Chapter 4: Building Hardware in Shenzhen When You Know Everyone and Still Have No Idea What You're Doing

People outside China have a version of Shenzhen in their heads that is half right.

They know it's where things get made. They know it's fast. They know that if you need a component, a supplier, a factory tour, a prototype run, you can make it happen here in days that would take weeks or months anywhere else.

Shenzhen Moves Faster Than Plans

The ecosystem is real and it's as good as advertised. MIT Technology Review described it as not a singular geography but a globally interconnected one, and that's accurate: its factories, component suppliers, service providers, and skilled workforce form something that took decades to build and remains genuinely difficult to replicate anywhere else on earth.

What they don't always understand is what it actually feels like to operate inside it. Especially when you've spent years here and you know the people and the rhythms and the shortcuts, and you're still, on your best days, improvising. Shenzhen moves faster than any plan you make for it. That's its greatest strength and the thing that will humble you if you're not careful.

We had advantages that most hardware startups don't have. Terrence, Elisa, Huilong, the industrial designers we'd brought in, all of us had histories in this ecosystem. We had supplier relationships, factory contacts, people we could call when something wasn't working and get an honest answer instead of a polished one. We knew how to read a production sample. We knew the difference between a problem that was cosmetic and a problem that would kill a launch. We thought we knew what we were doing. We were right about that, mostly. And we were surprised anyway.

The 3D printed prototype was a beginning, not an answer. Getting from a 3D print to a production-ready sample is a process that compresses and stretches time in ways that are difficult to explain until you've been through it. Some things move faster than you expect. A material decision that you thought would take two weeks of back-and-forth gets resolved in a single factory visit because the floor manager has a better idea and he's right. Some things move slower. A tolerance that looked fine on the model doesn't survive the transition to injection molding and now you're three weeks behind because the tooling has to be adjusted.

From 3D Print to Production Sample

You learn, during this process, what the product actually is. Not what you designed it to be. What it is. The object has opinions. It tells you things. The weight distribution you specified feels different when it's made from the real material at the real scale. The microphone placement that worked acoustically creates a problem with the band attachment. The button position that made sense on a screen makes less sense on a wrist.

Every one of these discoveries is a small crisis and a small education. You solve it or you accept it or you go back and redesign the thing that caused it. You do this dozens of times. And the product that comes out the other side is better than the product you started with, not because your original vision was wrong, but because the process of making something real is itself a form of design.

The supplier transition, the one that cost us time in early 2026, is worth saying more about. When you lose a manufacturing partner mid-process, the first thing you feel is the schedule. Everything you'd planned downstream, the sample run, the beta units, the pre-order timeline, the launch, all of it shifts. And in a startup, schedule is not an abstract concern. It's connected to capital. To team morale. To the commitments you've made to early supporters who are waiting to hold the thing in their hands.

We had beta founding members who'd signed up because they believed in what we were building. We had a community forming around the product. The idea of telling people it was going to take longer was not something any of us wanted to sit with. And so we made the mistake that pressure tends to produce. We stopped thinking and started moving.

The Supplier Transition

Chinese New Year was approaching. The holiday is the longest pause in the Shenzhen calendar, the point at which factories go quiet, supply chains freeze, and everyone goes home. For a hardware startup mid-launch, it feels like a wall. Whatever isn't done before the holiday waits until after it.

I felt that wall coming and I pushed. I pushed the team, I pushed the timeline, I pushed our strategy, and I didn't stop to ask whether the strategy was right. We defaulted to what we knew. We'd all done this before at Anker: presale, influencer outreach, media samples, paid advertising on platforms we understood. That playbook had worked for cables and batteries and charging docks. We knew how to execute it. So we executed it.

The ads went live. The cost per acquisition came back far higher than it should have been. Not slightly off, seriously elevated. A number that told us something was wrong, not with the execution but with the underlying assumptions. At the same time I kept feeling something I didn't have a clean name for: that the product-market fit wasn't fully proven yet. That we were pushing toward market with a question we hadn't fully answered.

It all felt wrong. The speed felt wrong. The channels felt wrong. The whole posture felt wrong. The day before the holiday, I called the team into a meeting room and said the thing that needed to be said. We have to chill. Not give up. Not pivot. Chill. Step back. Go back to the drawing board and look honestly at what we were doing and why.

We Have to Chill

The problem, I told them, was that we'd been treating Memoket like a hardware product. Like a battery. Like a cable. Like something with an established category, an established customer, an established path to market. And Memoket is none of those things. It's a new category. Entirely new. The people who need it most don't have a word for what they're missing yet. You can't reach them the way you reach someone who already knows they want a faster charger.

No one was pressuring us to go to market. No investor was banging on the door. No deadline existed except the ones we'd imposed on ourselves. We'd manufactured urgency out of competitive anxiety, watching other AI wearables enter the market and feeling like we had to match their pace. But their products weren't our product. Their customers weren't our customers. Their strategies weren't our strategies. I told the team to go home. Rest. Eat well. Sleep. Come back after the holiday with fresh eyes and an open mind.

We came back different. Not because the holiday solved anything, but because the distance did. When you've been inside something for months, moving fast, reacting, executing, you lose the ability to see it clearly. A week away gave us that back. What we saw, looking at it fresh, was that we had something genuinely new on our hands, and we needed genuinely new thinking to bring it to the right people. No crowdfunding. No Amazon listings. No traditional retail channels. None of the paths that work for established hardware categories, because Memoket wasn't an established hardware category. It was something the market didn't have a shelf for yet.

The new rule became: try everything. Not randomly, not without judgment, but without the constraint of doing things the way they'd always been done. Find the right people. Figure out where they are. Go there, in whatever form that takes.

Getting Out There

There's a Cantonese phrase I'd grown up hearing that came back to me during that period: 出来行,最紧要系"出来" — which translates roughly as: "If you're going to walk the streets, the most important thing is to actually get out there." It's a saying about action over hesitation, about the futility of waiting for perfect conditions before you move. The street doesn't reward the person who plans the walk forever. It rewards the person who takes the first step. That phrase became our motto.

Not recklessness. Not the frantic, unfocused energy of the weeks before the holiday. Something more grounded: the willingness to try things we'd never tried, in places we'd never looked, without the safety net of a playbook that had worked somewhere else before. We had the product. We had the algorithm. We had a team that knew how to build things. What we needed now was to find our people. That turned out to be its own education entirely.

China's manufacturing advantage gives hardware companies a unique edge in AI consumer electronics, one that most Western companies cannot match at speed or scale. That advantage is real and we felt it every day in Shenzhen. But manufacturing advantage only takes you so far. At some point the product is made, and the harder question begins: who is it for, and how do you find them?

The answer, we were learning, was not going to come from the channels we knew. It was going to come from somewhere we hadn't looked yet. We just had to get out there and find it.

Memoket Gem captures your conversations so you can stay present in them. See how