In a lab rack that looks more like a high-end audio system than a server, clusters of human brain cells are quietly learning to process information. Electrodes feed them signals, nutrients keep them alive, and software turns their flickers of activity into computation. The result is a “living” computer that behaves less like a rigid chip and more like a tiny, adaptable nervous system.
By wiring these mini-brains into digital hardware, researchers are testing whether biology can outperform silicon on some of the hardest problems in artificial intelligence and energy efficiency. The work is still experimental, but it already forces a basic question: when a computer is made of living neurons, where does the machine end and the brain begin?
From sci‑fi idea to rentable “living” hardware
The notion of a computer built from brain cells has long sounded like science fiction, yet it is now a commercial product that researchers can rent over the internet. Swiss startup FinalSpark describes its system as a new generation of bioprocessors, with brain organoids grown in the lab and wired into electronics so they can be trained to respond to inputs and generate outputs like a very unconventional CPU. On the company’s main site, the project is framed as a radical shift in how computation is done, with the FinalSpark team arguing that organic tissue can handle complex patterns with far less energy than conventional chips.
Another company, Cortical Labs, has taken a similar leap by embedding living neurons into a device it calls CL1, which it presents as a computer powered by human brain cells. In a short explainer, the CL1 is introduced as a machine that you can now buy, built by the Australian Cortical Labs team to showcase how biological networks can perform simple tasks when given the right feedback. Together, these efforts mark a turning point: biocomputers are no longer just lab curiosities, they are becoming platforms that outside scientists can access and experiment with.
Inside the world’s first “living computer”
The most visible example of this shift is a system often described as the world’s first “living computer,” built from tiny clumps of human brain cells grown in dishes. These organoids are arranged in clusters and connected to electronics so that they can receive electrical stimulation and send back measurable responses, effectively turning them into a biological processing unit. Reporting on the project notes that the company behind it is working with 10 universities worldwide, using these clusters to perform simple computational tasks and to study how living networks learn when they are treated like hardware, a setup detailed in coverage of the Oct World initiative.
What makes this machine so unusual is not just its living substrate but its global accessibility. Instead of shipping delicate organoids to every lab, the company keeps the mini-brains in a controlled facility and lets partners log in remotely, sending stimuli and recording neural activity over the cloud. That model mirrors the way cloud computing turned racks of servers into a shared resource, but here the “servers” are biological, and their behavior is shaped by growth, plasticity, and even reward chemicals like dopamine, as described in reports on brain-cell systems that are now training on dopamine to reinforce desired responses.
How a bioprocessor actually works
At the heart of these platforms is a deceptively simple architecture: brain organoids sit on top of microelectrode arrays, which act as both input and output channels. Each organoid contains thousands of neurons that spontaneously generate electrical activity, form synaptic connections, and reorganize themselves in response to stimulation. When researchers send patterns of pulses through the electrodes, the cells adapt, and over time the network can be coaxed into mapping specific inputs to specific outputs, a process that turns raw tissue into a programmable, if unpredictable, processor, as described in technical overviews of Dec Working on organoid computing.
Keeping these mini-brains alive and stable is as important as the electronics. The organoids must be constantly bathed in a nutrient-rich solution known as Neuronal Medium, which is circulated through a microfluidic system that delivers fresh supplies and removes waste. In one detailed description, the section labeled 3.4 explains how Micro-fluidics is used to sustain organoids on the MEA by continuously supplying Neuronal Medium through a closed-loop circuit. That infrastructure turns what would otherwise be a fragile biological sample into a semi-stable computing element that can run for extended experiments, bridging wet biology and dry code.
FinalSpark’s Neuroplatform and the rise of organic cloud computing
FinalSpark has wrapped these components into what it calls the Neuroplatform, a remotely accessible environment where researchers can run experiments on living processors without ever touching a pipette. The company describes this as the world’s first commercial bioprocessor, positioning it as a radical approach that could inaugurate a new generation of organic computing hardware. In one profile, the Neuroplatform is presented as a system built by a Swiss startup that integrates organoids, microfluidics, and software into a coherent stack, with Oct Swiss Neuroplatform described as the start of an “era of organic computing.”
From the user’s perspective, the Neuroplatform behaves like a specialized cloud service. Researchers can log in, select specific organoid cultures, and design stimulation protocols, while the system handles the messy details of keeping the cells alive and recording their activity. A separate deep dive into the project notes that the company has built a dedicated FinalSpark Lab to maintain the organoids, manage the microfluidic systems, and provide a stable interface for clients, with one analysis explaining what FinalSpark is actually doing and why it matters by walking through how the Aug You Here Lab keeps the cultures active and usable as computing resources.
Cortical Labs and the DishBrain lineage
While FinalSpark has focused on organoid clusters, Cortical Labs has built its reputation on flat cultures of neurons grown directly on electrode arrays, a configuration it calls DishBrain. These networks are trained using feedback loops that reward desired behavior, such as steering a virtual paddle in a game, and punish errors, gradually shaping the cells’ activity into something that looks like learning. The company’s own materials describe how its CL1 device packages this approach into a product, with Cortical Labs presenting DishBrain-style systems as a bridge between neuroscience and artificial intelligence.
