Toward Living Computers: The Dawn of Brain-Cell-Based Biocomputing
- alhinocoo
- Oct 16
- 6 min read
In a development that sounds like science fiction turned fact, researchers have begun building computers that use living brain cells as computing elements. Unlike conventional silicon chips, these “wetware” or biocomputers use neuronal networks cultured from stem cells or brain organoids. The idea is to harness the brain’s natural capacity for plasticity, learning, and energy-efficient computation to outperform—or complement—traditional AI hardware.
This paradigm shift in computing has recently seen a major milestone: an Australian startup has commercialized a biological computer called CL1, which integrates human neurons onto a silicon chip. Parallel efforts in Switzerland are pushing forward a concept called “organoid intelligence,” using mini brain-organoids as computational units.
In this article, we will explore:
the science behind brain-cell computing
the key players and recent projects
technical and ethical challenges
potential applications
implications for the future
The Science of Brain-Cell Computing
What is “wetware” or organoid intelligence?
“Wetware” refers to computing systems built with organic, living tissue rather than synthetic circuits. A common approach is to culture neurons (or mixed neural-glial organoids) from stem cells, maintain them in nutrient media, and interface them with electrodes to send and receive electrical signals.
Organoid intelligence (OI) is an emerging field that seeks to make these biological tissues more computer-like—i.e. able to receive inputs, process computations, and produce outputs—while preserving the inherent advantages of neural networks: learning, adaptability, energy efficiency, and parallelism.
In contrast to AI, which tries to emulate the brain on silicon, OI explores embedding computing architectures into actual brain-like tissue.
Early demonstrations: DishBrain and neuron-in-a-dish
Some of the foundational demonstrations include the DishBrain system by Cortical Labs, in which neurons cultured on a microelectrode array learned to play the arcade game Pong by adjusting firing patterns in response to electrical stimulation and feedback.
Researchers stimulate the neurons in specific spatial-temporal patterns, monitor their outputs, and iteratively adjust feedback to guide learning. These systems exploit the brain cells’ natural tendency to seek patterns and form efficient synaptic connections.
These experiments were not just novelties—they demonstrated the possibility of integrating living neural networks in hybrid computing systems.
Recent milestone: CL1, the first commercial biocomputer
In March 2025, Australian startup Cortical Labs unveiled the CL1, described as the first “code deployable biological computer.”
Key features:
Neurons are grown across a silicon chip and maintained in a nutrient medium within a sealed, life-support enclosure.
The chip both stimulates the neurons and records their responses in real time.
It is intended for research use—disease modeling, drug discovery, neuroscience studies—and not general-purpose computing (at least initially).
The initial price is about US$35,000 per unit for single installations; when deployed in racks (30 units), discounts apply.
The company also offers Wetware-as-a-Service: remote access to neural cultures for users via cloud-like infrastructure.
The CL1 is being framed not as a consumer computer, but as a specialized research tool that bridges neuroscience and computing.
Other players: FinalSpark and multi-organoid arrays
In Switzerland, startup FinalSpark is building systems combining multiple brain organoids to perform computation.
They maintain several organoids (e.g. 4) wired to electrodes, and experiment with more biologically inspired feedback: for example, releasing dopamine to reinforce certain neural behaviors.
Some reports say they have combined 16 mini brains into a living computing processor.
These systems aim to scale beyond a single organoid, exploring how networks of biological units might collaborate or specialize in computing tasks.
Technical Challenges
While the promise is high, there are numerous hurdles to overcome before brain-cell-based computers become practical or mainstream.
Viability, scaling, and maintenance
Life support: Neurons require precise conditions—nutrient flows, oxygenation, temperature, biocompatible encapsulation—and must be protected from contamination.
Scalability: Current organoids contain tens of thousands to hundreds of thousands of cells; to match meaningful computing power, researchers estimate millions or tens of millions of neurons may be necessary.
Longevity: How long can neural cultures remain stable and viable under continuous operation?
Interfacing density: The number of electrodes and connection points needed to read and write sufficiently high-resolution signals is huge.
Noise, drift, and error: Biological systems are noisy and variable; connecting them to deterministic electronics introduces mismatch.
Computation models and programming
Encoding and decoding: How to translate electrical signals between digital logic and spiking neuronal patterns.
