Over the past few years, we’ve spoken with hundreds of hardware startups, and there’s one phrase that has been showing up in almost every interview: “supply chains.”
They were always important for people shipping physical products, of course, but the pandemic and the international logistics chaos that resulted from it made the pain of getting components to factories and products from factories to warehouses, then to stores and to customers, all the more present.
A few of the founders TechCrunch spoke to at CES in Las Vegas this year seemed to be optimistic, saying that some of the supply-chain crisis is on its way to being resolved, while others laughed at the suggestion that all of this is behind us. We decided to find out what the status is.
I spoke with supply-chain experts Fictiv to figure out whether the light at the end of the tunnel is daylight or the headlights of an oncoming train. If anyone knows what’s going on, it’s Fictiv CEO Dave Evans. He’s been running the manufacturing platform for more than a decade; before that, he was part of the Advanced Manufacturing Group at the World Economic Forum and a lead hardware engineer at Ford. We talked to Evans about the status quo and the near future of hardware manufacturing, seen through the lens of startups.
“The main problem for early-stage startups is that it’s really difficult to get someone to care. You might care about your product, but nobody else does because you’re too small. Suppliers will ask if you have volume. If you don’t, why would they work with you?” Evans said, pointing out that relationships are super important with whoever your supplier base is. He noted that you have to be the right size for your suppliers. If you’re too small, they don’t want to deal with you. But if you’re too big, that’s also a problem.
“When I was at Ford, you know, there were a lot of great suppliers, but they only do $10 million a year. At Ford, we never worked with them, because there was simply too much risk.”
Why ButcherBox built two dry ice factories during the pandemic
I asked Evans if the supply-chain crisis was finally in the rearview mirror. When he was done laughing, he had a pretty simple answer: Nope.
“Is it over? I would say, unequivocally, no. I think we’ve shifted from a conversation about ‘if’ to a conversation about ‘when.’ I don’t think we are at the peak of it by any means,” he said, adding that while the pandemic was a big hit, supply-chain resiliency itself has been a conversation among his peers for a very long time. “We can go back to 2019 and talk about trade wars. We can go to 2020 and talk about pandemics, the Suez Canal, we can talk about shipping delays and massive global conflict in Europe. We can talk about recessions and all of these things that disrupt global trade and supply chains.”
The big change that happened during the pandemic, however, was that supply-chain disruptions started showing up in the lives of people who weren’t supply-chain professionals. Average consumers suddenly couldn’t buy the kind of ketchup they wanted, or the new car they were excited about was delayed, or they couldn’t get the right stuff at the hardware store to renovate their homes.
For businesses, disruptions were mostly hidden from view before the pandemic, something the chief supply chain officer would take care of. It went from a functional, practical problem that affected a part of the business to a management problem where the whole team suddenly started caring. Everyone had to know why there were delays in shipping, gaping holes in the financial statements and questions from the board.
“I think what this is doing is forcing companies to rethink their supply-chain strategies and modernize them, versus if we went back three, four years ago, people didn’t really care,” Evans said. “Before Amazon, there wasn’t this much pressure to ship in two days, but now the supply chain is a critical business function for anybody that ships a physical good. And I think we are probably at the peak or close to that peak of demand for change.”
Let’s take a look at how hardware startups can think about de-risking their products and supply chains and figure out how to think about resiliency.
What do you want to be good at as a startup?
If you’re running a software startup, you’re probably outsourcing a huge amount of the tech your company depends on. It’s unlikely that you run your own servers, for example — you run that on AWS or one of the many other options. You probably use Twilio instead of figuring out how to send your own text messages, and if you need to collect money, you probably use something like Stripe, so you don’t have to learn the minutia of charging credit cards across lots of different banking systems, and so on.
The same goes for hardware startups, to a degree, especially if you are creating high-volume products. The strength of the company will be in product development and in marketing the products once they roll off the assembly line.
