Electronics and robotics, in order. Every one of them took something that belonged to a few institutions and handed it to everybody. That is the whole story, and it is the thing we are trying to do next.
Thirty tons, eighteen thousand vacuum tubes, a room of its own and a staff to match. Reprogramming it meant physically rewiring it for days. Computation existed; access to it did not. For twenty years, using a computer meant belonging to an institution that owned one.
A germanium slab, a plastic wedge and two gold foil contacts. It is the most consequential object on this page, because everything after it follows from one fact: this could be made smaller, and then smaller again, more or less indefinitely.
Kilby put more than one component on a single piece of semiconductor. That is the whole idea. Every chip since is that idea repeated at a scale the people in the room would not have believed.
The first industrial robot went to work on a General Motors line lifting hot die castings — a job that hurt people. It is where the argument starts about what machines should do instead of us, and it has been running on factory floors ever since.
Four kilobytes of RAM, programmed by women weaving wires through magnetic cores by hand, and it landed people on the Moon. The most ambitious project in engineering history ran on less compute than a doorbell — and it worked because the people involved understood every part of it.
2,300 transistors, and for the first time a computer's whole processor was one part you could buy. Not a machine built for one purpose — a general-purpose engine sold by the tray. The microprocessor turns computing into a component.
A bare board sold without a case, keyboard or power supply, to people expected to finish it themselves. Two hundred were made. Its importance is the assumption behind it: that one person might own a computer, and that the interesting work would be done by whoever bought it.
Built from parts anyone could buy, with a published specification. IBM meant to sell computers and accidentally created an open platform that thousands of companies could build on. The hardware became a standard, and the standard is what actually scaled.
The first robot tens of millions of people actually let into their homes. Not a demo, not a research platform — a product, at a price, doing a chore. Robotics stopped being a laboratory discipline the day one of these shipped in a box.
An open board and a simple toolchain, and suddenly a designer or an artist or a fifteen-year-old could make something move. Arduino did not advance the state of the art. It collapsed the cost of entry, which turned out to matter far more.
A supercomputer, a camera, a radio and a full sensor package, made in volumes so large that every component inside it became cheap for everybody else. The accelerometer in a hobby project costs what it costs because a billion iPhones needed one first.
A whole computer for the price of a textbook, with pins on the side you can wire things to. Over sixty million sold. It put a real Linux machine into projects, classrooms and products that could never have justified one before.
A first stage comes back and sets itself down intact. The engineering is remarkable; the economics are the point. Reuse cut the cost of reaching orbit by more than an order of magnitude in a decade, and turned space from something nations do into something a university can afford.
A legged robot you can buy, that walks itself over ground no wheeled machine can cross. Forty years of locomotion research turned into a product with a price list. This is what the far end of hardware looks like when it finally ships.
Income per person, the price of light, transistors per chip, the cost of reaching orbit, the cost of a prototype circuit board. Five curves, one direction. It has never been cheaper or easier to build a physical product than it is right now.
Flat for eight centuries, then straight up.
Hours of work for an hour of light. Down 40,000×.
Eight orders of magnitude in one working lifetime.
$120,000 to $1,500 a kilogram.
Five two-layer boards, delivered. $1,200 in 1995. $9 today. The factory does not care who you are — the quote is the same.
Demand for designed hardware is up two hundredfold. The number of people who can design it has gone down. That gap is the whole reason Synqit exists.
We have watched software do it. Every curve on this page bent upward the moment the tools stopped asking permission — a compiler, a cheap board, an open standard, a $35 computer. Hardware has never had that moment.
It is overdue. And it is the only thing standing between a person with a real problem and the object that solves it.
Now it is hardware's turn. That is what Synqit is for.