Meet the Computer That Has Never Touched an Electron

Picture a machine that can add two numbers, remember a decision, and steer a rocket's flight path. Now take away every wire. Take away every transistor, every silicon chip, every volt of electricity running through it. What's left?

A whisper of air, moving through a maze of tiny channels, doing all the thinking.

That's not a thought experiment. It's a real technology, born in a U.S. Army laboratory in the late 1950s, and it still hums quietly inside washer nozzles, ventilators, and old missile guidance systems today. It's called fluidics, and it forces a question we rarely think to ask: does a computer actually need electricity at all?

What Fluidics Actually Is

Fluidics is the science of using a moving stream of liquid or gas to carry and process information, the same job electrons do inside a silicon chip. Instead of copper wires, fluidic circuits use narrow channels. Instead of transistors, they use fluidic valves and jets, precisely shaped chambers where one stream of fluid controls, redirects, or blocks another.

No electrons cross these circuits. No current flows anywhere in them. Everything happens because of pressure differences and the ordinary physics of moving fluid, air, water, oil, even exhaust gas, racing through channels no wider than a drinking straw.

Pressure Is the New Voltage

To understand fluidics, just swap out the vocabulary. In an electronic circuit, voltage pushes electrons through a wire, and current is the rate at which they move. In a fluidic circuit, pressure pushes fluid through a channel, and flow rate takes the place of current.

That single swap explains almost everything fluidics can do. A region of high pressure behaves like a battery. A narrow channel behaves like a resistor. And logic gates, the AND, OR, and NOT operations that let a computer make decisions, get built by aiming two fluid jets so they collide.

Here's where it gets interesting. When two jets strike each other at an angle, the combined stream deflects toward one output channel or another, depending on which jet is stronger. Shape the geometry correctly, and that deflection becomes a decision. Two jets firing at once might push the flow into an "AND" output. Either jet alone might trigger an "OR" output. Use a small side jet to block the main stream, and you've built a "NOT" gate. Nothing clicks or slides into place. The fluid does the arithmetic of logic simply by colliding with itself inside a carefully sculpted chamber.

The Trick That Makes It Work

None of this would be nearly as useful without one strange habit of moving fluid that gives fluidics its real switching power: the Coandă effect.

Named for Romanian inventor Henri Coandă, who patented a device exploiting it in 1934, the effect describes a jet's tendency to cling to a nearby curved or angled surface instead of shooting straight into open space. As the jet travels, it drags along the surrounding fluid, a process called entrainment. That entrainment lowers the pressure in the thin gap between the jet and the wall. Lower pressure on one side pulls the jet sideways, so it hugs the surface, sometimes curving around it entirely before finally letting go.

Fluidic engineers turned that quirk into a switch. Build a chamber shaped like the arms of a Y, and a jet fired down the middle will lock onto one wall almost at random, held there by the Coandă effect. Fire a small puff of air from a side channel, and you can knock the jet loose, sending it snapping over to the opposite wall, where it stays until nudged again. A jet that stays put until deliberately flipped is a bistable switch, the direct cousin of an electronic transistor or a memory flip-flop. Engineers at the Diamond Ordnance Fuze Laboratory (later renamed Harry Diamond Laboratories), notably Billy Horton and Raymond Warren, patented this wall-attachment amplifier around 1959, and it became the foundation of practical fluidic logic.

Where Silicon Fails, Fluidics Thrives

Why build a computer this way, when real electronic chips are staggeringly faster? Because there are places where electronics simply give out, and fluidics doesn't.

Consider intense heat. Silicon transistors start misbehaving long before they melt; their electrical properties drift once temperatures climb into the hundreds of degrees Celsius. A fluidic circuit, machined out of solid metal, keeps switching jets around as long as fluid keeps flowing through it.

Consider radiation. Ionizing radiation scrambles the delicate charge states that silicon memory depends on, which is exactly why spacecraft and nuclear facilities need expensive, radiation-hardened electronics. A fluidic circuit has no charge states to scramble. There's nothing there for radiation to corrupt.

Consider electromagnetic interference, including the pulse from a nuclear detonation, which can disable unshielded electronics across an entire region in an instant. Fluidic logic, having no electrical component whatsoever, doesn't even notice.

And then there's the matter of sparks. Inside an oil refinery, a pharmaceutical plant, or an explosives factory, one stray spark from a short circuit can trigger a catastrophe. Fluid circuits generate none. They can sit inside a room full of volatile vapor and keep computing without the faintest risk of igniting anything. Add in a practical bonus: with few or no moving mechanical parts to wear down or jam, a well-machined fluidic device can run for years with almost no maintenance.

Born of the Cold War

Fluidics didn't come out of a university curiosity project. It came out of Cold War anxiety.

