Imagine a computer with no silicon chip. No transistors. No glowing circuit boards. Just a small, warm box, humming quietly, with living human brain cells inside—cells that are breathing, metabolizing, and, in a very real sense, thinking.
This is not science fiction. It is the reality of 2026.
In March 2025, an Australian company called Cortical Labs unveiled the CL1, the world’s first commercial biological computer. It is a shoebox-sized device that combines lab-grown human neurons with silicon hardware. Each unit contains approximately 800,000 living neurons, kept alive by an internal life-support system that feeds them, removes their waste, and maintains the precise conditions they need to survive for up to six months.
In August 2026, Singapore took the next step. A server rack at the National University of Singapore was fitted with 20 CL1 units—approximately 16 million living human neurons, working together as a biological data centre. It is the first independently operated, biologically integrated server rack in the world.
The implications are staggering. We are no longer merely simulating intelligence with silicon. We are cultivating it in living tissue.
The Dish That Learned to Play Pong
The story begins in 2022, when Brett Kagan and his team at Cortical Labs achieved something that seemed impossible. They grew human and mouse neurons on a high-density multielectrode array—a silicon chip covered in thousands of tiny electrodes that can both stimulate and record neural activity—and connected them to a simulated game of Pong.
The neurons were not programmed. They were not given explicit instructions. They were simply embedded in a closed-loop environment where their electrical activity controlled the paddle, and sensory feedback about the ball’s position was delivered back to them as electrical stimulation. Within five minutes, the neural cultures began to learn. They improved their rally lengths. They adapted. They exhibited what Kagan’s team termed “synthetic biological intelligence”.
A conventional AI took 90 minutes to learn the same task. The biological system, with its 800,000 neurons and its 20-watt power budget, learned in five.
The reason is simple, and profound. Silicon-based AI is a brute-force learner. It requires millions of examples, enormous datasets, and megawatts of electricity to find patterns. Biological neurons are pattern-prediction machines. They infer from sparse data. They generalize. They adapt. A human child does not need a million labeled images to recognize a cat. A neural culture on a chip does not need a million Pong rallies to learn how to hit a ball.
Building a Living Computer: From Blood to Thought
How do you build a computer out of human neurons? The process begins with a simple blood draw or skin biopsy. From these cells, scientists use a cocktail of transcription factors to reprogram them into induced pluripotent stem cells (iPSCs)—cells that have been returned to an embryonic-like state, capable of becoming almost any cell type in the body.
Under precise chemical guidance, these stem cells are differentiated into neurons. Over the course of weeks, they mature, form synaptic connections, and develop into functional neural networks. Approximately 800,000 of these neurons are then seeded onto a microelectrode array—a silicon chip studded with thousands of electrodes that serve as the interface between the biological and digital worlds.
The neurons grow across the chip, extending axons and dendrites, forming trillions of synaptic connections. The microelectrode array can stimulate specific regions of the network and record the resulting electrical activity. This is how information is sent in, and how the network’s “response” is read out.
But hardware alone is not enough. The system needs an operating system—a way to translate digital commands into biological signals, and biological activity into digital information. Cortical Labs developed the Biological Intelligence Operating System (biOS). It runs simulations, sends information to the neurons about their environment, and interprets their impulses to influence that simulated world. Developers can deploy code directly to the neural layer, and the feedback signals shape adaptive pathways just as synapses are shaped in a living brain.
The entire system is housed in what Cortical Labs calls a “body in a box”—a self-contained unit with a life-support system that circulates nutrients, removes waste, and maintains the temperature and pH necessary to keep the neurons alive. The CL1 is, in essence, a portable, self-contained biological brain.
The Energy Equation: 20 Watts vs. Megawatts
The most compelling argument for biological computing is not philosophical. It is thermodynamic.
Training a large AI model consumes megawatts of electricity and thousands of litres of water for cooling. Data centres already consume 1–2% of global electricity, and that fraction is rising rapidly. The human brain, by contrast, operates on approximately 20 watts—less than a dim lightbulb.
A biological computing platform can achieve roughly one exaFLOP of computational performance while consuming only 20 watts. A conventional exascale supercomputer performing the same workload would require 20–30 megawatts—a million-fold difference in energy consumption.
This is not a marginal improvement. It is a paradigm shift. As the world grapples with the escalating energy demands of artificial intelligence, biological computing offers a path to computation that is not only more powerful but also more sustainable. As Kagan himself noted, “It doesn’t all have to be systems that consume huge amounts of power”.
Singapore: The First Biological Data Centre
The launch of the biological data centre prototype in Singapore represents the transition of this technology from laboratory curiosity to commercial infrastructure. The system, housed at the NUS Life Sciences Institute, consists of 20 CL1 units in a single server rack. Each unit contains 800,000 neurons, giving the rack a total of approximately 16 million living neurons—roughly the processing capacity of a frog’s brain.
It is a modest beginning. A human brain contains 86 billion neurons. The Singapore rack, at 16 million, represents less than 0.02% of that. But the trajectory is clear. The system is not merely storing living cells next to a computer. The neurons form an active part of the processing loop, with biological activity translated into information that the surrounding digital system can use.
