Li Wei sat alone in the Forge Lab that night, sketching out the architectural blueprint for the ForgeStation v1, a self-contained development terminal.
He wasn't thinking of profits anymore.
He was thinking of infrastructure.
If he wanted real innovation — not just devices cobbled together from existing parts — he needed full control over his own computing environment. Operating system. Compiler. Instruction set. Fabrication tools.
He needed to free himself from dependence on foreign chips, legacy software, and internet censorship.
That meant starting from the silicon layer upward.
ForgeStation: The Minimalist Dream
The ForgeStation v1 wouldn't be a full PC.
It would be a micro-terminal:
A small 4-inch monochrome display
A custom keyboard with programmable macros
Storage via EEPROM or SD card
Running ForgeOS Alpha, their scratch-built system with a minimal kernel and scripting shell
The goal was to create something that could:
Compile basic programs
Connect to sensors or motors
Operate independently of commercial platforms
"You're building a typewriter with attitude," Lan Jie joked when she saw his sketches.
"A typewriter that can launch rockets," he replied with a grin.
Materials, Materials, Materials
His biggest bottleneck wasn't code — it was materials.
Modern electronics needed:
High-purity silicon wafers
Gold or copper traces
Ceramic or tantalum capacitors
Rare earths for advanced components
He had none of these.
What he had was access to:
Industrial scrap from old appliances
Defunct PCBs pulled from garbage bins
Some raw copper, tin, and aluminum from junkyards
That was enough to start.
With Lan Jie's help, he set up a micro-refining rig in the back of the lab — an electric crucible to purify recovered copper and plate it onto salvaged boards.
It took them three days to produce a single palm-sized copper-layered PCB blank. Not elegant, but usable.
"This is painful," Lan Jie muttered.
"This is reality," Li Wei said. "We suffer now so we own the future."
The Storage Problem
While BlueFire worked, it lacked decent non-volatile memory.
They couldn't build a system if the OS disappeared every time power cut off.
Yanyue suggested ordering second-hand EEPROM chips from an old warehouse supplier. They were slow and small — only 256KB — but cheap.
Li Wei agreed, as long as they could get at least 50 units.
The order arrived five days later. Half were dead. A quarter were glitchy. Only 12 worked perfectly.
"We build with what we've got," he said, labeling the working ones with a permanent marker.
They programmed a minimal bootloader — just enough to initialize the board, load a kernel, and launch the ForgeShell interface.
It was crude.
But it booted.
Naming the Kernel
"Why don't we call it something catchy?" Lan Jie suggested. "Like NovaOS or DragonShell."
Li Wei shook his head. "This isn't a brand. It's a tool."
"Fine. What's it called?"
He opened the kernel boot log:
yaml
Copy
EditFORGEOS v0.01 Kernel Init: 2040.03.04 LiWei-Core Build
"ForgeOS," he said. "Built to endure."
They uploaded the OS to a working EEPROM. Then watched the terminal boot up in just 2.3 seconds on BlueFire.
No OS in the world ran that light.
It couldn't multitask. It couldn't connect to the web.
But it worked.
Spreading the Flame
By mid-March, they had ten ForgeStation kits built using:
BlueFire logic controller
ForgeOS boot EEPROM
EEPROM burner (handmade)
Assembly guide + tutorial
Cost: $12/unit in materials
They distributed three units to a local technical school and two to Zhou's workshop for testing.
They didn't take money. Instead, they asked for:
Feedback logs
Photos of use cases
List of component failures
A week later, Zhou sent back:
"Student used it to automate a chicken incubator. Display showed humidity and temperature. Very impressed."
Another student wrote:
"Used ForgeOS to control LED light patterns and buzzer alerts. Wrote my first script. Want more memory!"
Whisper in the Underground
Li Wei didn't advertise.
But word spread.
A few electronics hobbyists reached out through their anonymous bulletin board.
"Can I join your dev group?"
"Looking to help with PCB routing."
"I have an old milling machine. Want to help?"
Li Wei didn't build a community.
A community found him.
One by one, strangers began contributing:
Translating the manual into Vietnamese, Bahasa, and Hindi
Building test rigs and emulators
Creating replacement part lists based on regional supplies
By April, the Forge Society had grown from three to fifteen active members, most of whom Li Wei had never met in person.
"Decentralization is our armor," he told Lan Jie. "We never centralize. Never rely on a single factory or channel."
Looking Ahead
With stable memory and a bootable terminal, the ForgeStation v1 became their first true platform — a base from which all future technology would be built.
But Li Wei knew what came next would be even harder:
Designing their own instruction set
Building a compiler that didn't depend on GCC or LLVM
Creating a self-contained ecosystem for communication, encryption, and expansion
It would take months — maybe years — before they reached anywhere near modern computing power.
But in a world already relying too much on outsourced tech and crumbling global supply chains, they were planting the seed of sovereignty.
One board at a time.