Through the early 1960s, American engineers worked to build control systems that could survive a nuclear environment, one where an electromagnetic pulse might disable every conventional circuit in a missile's guidance package within microseconds. Fluidic amplifiers, immune to exactly that threat, looked like a serious answer. The first international symposium devoted to the field convened in Washington, D.C. in 1962, and the young discipline finally had a name: fluidics.

Soviet engineers pursued the same idea in parallel, developing their own pneumatic and fluidic control systems for industrial automation and military hardware, prizing the same ruggedness that let these circuits survive where electronics couldn't. Neither side was chasing elegance. They were chasing circuits that heat, radiation, and an enemy's first strike could not break.

Still Flowing, Decades Later

Fluidics never became the dominant computing technology; silicon won that race decisively. But it never vanished either. It just found the jobs silicon still struggles with.

Spacecraft and missiles: fluidic amplifiers saw early use in missile autopilot systems and artillery fuzes, where resistance to shock, vibration, and radiation mattered more than raw speed. Engineers at NASA and the U.S. Air Force also flight-tested purely fluidic flight-control systems aboard experimental aircraft during the 1970s, exploring them as backups immune to electrical faults.

Lab-on-a-chip diagnostics: microfluidics, the miniaturized descendant of classic fluidics, now drives many medical diagnostic devices that separate, mix, and analyze tiny fluid samples on a chip the size of a postage stamp. Researchers have pushed the logic side further still. In a study published in Science in 2007, Manu Prakash and Neil Gershenfeld at MIT demonstrated a bubble-based logic system, in which precisely timed bubbles traveling through a microfluidic channel perform digital operations, the same jet-and-collision principles fluidics pioneers worked out decades earlier, shrunk down to a scale visible only under a microscope.

Automotive fuel and washer systems: fluidic oscillators, tiny chambers that make a jet flap back and forth entirely on its own, show up in automotive fuel injectors and, far more commonly, in windshield washer nozzles, fanning a stream of washer fluid into a wide spray using nothing but the shape of the nozzle itself. No motor, no moving parts, no battery required.

Non-electric ventilators: because they generate no sparks and need no power source, purely pneumatic ventilators built around fluidic logic elements have been used in operating rooms and field hospitals, places where an electrical fault near pressurized oxygen is not a risk anyone wants to take.

The Catch

None of this makes fluidics a rival to your laptop, and it's worth being honest about why.

Speed is the first wall. Electrons move through a wire at a meaningful fraction of the speed of light. Fluid, even fast-moving fluid, obeys ordinary fluid dynamics, viscosity, inertia, turbulence, all of which cap how quickly a fluidic switch can flip states. A modern processor executes billions of operations every second. A fluidic circuit is fortunate to manage a few thousand.

Miniaturization is the second wall, and arguably the harder one. Modern chips pack transistors just a few nanometers across, smaller than most viruses. Push a fluidic channel down anywhere near that scale, and the physics itself resists you. Viscosity dominates over inertia at tiny scales, flow becomes difficult to control with precision, and manufacturing channels that small, reliably and cheaply, remains a genuine struggle even in today's best microfluidics labs.

Not a Replacement. A Specialist.

Fluidics was never meant to outcompete electronics on its home turf. It's what engineers reach for exactly where electronics quietly, predictably fails: deep heat, hard radiation, explosive atmospheres, anywhere hostile to a charge-carrying circuit.

That specialist role isn't shrinking. It's mutating. Researchers in biotechnology increasingly treat the fluid channel itself as the computer, using microfluidic logic to sort cells, screen drugs, and run diagnostic assays without a single silicon component ever touching the sample. Soft robotics researchers, building robots from silicone and rubber instead of metal and motors, have started borrowing fluidic valve designs to let a robot's own limbs "decide" how to move using nothing but internal air pressure, no onboard processor required.

Here's the vocabulary worth keeping in your back pocket, since fluidics essentially translates the entire language of electronics into the language of flow.

Electric current - Fluid flow

​Voltage - Pressure

​Wire - Pipe or channel

​Transistor - Fluidic switch (wall-attachment amplifier)

​Battery - Pressure source or pump

​Resistor - Narrow channel or orifice

​Capacitor - Flexible chamber or fluid reservoir

And yet something about fluidics still refuses to sit still, scientifically speaking. Engineers can explain, in exhaustive mathematical detail, why a jet clings to a wall under the Coandă effect. What they still can't fully predict, without heavy computation and a fair amount of trial and error, is exactly when a given fluidic switch will flip under messy, real-world turbulent conditions, because turbulence itself remains one of the great unsolved problems in classical physics. Somewhere in that gap, between a phenomenon we can harness and one we can fully model, an ordinary jet of air is still making decisions nobody can predict in advance.