Mechanical computer built from springs and metal bars proves basic computing without electricity

Researchers in the United States have assembled a mechanical computer from steel bars and springs that can perform simple computing tasks without any electrical power. This line of inquiry could pave the way for technologies that improve lives through more responsive artificial limbs or tactile environments, according to project notes.

Could you build a completely mechanical PC? The straightforward answer is no for a fully general computer, but it is possible to create devices that perform basic computations entirely without power. The project was highlighted by Interesting Engineering and conducted by teams at St. Olaf College and Syracuse University, with findings published in Nature under the paper “Mechanical hysterons with tunable interactions of general sign.” The platform relies on interconnected metal bars and springs to form its computational primitives.

Joey Paulsen, Associate Professor of Physics at St. Olaf College, explains that memory isn’t confined to hard drives or brains. Everyday materials can remember past states too—for example, how much a rubber element has been compressed or stretched. The researchers asked whether such materials could also process information, not just recall past states, effectively performing computation with no external power.

To date, the team has built three mechanical computers. One can count up to three, another can decide whether a sequence of pushes is odd or even, and a third can remember whether a medium or large force was applied—all without any electrical energy.

These early devices are basic, but they demonstrate a concept with potentially broad implications. A key next step is scalability: scientists are examining how the state of one rotor influences a second rotor to pave the way for a third rotor and more complex operations. Mechanical computing could find real-world use in harsh environments where traditional silicon-based systems falter or fail, such as extreme heat where chips would melt.

Paulsen notes that these results are a step toward materials that sense their surroundings, make decisions, and respond accordingly. Such smart materials could eventually lead to more capable assistive devices or environments that adapt to users in real time.

The idea isn’t entirely new. A few years ago, researchers at North Carolina State University explored a mechanical computer design built from plastic cubes that are pushed and pulled to input, store, and manipulate data. It’s another reminder of how physical systems—like a Lego-inspired computer—continue to inspire unconventional computing approaches.

With ongoing work on rotor interactions and multi-rotor networks, these mechanical platforms could become a foundation for resilient computing technologies in contexts where electricity or traditional semiconductor devices aren’t viable.