. And it keeps us independent. Lucas: Absolutely. So back to the problem — when a component goes end of life, what do you actually do? Most engineers instinctively start searching for a drop-in replacement, but it's rarely that simple. The first step is understanding your timeline. Luna: Right — you get the PCN, and typically there's a last-time buy window. Sometimes it's 90 days, sometimes it's a year. How do you decide whether to use it? Lucas: The last-time buy is your safety valve. You can purchase enough of the component to cover your projected production for the next 12 to 24 months, or even longer. But you're tying up cash in inventory, and if your product pivots or demand flops, you're stuck with obsolete parts. There's a real trade-off there. Luna: So you need a forecast you trust. Easier said than done for a startup. Lucas: Exactly. That's why the smarter play is to invest in obsolescence monitoring from day one. There are services — like SiliconExpert or IHS Markit in the past — that aggregate PCN data and give you early warning. For a small startup, it might be as simple as subscribing to distributor notifications on your critical parts. The point is you don't want to find out about an EOL when you go to place a replenishment order and the part is suddenly unavailable. Luna: And if you can't do a last-time buy — maybe the window already closed or the volumes don't justify it — you're forced into a redesign. That's where the real cost hits. Lucas: Let me give you a concrete example. I spoke with a startup that was building a wearable medical device for continuous glucose monitoring. They selected an analog to digital converter from Analog Devices — a specific 24-bit sigma-delta converter that was perfect for their low-power requirements. They were about two years into development, had already done clinical validation, and were scaling up production. Luna: Ouch. So the whole system was tuned around that ADC. Lucas: Exactly. They got a PCN saying the part would be discontinued in six months. The last-time buy window was 90 days, and their lead time for the component was already 16 weeks. So realistically, they had about two months to place an order. They ended up buying a two-year supply — about 200,000 units — but that ate up a huge chunk of their Series A funding. And they still had to plan for a redesign because two years later they'd need a new part. Luna: So they kicked the can down the road, but the redesign was inevitable. Lucas: Right. They started searching for a pin-compatible replacement. They found a part from Texas Instruments that had similar specs, but the offset voltage was slightly different, so they had to tweak their calibration algorithm. That took another nine months of testing and regulatory revalidation. The whole delay pushed their market entry by almost a year. Luna: Nine months for a single component swap. And that's assuming you find a replacement. What if you don't? Lucas: Then you're looking at a major board spin — new layout, possibly a different package, new thermal analysis. I've seen startups that had to switch from a QFN package to a BGA because the original part was the only one in that small QFN footprint. That changed their entire assembly process. Luna: And every spin adds months and hundreds of thousands of dollars. So the lesson is you need to plan for obsolescence during the initial design. What's the best practice? Lucas: First, whenever possible, choose components that have multiple second sources. For example, if you can use an industry-standard op-amp that's made by three different manufacturers, you're in a much better position than if you rely on a unique part from one vendor. Second, design for adaptability — leave room on the board for alternative footprints, or use socketed components for critical chips. Luna: That sounds like it adds complexity to the layout. Lucas: It does, but it's insurance. The IPC released a survey in 2025 — they asked hardware startups about their biggest supply chain headaches, and 60% reported at least one critical component obsolescence event during their first product lifecycle. The average delay was seven months. The average cost — between engineering time and respins — was about $350,000. That's a serious hit when you're a startup. Luna: So it's not just a 'nice to have' — it's a core risk management practice. Lucas: Exactly. I'd argue that obsolescence planning is as important as thermal management or EMC compliance. If you don't think about it until you get the PCN, you're already behind. Luna: And what about the software side? If the microcontroller goes EOL, you're not just changing a chip — you're porting firmware to a new architecture. Lucas: That's a whole other layer. Some startups use abstraction layers or RTOS that can be ported, but if you've written bare-metal code tightly coupled to the hardware, it's a major rewrite. That's why we see more teams using higher-level frameworks like Zephyr or FreeRTOS, which have driver abstraction. It's still work, but it's less painful. Luna: So the takeaway is: choose parts with second sources, monitor for PCNs early, and design for flexibility. Anything else? Lucas: One more thing — build relationships with distributors. If you're a small startup, you might not get the same attention as a large OEM, but even a good relationship with a distributor's field applications engineer can give you early warnings. Sometimes they know about imminent EOLs before the PCN goes public. Luna: That's a good tip. It's all about reducing surprises. Lucas: Right. And if you do get caught in a redesign, try to use it as an opportunity to improve — maybe that new ADC has better performance or lower power. It doesn't always have to be a pure setback. Luna: That's a glass half full way to look at it. But in a startup, every delay hurts. Lucas: Absolutely. The goal is to avoid the delay altogether. With a bit of upfront planning, you can keep your product on schedule even when parts go away. And that's a competitive advantage.