When Living Cells Become the Computer
Silicon has run this show for seventy years. Transistors kept shrinking, clock speeds kept climbing, and Moore's Law quietly became a law of physics in our heads. But we're bumping into real walls now heat, power draw, and the sheer physical limits of how small a transistor can get. So researchers have started looking somewhere unexpected for the next leap: biology itself.
That's the premise of biocomputing. It is an emerging field that uses biological components like DNA, proteins, or even living neurons to perform computation. It's easy to confuse with computational biology, but the two are almost mirror images of each other. Computational biology uses traditional computers to model and analyze biological data (think protein folding simulations or genome sequencing pipelines). Biocomputing flips that relationship, it uses biological material as the computer.
Biological systems are already exceptionally good at massively parallel, energy-efficient information processing. So people thought, "Why not put that machinery to work on our computational problems, instead of just simulating it in silicon?"
The Three Main components of Biocomputing
"Biocomputing" isn't one technology it's an umbrella over a few different approaches.
1. DNA Computing
This is where it all started. In 1994, computer scientist Leonard Adleman ran a landmark experiment where he used strands of DNA and a series of biochemical reactions to solve a small instance of the Hamiltonian path problem — a classic combinatorial puzzle. The idea was that trillions of DNA molecules could explore many possible solutions simultaneously, exploiting the parallelism baked into molecular biology.
DNA computing isn't just about solving logic puzzles, though. It's also emerging as a serious contender for data storage. DNA is astonishingly dense — a single gram of it can theoretically hold over 200 petabytes of data. For a world drowning in data and running out of places to archive it, that's a genuinely compelling pitch, even if practical read/write speeds are still far from where they'd need to be for everyday storage.
2. Molecular and Protein Computing
Here, researchers build logic gates out of biochemical reactions, enzymes, or synthetic biology circuits rather than DNA strands specifically. Think engineered bacteria or synthetic molecular circuits designed to respond to specific chemical inputs and produce predictable outputs — essentially wet-lab logic gates. It's a smaller, less headline-grabbing corner of the field, but it underpins a lot of the synthetic biology work happening in biosensors and diagnostics.
3. Organoid and Neuromorphic Computing
This is the part of biocomputing making the most noise right now, and for good reason.
Researchers have been growing brain organoids, these are small, lab-grown 3D clusters of human brain cells and connecting them to electrodes via microelectrode arrays (MEAs). The resulting field has picked up a name of its own: Organoid Intelligence (OI), formally proposed as a discipline in a 2023 Frontiers in Science paper by Thomas Hartung, Lena Smirnova, and collaborators.
The history here is wilder than you'd expect from an academic field:
- 2001 — Steve Potter's team at Georgia Tech used 2D cultured rat neurons to control a robot, letting it explore its environment.
- 2022 — Cortical Labs trained a dish of living neurons to play table tennis, in a project that went semi-viral as "DishBrain."
- 2023 — "Organoid Intelligence" is formally coined as a multidisciplinary field.
- 2025–2026 — The first commercial biocomputing platforms are starting to reach the market, marking a real inflection point from lab curiosity to product. Companies like Cortical Labs are building commercial neuron-based computing systems, while FinalSpark has claimed its biocomputing platform could be up to a billion times more energy-efficient than traditional silicon hardware — a figure notable enough that Forbes picked it up.
The Energy Argument
If there's one number that sells biocomputing better than any other, it's this: the human brain performs extraordinarily complex computation on roughly 20 watts of power, about what it takes to run a dim light bulb. Modern supercomputers doing comparable pattern-recognition tasks can burn through megawatts.
Biological systems also don't need the same kind of retraining cycles that silicon-based AI models do. Where a neural network typically needs to be retrained on fresh data, organoid-based systems can potentially learn continuously, adapting in something closer to real time — the way an actual brain does.
Combine that with massive inherent parallelism (millions of neurons or trillions of DNA strands all "computing" at once) and you get the core appeal of biocomputing: not necessarily raw speed, but staggering efficiency and adaptability.
What It's Actually Being Used For
Beyond the classic Boolean-logic demos, real applications are starting to take shape:
- Drug discovery and toxicity testing — organoids can be used as living testbeds for how tissue responds to new compounds, long before human trials.
- Disease modeling — particularly for neurodegenerative diseases like dementia, where "intelligence-in-a-dish" systems let researchers study learning and memory breakdown directly.
- Ultra-dense data archival — DNA storage as a long-term, high-density alternative to tape or cold storage.
- Environmental applications — early-stage research is exploring how bio-integrated computing systems could assist with carbon capture, bioremediation, and pollution monitoring. ## The Catch: Challenges and Open Questions
It wouldn't be a fair blog post without the reality check.
- Scalability and reproducibility — biological systems are messy. Getting consistent, standardized results across different organoid batches or DNA reactions is genuinely hard, and the field still lacks widely agreed-upon protocols.
- Keeping it alive — living components need nutrients, temperature control, and maintenance in ways silicon chips simply don't. A biocomputer is, in a very literal sense, alive, and that comes with upkeep.
- Speed — for most everyday computing tasks, biological systems are currently far slower than silicon. The advantage is efficiency and parallelism, not raw throughput.
- Ethics — this is the big one, especially for organoid intelligence. As brain organoids grow more complex and more capable of learning, questions about their moral status become unavoidable. Could a sufficiently advanced organoid have some rudimentary form of experience? Researchers are actively debating this in parallel with the technical work, and it's not a question anyone's rushing to answer casually. ## Biocomputing vs. Quantum Computing vs. Classical Silicon
It's easy to lump "alternative computing paradigms" together, but they're solving different problems in different ways:
| Classical Silicon | Quantum Computing | Biocomputing | |
|---|---|---|---|
| Core mechanism | Transistors, binary logic | Superposition & entanglement | Biochemical/biological processes |
| Key advantage | Speed, maturity, reliability | Exponential speedup on specific problem classes | Energy efficiency, massive parallelism, adaptability |
| Key limitation | Physical/thermal scaling limits | Extreme fragility, needs near-0K cooling | Slow, hard to standardize, ethical questions |
| Best suited for | General-purpose computing | Cryptography, optimization, simulation | Pattern learning, data storage, bio-integrated tasks |
They're not really competing for the same crown — they're complementary tools likely to coexist, each suited to different problem shapes.
Where This Is Headed
Biocomputing is still very much in its early, messy, exciting stage — closer to where quantum computing was a decade or two ago than to something you'll deploy in production tomorrow. But between DNA's absurd storage density, organoids learning to play table tennis, and efficiency claims measured in billions of times better than silicon, it's hard not to be curious about where this goes next.
We spent decades teaching silicon to mimic biology. Biocomputing flips the question: what if we just let biology compute for itself?
Sources referenced: Frontiers in Science (2023) on Organoid Intelligence, ScienceDirect review on OI and biocomputing advances, and recent industry coverage of Cortical Labs and FinalSpark (2025–2026).
Top comments (0)