Researchers at Maynooth University in Ireland have demonstrated a DNA computer that performs calculations as it settles into thermodynamic equilibrium, executing more than 700 individual computations across 10 different programs — including addition, multiplication and division — on a single reusable molecular system.

The team, led by Damien Woods, reports the work in Nature. Simple computations completed in as little as one minute, though larger programs took considerably longer. In one test of durability, the researchers repeated an experiment after 15 months: they added water to a partially dried setup and still obtained correct answers.

The result is notable because conventional computers — and most molecular computers built to date — operate far from thermal equilibrium. They consume energy to maintain and switch encoded bits. Back in the 1970s, IBM physicist Charles Bennett showed that any computation can in principle be made reversible, allowing a machine to run close to equilibrium and cut its energy demand sharply. Subsequent work pushed the idea further: a system could deliver an algorithm's output in its equilibrium state itself, drifting naturally toward the solution as it reaches its most stable configuration.

"The idea presents multiple challenges that include finding a physical implementation that is computationally expressive and programmable, has easily prepared initial states and has a controllable energy landscape for rapid navigation to target outputs with high probability," Woods explains.

Earlier DNA computing efforts mostly relied on out-of-equilibrium systems, driven either by molecular fuel or by carefully prepared initial states under tightly controlled conditions. Woods' group took a different route, building on two established lines of research.

The first is DNA origami, which Paul Rothemund at Caltech demonstrated in 2006: a long single-stranded DNA scaffold folds into chosen shapes as it settles into equilibrium, guided by a set of shorter binding strands. The second is molecular tile competition, which Erik Winfree devised in the 1990s.

The Maynooth team combined both. They engineered a DNA scaffold carrying a designed sequence of binding domains and mixed it into a solution of shorter strands — "tiles" that represent the possible values at each step of a calculation. The tiles compete to bind to the scaffold. At each binding domain, the winner is the tile that binds most favourably both to the scaffold and to its neighbouring tiles. Because DNA strands follow simple, predictable binding rules, the researchers could engineer the energy landscape directly: mismatched tiles detach and better-fitting ones replace them until the system reaches its most stable state.

"Eventually, the system settles down into its energetically preferred state which encodes the answer to the computation: a sequence of tiles each bound to the scaffold and to neighbouring tiles on its left and right," Woods describes. "The competitive process of binding executes the computation."

The setup is operationally simple. Running a computation required mixing all strands together, heating the solution and letting it cool — no molecular fuel, no specially prepared initial states.

Reuse worked as well. "The same molecular computer can be reused to perform new calculations, running the same program up to 25 times on different inputs," says Maynooth's Abeer Eshra.

Constantine Evans, also at Maynooth, frames the significance in terms of reliability rather than speed. "Our system uses just a handful of different kinds of molecules, never really following an organized process of steps, never making irreversible steps, and yet ending up with the right answer," he says. "When thinking about computation at a molecular level, reliably making even those seemingly simple computations is very hard."

For R&D managers tracking unconventional computing architectures, the caveats are as concrete as the results. The measured performance covers 10 programs and roughly 700 computations; only simple ones finished within a minute, and the team describes the design as only partially optimized. Scaling to larger programs, or to computations beyond arithmetic, remains undemonstrated. The researchers themselves position the work as one of the most complex DNA computers built to date — an arguable claim, and one that competing groups in the field will test.

Still, the economics of the approach differ from fuel-driven molecular computing: a stable, reusable, water-tolerant system that reaches its answer by cooling lowers the experimental overhead per computation. That may matter for applications where wet-environment operation, rather than raw throughput, is the constraint.

"Although we optimized some aspects of the design, there remain many ways it could be improved and generalized," Woods says. "The work opens the door to new ways of thinking about computation in a wet environment, and about energy use in computation overall."