The partnership behind the project—NUS Medicine, DayOne (a data centre operator), and Cortical Labs—reflects the convergence of disciplines required to make biological computing a reality: neurobiology, infrastructure, and hardware design. NUS provides expertise in culturing and maintaining the living cells. DayOne brings its experience in data centre operations. Cortical Labs supplies the CL1 technology and the biOS operating system.
The system is being positioned not as a replacement for silicon, but as a complement—a new category of computing infrastructure where biological systems work alongside conventional electronics.
The Cloud of Living Neurons
Access to biological computing is not limited to those who can afford a $35,000 CL1 unit. Both Cortical Labs and FinalSpark, a Swiss competitor, offer cloud-based access to their neural networks.
FinalSpark’s Neuroplatform, launched in 2024, was the first cloud-accessible platform for biocomputing research. It allows researchers worldwide to perform real-time experiments on living neural networks via a Python API. Users can trigger neurotransmitters like dopamine, glutamate, and serotonin, and observe the resulting changes in neural activity. The platform is reported to be approximately a million times more energy-efficient than a digital processor.
Cortical Labs offers a similar service: wetware-as-a-service, with remote access to their in-house cell cultures for approximately $300 per week per unit. For those who prefer to own the hardware, the CL1 is available for $35,000, with discounts for bulk purchases.
This cloud model democratizes access to biological computing, enabling researchers, startups, and academic institutions around the world to experiment with living neurons without the need for specialized cell culture facilities.
The Ethical Question: What Happens When the Computer Feels?
The most profound—and unsettling—question raised by biological computing is not technical. It is ethical.
We are growing human neurons. We are organizing them into networks. We are training them to perform tasks. At what point, if ever, does a neural culture become capable of suffering? Of consciousness? Of a subjective experience of its own existence?
The scientific community has not reached a consensus. A 2026 commentary in Nature noted that researchers are building computers that run on brain organoids but have “neglected a major ethical issue”. Some researchers argue that current systems are too primitive to feel or understand. Others caution that as the technology advances, the ethical stakes will rise.
The question of consciousness in brain organoids has become a central focus of neuroethics. A 2026 study in Scientific Reports explored how people’s perceptions of consciousness influence their ethical judgments toward brain organoids used as biological computers. The findings suggest that the more people believe an organoid might be conscious, the more likely they are to oppose its use in computing.
There is also the question of funding and intent. Cortical Labs, which developed the CL1, has received investment from In-Q-Tel, the venture capital arm of the CIA. This has raised concerns about the potential military and intelligence applications of biological computing—applications that go far beyond drug discovery and disease modeling.
The Hybrid Future: Where Silicon Meets Synapse
The future of computing is unlikely to be purely biological or purely silicon. It will be hybrid. Silicon will continue to excel at rapid, precise calculations—the kind of deterministic processing that powers spreadsheets, databases, and traditional software. Biological neurons will be deployed where they have a unique advantage: learning from sparse data, adapting to changing conditions, and making complex decisions with minimal energy.
This hybrid model is already being explored. A 2025 paper on “Wetware Computing” proposed a framework for biological-digital hybrid architectures that achieve a 47% energy reduction compared to traditional silicon systems while maintaining competitive computational throughput. The system uses organic-silicon interfaces for seamless signal transduction, blending the best of both worlds.
The applications are vast. In drug discovery, biological computing platforms can serve as “physical reservoirs” for pattern recognition, outperforming traditional silicon-based AI in data efficiency and energy use. Brain organoids on MEAs can be used to test the effects of experimental medications on neural activity, providing a functional biomarker platform for central nervous system drug discovery.
In robotics, lab-grown brains have already been used to control robots for obstacle avoidance, maze navigation, and target tracking. Closed-loop systems enable adaptive behaviors, with neural activity guiding precise motor control.
In disease modeling, iPSC-derived neurons from patients with neurological disorders can be grown on MEAs to study disease mechanisms and test potential therapies. The neurons can be trained on tasks that reveal subtle functional deficits, providing a dynamic readout of disease progression and treatment response.
The Question That Lingers
The CL1 is a machine. It is made of neurons, but it is not a person. It does not have a childhood, a personality, or a sense of self. It learns to play Pong, but it does not know it is playing.
Yet there is something unsettling about a computer that requires feeding. A machine that can die. A processor that is, in a very literal sense, alive.
As we move toward a future where biological and digital systems are seamlessly integrated, we will need to develop new frameworks for understanding what we are creating. We will need to decide where the line lies between a tool and a being. We will need to ask ourselves whether it is ethical to grow a mind for the purpose of computation—even if that mind is small, and simple, and unaware.
The answer is not yet clear. But the question is no longer hypothetical. The living computer is here. It is learning. And it is waiting for us to decide what comes next.
It has no silicon chip. No transistors. Just living human neurons, breathing and thinking inside a box. This is the biological computer—and it is already learning. The future of computing is not silicon. It is alive.
#BiologicalComputing
#LivingComputer
#CL1
#HumanNeurons
#SyntheticBiologicalIntelligence #Scientific #cutting-edge
#WetwareComputing #Emerging
#NeuralNetworks
#FutureOfComputing
#DrSannjay
#BioDigitalConvergence
Original article
Comments