Definition
Biocomputing: Biocomputing is computing that uses biological materials, such as DNA molecules, engineered cells or living neurons, to store or process information. It ranges from DNA computing and synthetic gene circuits to neuron-based approaches such as organoid intelligence, where cultured human neural tissue on electrode arrays acts as the processing substrate.
Also known as: biological computing, biocomputation, biological computation
Biocomputing is the use of biological materials, such as DNA, engineered cells or living neurons, to store and process information. Instead of transistors switching on silicon, a biocomputer relies on chemical reactions, gene expression or electrical activity in living tissue to carry out a computation, usually with conventional electronics handling input and output.
Key takeaways
- Biocomputing is an umbrella term covering DNA computing, cellular computing and neuron-based computing.
- Adleman’s 1994 Science paper is widely cited as the first experimental demonstration of DNA computing.
- Neuron-based approaches, including organoid intelligence, use cultured neural tissue that can adapt and learn.
- Interest is driven partly by the energy cost of conventional AI and partly by what biology can teach about efficient computation.
- All forms of biocomputing remain at the research stage, with serious challenges in speed, reliability and scale.
What does biocomputing mean?
The word joins “bio” (life) and “computing”. In practice it describes any system where a biological component does meaningful information processing, not merely where biology is the subject of computation. That distinction separates biocomputing from bioinformatics, which uses ordinary computers to analyze biological data.
Three broad families are usually grouped under the term:
- Molecular computing, where DNA, RNA or proteins encode data and their reactions perform operations.
- Cellular computing, where engineered bacteria or other cells run synthetic gene circuits that behave like logic gates or counters.
- Neural computing, where living neurons, grown as 2D cultures or as 3D brain organoids, process electrical signals. This branch is often called wetware computing.
How does biocomputing work?
The details depend on the substrate, but every biocomputer needs a way to encode inputs, a biological process that transforms them, and a way to read out the result.
Molecular computing
In 1994 Leonard Adleman reported in Science that he had encoded a small instance of the directed Hamiltonian path problem in DNA strands. Mixing the strands let molecules combine in parallel, and laboratory steps filtered out the strands representing valid answers. The demonstration showed that chemistry could carry out a search, though reading out the answer required lengthy manual laboratory work.
Cellular computing
Synthetic biologists build circuits from genes and regulatory proteins so that a cell produces an output, such as a fluorescent protein, only when certain chemical inputs are present. These systems are useful as biosensors and as testbeds for programming living cells, but they are slow because gene expression takes minutes to hours.
Neural computing
Neurons communicate with fast electrical impulses called action potentials, and their connections strengthen or weaken with use through synaptic plasticity. In neural biocomputing, cultured neurons sit on a microelectrode array that both stimulates and records them. Software encodes a task into stimulation patterns, decodes the responses, and in many experiments closes the loop with feedback. Kagan et al. reported in Neuron (2022) that cultured neurons improved at a simplified Pong game under such a closed-loop setup. Cai et al. reported in Nature Electronics (2023) that a brain organoid used as a reservoir computing substrate could support a speech recognition task. The full picture is in what is organoid intelligence.
Types of biocomputing compared
| Type | Substrate | Typical operation | Speed | Main strength | Main limitation |
|---|---|---|---|---|---|
| DNA computing | DNA strands in solution | Hybridization and enzymatic reactions | Hours to days per run | Massive molecular parallelism, dense storage | Slow readout, error-prone chemistry |
| Cellular computing | Engineered bacteria, yeast or mammalian cells | Gene regulatory circuits | Minutes to hours | Runs inside living systems, useful for sensing | Limited circuit size, noise |
| Neural computing | Cultured neurons or brain organoids on electrodes | Electrical spiking and plasticity | Milliseconds | Adaptive, can learn from feedback | Lifespan, variability, interface limits |
The table explains why neural approaches attract attention for AI-like tasks. Neurons operate on millisecond timescales and already learn, while molecular and cellular systems are better suited to storage, sensing and diagnostics.
Why does biocomputing matter?