Training and feedback: Traditional backpropagation is not directly applicable; methods for reward-based or reinforcement learning in biological networks must be developed.
Reliability: Ensuring reproducible behaviors across multiple units or over time is challenging.
Ethical and philosophical concerns
Consciousness and sentience: As neuronal systems become more complex, at what point (if any) might they exhibit rudimentary awareness or capacity for suffering?
Tissue sourcing: Use of human-derived stem cells raises consent, privacy, and regulatory issues.
Dual use and security: Could such systems be misused (e.g., bio-hacking)?
Regulation and oversight: Clear frameworks will be needed to govern experimentation and deployment.
Integration with existing systems
Bridging between conventional digital systems and bioware in real time is nontrivial.
Latency, bandwidth, and compatibility: The hybrid system must not be bottlenecked by mismatches between biological and silicon domains.
These challenges mean that early systems will likely remain niche, used mostly in laboratories for research rather than consumer or industrial applications.
Potential Applications
Given the early stage, the most likely and impactful uses are in domains that benefit from the unique strengths of biological computation.
Neuroscience, disease modeling, and drug discovery
Because the computing elements are neurons, it becomes possible to model neurological diseases, screen drugs, and study dynamics in a more biologically faithful environment than pure simulation. The system could reflect real electrophysiological responses to stimuli or perturbations.
For example, one could introduce disease-associated mutations in the neurons and observe emergent network behavior or test candidate therapies in situ.
Low-power, adaptive computing
Biological neurons operate at remarkably low energy (in the order of milliwatts) compared to high-power AI training. This opens the possibility of computing systems that adapt, self-repair, and learn while consuming far less energy than comparable silicon-based AI.
Novel AI architectures
Rather than simulating neural networks on digital hardware, one could offload certain pattern recognition, reinforcement learning, or adaptive control tasks to living neuronal subsystems. These might operate in parallel to traditional AI, complementing or accelerating them in hybrid architectures.
Brain–machine interfaces and prosthetics
In the longer term, neurons as computing elements could provide more seamless integration between biological and electronic systems—e.g., in prosthetic limbs, sensory augmentation, or closed-loop interfaces.
Biocomputing clouds and services
Just as we use GPU clouds today, future “wetware-as-a-service” (WaaS) could let users access neuron-based computational resources remotely. Cortical Labs already offers this with CL1 units.
Implications and Outlook
If successful, living computers could overturn assumptions about the future of computing. Here are some key takeaways and predictions:
Energy efficiency and scaling
Biological systems may one day deliver orders-of-magnitude improvements in energy-per-operation over silicon AI. This could be critical in the era of large-scale neural models and pervasive AI use.
Hybrid architectures
Rather than replacing silicon, biocomputing will likely augment it—certain tasks or submodules may be delegated to living systems, creating hybrid AI architectures with dynamic adaptability.
New software paradigms
Soft computing, reinforcement, unsupervised learning, and neuromorphic approaches may become more central, bypassing rigid Boolean logic in favor of emergent, plastic computation.
Ethical, societal, and regulatory complexity
As neuron-based systems grow more complex, we must grapple with whether they merit moral status, regulation, or oversight. Policies will lag behind technology if not anticipated.
Long timeline and incremental progress
This is not a leap all at once—progress will come gradually: higher-density electrodes, more stable organoids, improved interfacing, better training protocols, and more robust life systems.
Open research and collaboration
Advances will depend heavily on multidisciplinary work across neuroscience, bioengineering, computing, ethics, and materials science. Open datasets, platforms, and standards will be pivotal.
While today’s systems (e.g. CL1 or small organoid arrays) are primitive by conventional standards, they already mark the beginning of a new frontier. They prompt us to ask: What even is a computer when neurons themselves compute?
The emergence of computers built on live brain cells is no longer speculative—it is becoming tangible. With innovations like Cortical Labs’ CL1 and the organoid intelligence paradigm, humanity is venturing into a hybrid realm between biology and technology.
Though many challenges remain—scaling, interfacing, reliability, ethics—the potential rewards are profound: ultra–energy-efficient AI, new paths to simulate disease, adaptive hybrid systems, and novel computing models grounded in life itself.
We are witnessing the early genesis of biocomputing, where circuits pulse with living cells, learning and adapting like brains. This is not simply a next step in hardware—it may be a new computing paradigm. The question now is: how fast can we grow it, and how responsibly?
