# Biological computing: Singapore tests 20 computers with human neurons

> Biological computing: Singapore tests 20 computers with millions of living human neurons to process information
> By Observatorio Blockchain · Technology · 2026-09-16

The National University of Singapore, DayOne and Cortical Labs have put into operation a prototype made up of 20 CL1 biological computers. Each unit contains at least 200,000 human neurons grown in the laboratory and connected to silicon chips via electrodes.

By using living matter to process information, the project opens a new stage in computing. Instead of relying exclusively on processors manufactured from silicon, it incorporates living human neurons grown in the laboratory, connected via electrodes to electronic systems and used to process information.

The project is installed at the [Life Sciences Institute of the National University of Singapore](https://medicine.nus.edu.sg/news/nus-medicine-dayone-and-cortical-labs-unveil-biological-data-center-prototype-in-singapore/) (NUS) and has been developed by the university's Yong Loo Lin School of Medicine, the data center company DayOne and the Australian biological computing firm Cortical Labs.

The facility brings together 20 CL1 biological computers developed by Cortical Labs in a single rack. NUS Medicine describes the set-up as the first biologically integrated server rack capable of operating independently and considers it the prototype of a future Biological Data Centre.

**[How to replace ChatGPT on Linux with local AI to protect your prompts and documents](https://observatorioblockchain.com/en/ai-agents/chatgpt-frente-ia-local-linux-datos/)**

The facility was shown publicly on August 6, 2026, before more than 80 representatives from academia, technology, business and digital infrastructure. During the presentation, a live demonstration was given of the CL1 and Cortical Cloud systems, their integration via microelectrode arrays and the activity of the neural networks in real time.

The story is particularly striking because the neurons are not a computer simulation of the brain. They are living human cells.

## Four million neurons

Each of the 20 CL1 computers contains at least 200,000 neurons grown in the laboratory, according to information gathered by the Neuroscience programme at NUS Medicine. That allows a minimum figure of approximately four million neurons across the entire facility to be established.

The cells are obtained from human cells reprogrammed as stem cells and subsequently differentiated to produce neurons. The cultivation and growth of the project's cells is carried out under the supervision of Rickie Patani, Professor of Neuroscience at NUS Medicine, director of the Neurobiology Programme at the NUS Life Sciences Institute and head of the university's Neuroscience Translational Research Programme.

The CL1's operation combines biology and electronics. The neurons grow on specially designed chips and communicate with the computer system via microelectrode arrays. These electrodes make it possible to send electrical signals to the cells and subsequently record their activity.

Communication therefore works in both directions. The digital system provides electrical stimuli to the neurons and the neural network responds by generating its own electrical activity, which can be captured and interpreted by the computer.

## Biological computing

Cortical Labs calls this combination of living cells and computer systems biological computing and presents the CL1 as the first commercial biological computer on which developers can deploy code directly onto networks of cultured neurons.

The company also uses a system called Biological Intelligence Operating System, biOS, which enables interaction between the digital environment and the neural network. Instead of mathematically simulating the operation of a neuron, as happens in the neural networks used by artificial intelligence, the system works directly with biological cells.

It is a fundamental difference. An artificial neural network is made up of mathematical structures executed by processors. In the CL1, part of the processing is carried out by real neurons connected to the electronics.

Cortical Labs sums up its technology as the union of laboratory-grown neurons with silicon chips. The cells form a network on the chip, receive information via electrical signals and generate responses that can be used within the computing system.

## From hardware and software to wetware

The emergence of these systems introduces a third category into computing vocabulary. Alongside hardware and software appears the so-called [wetware](https://pmc.ncbi.nlm.nih.gov/articles/PMC12412539/), a term used to describe living biological components integrated into computing systems.

NUS Medicine expressly uses this concept to explain its Biological Data Centre. The university notes that the infrastructure uses living biological neurons derived from stem cells to process information and support artificial intelligence systems.

Although the biological computer still needs chips, electronics and software, the difference is that it incorporates an additional layer made up of living cells. The resulting architecture is hybrid, since biology and silicon work together within the same system.

Traditional computing has for decades used inert matter and electricity to perform operations. Biological computing introduces living matter into that equation.

## Learning to use neurons that learn

The origin of Cortical Labs is precisely related to neuronal learning. The company has spent years researching how networks of cultured neurons can interact with digital environments and modify their behavior in response to the stimuli they receive.

One of Cortical Labs' best-known experiments is DishBrain. The experiment was initially carried out in 2021 and its results were published [in 2022](https://pubmed.ncbi.nlm.nih.gov/36228614/), in the scientific journal Neuron. The researchers cultured networks of human and rodent neurons on high-density microelectrode arrays and integrated them into a digital environment that simulated the video game Pong. Through electrical stimulation and recording of neuronal activity, they established a closed loop in which the networks received information about the environment and modified their activity during the task.