Independent reporting has highlighted how these cultures can be pushed to perform surprisingly sophisticated tasks when given the right training signals. One account describes how neurons in a dish were taught to play a simplified video game using a feedback system that rewarded correct moves and penalized mistakes, a process summarized under the phrase Teaching Mini Brains with Feedback in coverage of the world’s strangest computer that is alive and blurs the line between hardware and biology. That same report notes that these networks can generate electrical activity, form connections, and import information from their environment, illustrating how Dec Teaching Mini Brains Feedback can turn a sheet of neurons into a rudimentary information processor.
Why brain-cell computers matter for energy and AI
The appeal of these living processors is not just novelty, it is efficiency. Neurons are extraordinarily frugal compared with transistors, and a small cluster of brain cells can, in principle, perform complex pattern recognition while consuming a fraction of the energy used by current systems. One analysis invites readers to imagine a world where computers are made from human brain cells, consuming a fraction of the energy used by today’s hardware, and frames this as the dawn of a new era in computing in which organoid-based platforms become the first rentable systems of their kind, a vision laid out in a piece titled Aug Imagine.
There is also a sustainability argument. Traditional data centers draw enormous amounts of power and generate significant heat, while a biological processor operates at body-like temperatures and leverages the inherent parallelism of neural tissue. A separate commentary on The Living Computer frames this as a way of merging biology with technology for a sustainable future, suggesting that such systems could run advanced artificial intelligence with great efficiency if they can be scaled and controlled. In that account, the project is explicitly described as The Living Computer, Merging Biology and Technology for a Sustainable Future, with advocates arguing that Aug The Living Computer Merging Biology Technology for Sustainable Future could help address the energy demands of AI if the technology matures.
Training mini-brains with dopamine and feedback
Teaching a living network to compute is very different from programming a chip, and researchers are borrowing heavily from neuroscience to make it work. Instead of writing code, they send patterns of electrical pulses and adjust them based on how the neurons respond, using reward signals to reinforce useful activity. In some experiments, this reward takes the form of dopamine, the same neurotransmitter that shapes learning in the human brain, which is delivered to the organoids when they produce desired outputs, a strategy described in reports on living brain-cell biocomputers that are now training on dopamine to guide their development.
Other teams rely on purely electrical feedback, turning correct behavior into pleasant stimulation and errors into disruptive noise. This is the logic behind the Teaching Mini Brains with Feedback approach, where neurons learn to control a simple system, such as a virtual paddle, by associating certain firing patterns with better outcomes. Over time, the network reorganizes itself to favor those patterns, effectively encoding a policy in its synaptic weights. That process is central to the experiments described in coverage of the world’s strangest computer that is alive, where Dec Teaching Mini is used as shorthand for this feedback-driven training.
Ethical questions and the “mini-brain” line
As these systems become more capable, they raise uncomfortable questions about what, exactly, is being built. The organoids used in these computers are often described as “mini-brains,” but they lack the structure and sensory input of a full nervous system, and researchers stress that they are far from anything like a conscious mind. Even so, the language of living computers and brain-cell CPUs forces a reckoning with how society defines moral status in the lab, a tension that surfaces in discussions of The Living Computer and its promise to merge biology and technology for a sustainable future, where advocates of The Living Computer also acknowledge the need for ethical guardrails.
There are also concerns about how such technology might intersect with broader debates over artificial intelligence and energy use. One detailed report on brain-cell computers asks bluntly, Is AI making it worse, and notes that Cortical Labs is renting out its biological computers over the cloud while also training them to recognize characters at 83 percent accuracy. That same account highlights how Jul Is AI Cortical Labs is positioning its technology as a more sustainable alternative to power-hungry AI models, even as it prompts new questions about the treatment of living tissue in commercial systems.
Where this “living” hardware could go next
For now, these biocomputers are fragile, small scale, and limited to relatively simple tasks, but the trajectory is clear. Companies are already talking about scaling up the number of organoids, improving electrode density, and refining microfluidic support so that larger, more stable networks can be trained for longer periods. One Instagram reel, for example, highlights how 16 lab grown human brains make up what is described as the world’s first living computer, built by the startup Final Sp so that researchers can conduct experiments with the computer remotely, a snapshot of how Jun What the Final Sp is already pushing the scale of organoid clusters.
Looking ahead, proponents imagine hybrid systems where silicon handles precise arithmetic and control, while living tissue tackles pattern recognition, adaptation, and other tasks that brains excel at. Commentators on biocomputing breakthroughs argue that this could mark the dawn of a new era in computing, with human brain cells acting as the new CPUs in specialized contexts, and with rentable platforms like FinalSpark’s Neuroplatform and Cortical Labs’ CL1 serving as early testbeds. In that vision, the line between brains and machines does not disappear, but it becomes a design choice, with engineers deciding how much of their system should be alive, a prospect that is already being explored in detail on sites that invite readers to The Dawn of a New Era in Computing.
More from MorningOverview