Scientists Built a Working Computer Out of Springs That Doesn't Use a Single Watt of Electricity: This bizarre mechanical computer has no wires or chips.

It has no wires, no silicon chips, and needs zero electricity. Yet, it computes. Researchers from St. Olaf College and Syracuse University have built a functioning computer using only rigid steel bars and springs.

This battery-free machine performs basic logic and memory tasks using only physical components. It transforms abstract physics into a tangible platform capable of processing information through sheer mechanical movement and tension.

But why build a mechanical computer today? After all, engineers built magnificent, brass-geared mechanical calculators, which reached their heyday during World War II. Those old-school machines, which faded away after the invention of the transistor, were technically vastly superior at raw number-crunching than this 2026 version.

The difference lies in the ultimate goal. The World War II machines were dedicated, standalone mechanical calculators built to solve equations for humans using fixed, continuous gear ratios. This new spring-based platform is radically different. It attempts to mimic the messy, non-equilibrium physics found in crumpled paper or amorphous solids.

Rather than building a better calculator, the scientists are trying to create the fundamental building blocks for a new generation of smart materials. They want to give everyday structures the innate ability to compute and adapt to their physical environment.

You might wonder how a piece of metal can possibly store information.

“We typically think of memory as something in a computer hard drive, or within our brains,” said Joey Paulsen, an associate professor of physics at St. Olaf College.

“However, many everyday materials retain some kind of memory of their past, for example, rubber can ‘remember’ how far it has been squeezed or stretched in the past,” Paulsen explained.

The research team set out to harness this type of mechanical logic. Led by Paulsen, a team of St. Olaf undergraduates — Faten Abu Al Ardat, Harry Maakestad, Alex Walk, and Jack Feider — built bistable mechanical units called hysterons.

Each hysteron is remarkably simple. It is a rigid bar that swivels on a pivot, corralled between two stops. A spring links the bar to a sliding rod. When you pull the rod, the bar holds its position until the force crosses a specific threshold. Then, it violently snaps into a new state.

Push the rod back, and it requires a completely different threshold of force to snap back. This gap between switching points is called hysteresis. It means the current state of the bar depends entirely on what just happened to it.

The Power of Mechanical Frustration

A single snapping bar is just a switch. True computing requires interaction between these on and off switches.

The researchers linked multiple hysterons together using additional springs. This is where the platform drastically separates itself from historical mechanical computers.

By tuning the springs, the researchers created highly complex interactions between the bars. If they left the springs uncrossed, the bars “cooperated” and wanted to point in the same direction. If they crossed the springs, the interactions became “frustrated,” forcing the bars to seek opposite states.

Even more bizarrely, they engineered non-reciprocal interactions. This means one bar could strongly dictate the movement of its neighbor, without the neighbor having much influence in return. This dynamic tunability is something classical gear-based computers simply could not achieve.

“We now have a rational way of building these machines that can perform simple computations without a computer chip or a power source,” said Paulsen.

Counting, Latching, and Logic Gates

To prove their springs could actually do computing operations, the team built three distinct computers.

The first machine acted as a physical counter. It tracked how many times a user pulled its drive rod. A chain of hysterons acted as a moving boundary, shifting one step down the line with every half-cycle of motion.

The second machine functioned as a logic gate that counted modulo 2. It could successfully distinguish between an odd or even number of inputs. Using four interconnected bars, a repeated physical push forced the system into a repeating two-cycle rhythm.

The third machine performed a strange trick called latching. It locked into a changed state after a medium push, but reset entirely when subjected to a massive, larger push. The machine effectively stored information about the intensity of the force applied to it.

Surviving the Extremes

No one is saying these clunky machines will replace highly efficient silicon chips. That’s not the point. However, there are select applications where a spring-based, basic computer could prove useful.

Silicon chips are incredibly fragile. They melt in extreme heat, fail under heavy radiation, and quickly dissolve in corrosive chemical vats.

These new mechanical computers, however, harvest their energy directly from physical force. They require absolutely no electricity. Because they find their intelligence purely in the tension of springs and the movement of steel, they are highly durable. So, they could easily survive inside a vibrating jet engine or on a probe exploring a hostile planet.

But the true prize lies much closer to home.

“Our results are one step toward designing materials that can sense their environment, make a decision, and then respond,” said Paulsen.

If computation can be baked directly into the geometry of a material, it opens the door to a new world of physical design.

“Frequently called smart materials, what we learned could help improve people’s lives by having more responsive artificial limbs or tactile rooms,” Paulsen noted.

A Stepping Stone, Not a Finished Product

The researchers caution that scaling up these spring-based systems is highly difficult. Even to get the four-bar modulo-2 machine working, the team had to iteratively adjust spring mounting positions and post angles. It was a grueling process that the authors likened to supervised machine learning.

Currently, St. Olaf students are continuing to investigate how multiple rotors influence one another, hoping to build larger, more complex networks.

For now, the device remains beautifully simple and serves more like an academic demonstration. The device counts. It sorts odd from even. It remembers the strength of a push. And it does it all with springs, proving that matter itself is capable of making decisions.

The findings appeared in Nature Communications.

This article originally appeared in April 2026.