The ball existed only within a digital environment. The object's position was communicated to the cells via electrical stimuli and neuronal activity was used to control the virtual paddle. With the feedback provided by the system, the activity of the neural networks changed progressively during the task.

Those experiments showed the possibility of creating a closed loop between living cells and a digital world. The CL1 seeks to turn that experimental principle into a computing platform accessible via software. The company currently allows these networks to be used locally via the CL1 or accessed remotely via Cortical Cloud, a platform that offers researchers and developers the possibility of running experiments on real neurons without having their own laboratory.

## What is a biological computer for?

The big question is what tasks a biological neural network can perform better than a conventional processor or an artificial intelligence system. Cortical Labs argues that the main potential advantage lies in problems where there is little data and it is necessary to learn and adapt quickly to changing conditions. Among the areas the company intends to research are drug discovery, humanoid robotics, cybersecurity and fraud detection.

NUS Medicine also envisages applications related to neurological research. Working with networks of human neurons in controlled environments could make it possible to directly study processes related to learning, adaptation, neurological diseases and response to drugs.

Rickie Patani believes this combination makes it possible to investigate two questions simultaneously: new forms of computing and the biological mechanisms that enable learning and adaptation.

This dual condition makes biological computing a field located at the intersection of computer science, neuroscience and biotechnology. The same system that can be researched as a computational architecture can also be used as an experimental platform to study the behavior of human neural networks.

## Energy efficiency

One of the main arguments used by NUS Medicine, DayOne and Cortical Labs to promote this technology is its potential energy efficiency. Biological systems are capable of carrying out complex processes using extraordinarily small amounts of energy, something especially attractive at a time when the growth of artificial intelligence is increasing the electricity needs of data centers.

NUS Medicine argues that biological computing could operate using a fraction of the energy required by digital computers in certain tasks. DayOne also presents the project as a way to explore new forms of increasing computing capacity while reducing energy intensity.

However, that advantage must for now be considered a technological hypothesis that needs to be measured through equivalent comparisons. Neurons may display extraordinary biological efficiency, but a computer like the CL1 also needs electronics and systems capable of keeping the cells in suitable conditions. To determine real efficiency it will be necessary to compare consumption, performance and type of computational workload using homogeneous criteria.

Energy efficiency is therefore one of the project's great promises, but there is still not enough independent evidence to turn it into a demonstrated general superiority over conventional computing.

## A data center that needs to feed its cells

The presence of living tissue also introduces needs that do not exist in a conventional server. Neurons need nutrients and certain environmental conditions to survive. According to the published description of the Singapore facility, technicians periodically provide the cells with a mixture of nutrients and use support systems to maintain the conditions necessary for their survival. Computing thus literally acquires a biological component.

The rack installed at NUS is part of a broader strategy by DayOne and Cortical Labs. Both companies announced in March 2026 their intention to use this first phase to validate the technology and subsequently study its deployment within a DayOne commercial facility in Singapore.

DayOne considers Singapore to offer a particularly suitable environment for these tests due to the combination of scientific capacity, technology industry and strong restrictions on the energy consumption associated with data center expansion.

Cortical Labs has also made clear its intention to identify commercial applications. Its founder and CEO, Hon Weng Chong, believes biological computing could complement artificial intelligence in areas where there is little data and it is important to learn and adapt to changing conditions.

The company intends to use Singapore as a testing ground for a model that, if it works, could subsequently be reproduced and scaled elsewhere.

## Part of the computer is alive

The Singapore project represents a profound conceptual shift in the history of computing. For decades, biology has served as inspiration for building machines. The artificial neural networks that underpin much of contemporary artificial intelligence were born precisely from the intention to mathematically reproduce some extremely simplified principles of neuronal operation.

Biological computing now travels the opposite path and, instead of merely imitating certain properties of neurons through software, incorporates real neurons into the computational architecture.

It does not mean that these systems possess a brain or that the networks currently used are conscious. The cells used in the CL1 form cultured neural networks connected to electronic systems to study learning and information processing. But the matter participating in the computational process is alive.

At the Life Sciences Institute of the National University of Singapore there is already a rack with 20 CL1 computers in which millions of cultured human neurons exchange electrical signals with silicon chips.

Computing was born using machines to process information. It later managed to build machines capable of learning through artificial intelligence. Biological computing opens a third possibility, which is to use living systems directly as part of the computing infrastructure.