“The highest-valued company in the world is Apple. Do they manufacture anything? Nope. It says ‘designed in California’ — they design and market it. It is working out pretty well for them. I know I’m being a little cheeky here, but what I would say is that most companies don’t start because they want to be great at HR or accounting or supply chains. Some do, but most don’t. They want to be great companies,” Evans said. “I would say, focus on your product. Focus on how people discover you — marketing or sales. Everything else, de-risk it as much as possible. De-risk your HR so you don’t get sued by your employees or have legal issues. De-risk your supply chain so you don’t have a single point of failure and go out of business. De-risk your time to market by having faster cycle times and working with smart partners. If you do all of those things, your product survives, and you get to make a dent in the world.”
The time-to-market aspect of product development is worth dwelling on for a moment.
De-risking product development
When you are building hardware, it is all about getting the product to market as quickly as possible while de-risking the design process all the way through. In product development, the usual way of doing that is through a staged process called EVT/DVT/PVT.
The first stage is to make sure that it works the way you want it to work, to ensure that what you are trying to build is actually possible; this is known as an engineering validation test, or EVT. At this point, you’d figure out if all the features of a product work the way you want them to: Does it use the right amount of power, does it stay cool enough, and does every function of the device work the way you were expecting? At this stage, it’s a good idea to start talking to suppliers and contract manufacturers to figure out what the availability is of the components you’ve chosen.
It’s much easier to switch out a component at the EVT stage because you’re still at an early prototype phase, and swapping out one component is often not straightforward; you’d typically have to switch out a whole lot of them and then ensure that they play nicely together. A simple example: If a particular microchip that includes Wi-Fi on-chip isn’t as abundantly available as you had hoped, you’d need to add a second Wi-Fi chip, which may change how much power you need and might necessitate using a different type of power controller or battery.
The second stage is ensuring that it looks the way you want it to look: your design validation test, or DVT. Now you have all the functionality you want, but it also looks pretty. This is where you need to get the device tested to pass any certifications (CE-marking, FCC approvals, etc.). You’ll also want to figure out if the device is as weatherproof as you were hoping, whether it can withstand high and low temperatures, whether the finish fades in sunlight, whether it survives being dropped, how easily it gets scratched, and so on. From the supply-chain point of view, the question is whether all the fasteners and materials you want to use are easily available.
Amazon introduces AWS Supply Chain to help bring order to supply chain chaos
The final stage is the production validation test, or PVT. Ideally, throughout the process, you will have made choices to ensure the product can be mass manufactured (sometimes referred to as design for manufacture, or DFM).
For example, if you are a hobby electrical engineer, you could choose to use through-hole components for your printed circuit board because they are easy to work with and generally have generous tolerances. When you want to mass manufacture by the tens of thousands and beyond, however, you want components that can be placed and soldered by machines, and they tend to be surface-mounted. In any case, at the PVT stage, the rubber really hits the road, and you try to run a small batch of production to figure out whether the product can be easily manufactured.
“The analogy I love to give is that if I’m just a software company, the way I de-risk is to write some code, push it to production, and I find out almost instantly if something breaks or not,” Evans said. “In hardware companies, if you make a change and something breaks, that’s a problem. That’s why when we think about new product development and risk reduction, it’s all about speed. If you can get more iteration cycles to prove or disprove the risk, that makes a better product. If you can do more iteration cycles in a shorter amount of time, that’s going to get you to market sooner, and you’ll get to revenue sooner. You are able to get customer feedback faster, and reduce risk in a faster and more cost-effective fashion. It isn’t a question of ‘How cheap is your prototype?’ but ‘How much money did you have to burn to get that prototype to where you want?’ That’s how digital aggregators, such as Fictiv, give a lot of leverage to the startups of the world, because you’re able to access a global supply base without having to have all the connections themselves.”
Demand planning as supply chain risk reduction
If you always knew how many units you were going to sell of something, you could order the right number of gadgets, but that is notoriously hard, especially for new startups. You can make a plan for how you’re going to sell 10,000 speakers, but that costs a tremendous amount of money in design, prototyping, tooling and manufacturing. Then if you only sell 2,000, you have a warehouse full of products that will soon be outdated. The opposite might be even worse, at least from the CFO’s point of view.
“I might be like, ‘OK, I’m [going to] build 10,000 in year one,’ but then Haje writes an article on me on TechCrunch, and suddenly the demand skyrockets, all because of this amazing article, and you have the opposite problem,” Evans said. “Uncertainty is risk. Reducing uncertainty is reducing risk. If you can shorten the time it takes to make something, you can run a just-in-time strategy. That means that you can improve your cash outlay, and you’re making things based on what consumers actually want. That’s probably one of the biggest trend highlights right now, in fact.”
