In 2003, a group of Linux developers discovered that the Linksys WRT54G router, a $60 plastic box sold at Best Buy, was running Linux under the hood. Linksys had violated the GPL by not releasing the source code. The legal pressure forced them to comply, and the source code that came out sparked a community that would spend the next two decades reverse-engineering every router they could get their hands on, stripping out vendor firmware, and replacing it with software that actually worked.
Twenty years later, that community shipped the OpenWrt One. It is the first router designed from the ground up by the OpenWrt project, for the OpenWrt project. It is not a very good router by the standards of the consumer networking market. Its WiFi is WiFi 6, not 7. It has no 2.5GbE ports. By the numbers, a $70 TP-Link box from 2025 will beat it on throughput. But the OpenWrt One is not competing on throughput, and that is the whole point.
Open hardware has a terrible track record. OpenMoko tried to build an open phone in 2007 and failed because it tried to compete with the iPhone on specs. The Novena laptop was a beautiful dead end. Every attempt at an open smartphone has fizzled. The pattern is consistent: someone decides to build open hardware, tries to compete with the mass market on the mass market's terms, and loses. Consumers do not care about open firmware. They care about whether the camera is good.
OpenWrt One works because it does not make this mistake. It was not designed for consumers. It was designed for the person who has bricked a router by flashing experimental firmware at 2am and knows exactly where the serial console pins are on a WRT54G v1.1 motherboard. That person is not a market segment in any spreadsheet. But they are a real community that has been alive for two decades, and they knew exactly what they wanted because they had been living without it since 2003.
What they wanted was a router with a physical recovery switch, a serial console accessible from the case, a NOR flash recovery chip, and silicon that the OpenWrt developers had already written drivers for. Consumer routers compete on the spec sheet. Community hardware competes on driver maturity and longevity. The boring choice is the radical choice because it means the hardware will still be supported in 2030.
The supply chain reality is worth noting. Most open hardware projects die at the 1,000-unit minimum order phase. That Banana Pi tooled up production for a community-designed router with niche features like a recovery switch tells us that contract manufacturing for small-batch open hardware has become viable in ways it was not ten years ago. The economics work because the OpenWrt One does not compete on price either. It is a more expensive router that saves you time, and its customers value hours over MHz.
There is a deeper shift here. The WRT54G existed because consumer router hardware was built on a thin margin and the only way to differentiate was to slash costs, which meant using off-the-shelf Linux SoCs. That created the conditions for the OpenWrt project to exist. But that same market dynamic is now squeezing manufacturers so hard that the margin is zero at the low end. Everyone competes on WiFi throughput specs with the same MediaTek chips. The OpenWrt One targets the niche that values what the market no longer measures: trust.
Trust is the spec that cannot be written on a box. The OpenWrt One trusts its owner. It has a serial port that does not require disassembly. It has a dedicated unbricking flash that cannot be overwritten by bad firmware. It is the first router designed for the person who breaks things, not the person who buys them at Best Buy. The owner is trusted to SSH in, to flash an experimental kernel, to fix what they broke, to learn.
What comes next is the interesting question. The Nerves meetup is already featuring a project that uses OpenWrt One as its hardware platform. Nerves is an Elixir framework for embedded devices, and an Elixir community using a router designed by the OpenWrt community is the first sign that OpenWrt One is becoming a general-purpose open hardware platform other projects build on. It is not just a router for OpenWrt users. It is the first piece of consumer-grade hardware that assumes the person using it has root.
There is also a paradox. OpenWrt's whole reason for being is that routers ship with locked firmware. What happens when the hardware is already open from the factory? If every router shipped with the GPL source and an unlocked bootloader, the OpenWrt project would have less to do. But the market is not going that way. The pressure on router margins is driving manufacturers to lock down even harder, to extract more value from firmware, to treat the device as a disposable service endpoint. The OpenWrt One is the countertrend, and it works because it is aimed at the people who hate the trend enough to pay for the alternative.
The 20-year arc from a GPL violation lawsuit to shipping your own hardware is not about open source politics or ideological purity. It is about what happens when a community that has been fixing other people's broken products for two decades finally gets to build their own. They did not try to build something for everyone. They built the router they always wanted, and it turns out there were enough of them to make it real.