The energy question
Conventional computing is consuming more electricity as AI grows. The IEA reported in Energy and AI (2025) that data centers used around 415 TWh in 2024, roughly 1.5% of global electricity. Biological brains perform perception and learning on far less power than the hardware used to train large models. That contrast motivates research into whether living neurons could do some kinds of computation more efficiently. The argument, and its caveats, are covered in biocomputing vs silicon: energy efficiency.
Learning from biology
Biocomputing is also a scientific tool. Running computations on real neurons tests theories of how brains learn, such as Hebbian learning and the free energy principle, in a controlled setting.
New kinds of hardware
Silicon chips inspired by the brain, known as neuromorphic computing, borrow principles from neurons. Biocomputing goes further by using the biology itself, which may reveal computational properties that engineers have not yet thought to imitate.
Biocomputing vs neuromorphic computing
These terms are easy to confuse. Neuromorphic systems are built from conventional semiconductors arranged to emulate spiking neurons, so they are manufactured, reproducible and programmable. Biocomputing systems contain living or molecular material, so they grow, vary and age. Neuromorphic chips are much closer to deployment, while neural biocomputing offers access to genuine biological learning that silicon only approximates. Many researchers see them as complementary lines of work.
What are the limitations of biocomputing?
- Speed and throughput. Molecular and cellular systems are slow, and neural systems are limited by how many electrodes can reach the tissue.
- Reliability. Biological components are noisy, and no two cultures behave identically.
- Lifespan. Living tissue needs nutrients, temperature control and sterile handling, and it changes over time.
- Programming. There is no standard instruction set. Encoding a problem for a biological system is itself a research task.
- Ethics. Neural biocomputing uses human cells, raising questions about donor consent and, at least in principle, moral status. See the ethics of organoid intelligence.
Where is biocomputing heading?
Smirnova et al. (2023) laid out a roadmap for organoid-based biocomputing that calls for larger and longer-lived organoids, better electrode interfaces, data standards and embedded ethics. Progress in all branches of biocomputing is likely to come from hybrid systems, where biological components handle what they do well and silicon handles the rest. Read more in the future of biocomputing.
At Organoid Programming we are building software and cloud access for neuron-based biocomputing; you can follow the work through our early access program.
Frequently asked questions
What is a biocomputer?
A biocomputer is a system in which biological material performs part of the computation. Examples include test tubes of DNA strands that encode and solve a combinatorial problem, bacteria engineered with gene circuits that act like logic gates, and cultured neurons on electrode chips that respond to and learn from electrical inputs. Most biocomputers today are research prototypes.
Is biocomputing real or theoretical?
Biocomputing is real but early. Adleman demonstrated DNA computing experimentally in 1994, synthetic biology has produced working gene circuits, and labs have shown cultured neurons learning simple tasks in closed loop. None of these yet competes with conventional computers on general workloads, so biocomputing is best described as an active research field with early prototypes.
Why is biocomputing considered energy efficient?
The interest comes mostly from the brain, which performs complex learning and perception on very little power compared with data-center hardware running AI models. Molecular systems such as DNA can also act in massively parallel ways. Whether practical biocomputers will be energy efficient once life support, electronics and cooling are counted is still an open research question.
What is the difference between biocomputing and bioinformatics?
Bioinformatics uses ordinary computers to analyze biological data, such as genome sequences or protein structures. Biocomputing reverses the relationship: biological material itself does the computing. A bioinformatics pipeline runs on silicon, while a biocomputer might use DNA reactions or living neurons as its processing elements, often with silicon electronics around it for input and output.
How does organoid intelligence relate to biocomputing?
Organoid intelligence is one branch of biocomputing. It focuses on three-dimensional brain organoids grown from human stem cells as the computing substrate. Other branches use DNA, proteins or engineered microbes. OI stands out because neurons are naturally adapted to process signals and to learn, which makes them a candidate for adaptive, brain-like computation.
Sources
- Adleman (1994). Molecular computation of solutions to combinatorial problems. Science.
- Smirnova et al. (2023). Organoid intelligence (OI): the new frontier in biocomputing and intelligence-in-a-dish. Frontiers in Science.
- Kagan et al. (2022). In vitro neurons learn and exhibit sentience when embodied in a simulated game-world. Neuron.
- Cai et al. (2023). Brain organoid reservoir computing for artificial intelligence. Nature Electronics.
- IEA (2025). Energy and AI.