One of the big shifts in product design is the customization that’s happening — and there’s a lot more of it than five years ago. That is only possible with just-in-time supply chains.
“Just-in-time supply chains are the future,” Evans said. “As we look at digital and AI helping us, the whole tool stack improved. You won’t have to write off a billion dollars’ worth of additional supplier costs because you don’t have the right $20 chip, or whatever the issue was.”
Why Ford will have to eat an extra $1B in supplier costs this quarter
Fictiv focuses on the mechanical parts of the supply chain — plastics and metals, for the most part, and there are other suppliers who can be their counterparts for other parts of manufacturing. Macrofab, for example, can help with circuit boards and electronic components. Vention can help you build manufacturing automation more easily, and there are scores of other companies that have popped up to make hardware manufacturing much easier than it was.
“Our job is to simplify sourcing for customers, whether that’s small customers or big ones, and we focus on the custom and mechanical part of the supply chain,” Evans said as I took a sip of my coffee. “That cup you’re drinking from, I made all the prototypes for products like that. We don’t own any machines or factories, but we are helping you get the quote for the manufacturability you need. We have boots on the ground in Asia that help with the factory relationships, and we run a pre-vetted network of around 250 partners all around the world.”
Cottage industry in the year 2023
In the pre–Industrial Revolution days, there used to be “cottage industries,” where different parts of a product were made by hand or by simple machines by more and more specialized experts.
A pair of boots, for example, requires a lot of different skills. Making the soles is different from making the leather, which in turn is a different set of skills than making laces. Each of these processes was done separately at people’s homes (or small work cottages, hence the name) before the final boot was assembled. The Industrial Revolution changed that, and everything was put under one roof. Then in the 1950s through the1970s, a number of manufacturers started working on just-in-time-type systems, the most famous of which was the Toyota Production System.
Then the internet and advanced computer-aided design and manufacturing tools came along, and it suddenly became possible to have vastly more complex production and manufacturing systems that didn’t have to be in the same place.
“We’re still in these vertically integrated factories today, where you have the Ford example where the car can’t ship off the line because it doesn’t have the chip showing up on time,” Evans said. “I want to remind folks that we can now actually take geographically separate entities and have them be under a virtual factory, meaning you can start to run systems where the factories are distributed, but they are connected together through the internet. I think that’s where we are today, with history repeating itself as it always does, where we move towards a cottage-based industry connected with a digital thread. We are seeing this happen in cloud computing; one server fails, and another server in a completely different place picks up the slack. We now have fail-overs in manufacturing, too, so if a tsunami hits one region, it shuts down production, but it moves over to another region and it keeps moving. Then there’s a labor shortage somewhere, and you shift 30% of the work to somewhere else.”
The weakest link: Charting supply chain risk in an era of globalization
Of course, building that level of resiliency into manufacturing is exponentially harder than finding and working with certain factories in the first place; if you have a huge manufacturing operation, you might be able to build something like that out, but this is where we come full circle: What do you want to be good at as a startup? Running supply chains or creating and selling cool products?
“You need to ask yourself what your competitive advantage is. What is the industry secret that you know that nobody else has uncovered? If you’re gonna do that, that’s probably where your core competency is. If you’re creating a new product, you’re making something that Toshiba or Sony or LG or Amazon couldn’t do; that’s why you are tackling it,” Evans explained. “My favorite example is Juicero. That’s what happens when you hire a bunch of Apple engineers to go make a $700 juicer. Nine hundred mechanical parts later … well, what was the competitive advantage of this juicer? Why did Breville not make one? Why hasn’t Cuisinart done this? Your core competency is answering those questions, and to solve the pain point in the world for whatever you are trying to do.”
Fictiv has more than 300 people operating in four different countries at the moment, suggesting that they, too, didn’t get to where they are today by being good at everything.
“Focus on solving your users’ pain points and partner for everything else. Hold them accountable, but to survive, you have to focus,” Evans concluded.
Comment