Singapore’s next experiment in AI infrastructure may not look like a conventional server farm. DayOne Data Centers, Cortical Labs and the Yong Loo Lin School of Medicine at the National University of Singapore have launched a biological data-center prototype built around living human neurons. The project puts “wetware” computing into a real research environment, offering an early test of whether biological neural networks can complement silicon systems where energy efficiency, adaptive learning and scientific modelling matter.

DayOne Tests Whether Living Neurons Belong in the Data Center

The data-center industry has spent years squeezing more computing performance from silicon. The AI boom is now making that strategy increasingly expensive in power and cooling.

The International Energy Agency estimates global data-center electricity consumption will more than double to about 945 terawatt-hours by 2030, with accelerated servers used heavily for AI accounting for almost half of the projected net increase in data-center electricity consumption.

Against that backdrop, DayOne is backing a radically different proposition: some computing workloads may eventually be better served by living neurons than by conventional processors.

The Singapore-headquartered digital infrastructure company has launched what it describes as Singapore's first Biological Data Center Prototype in partnership with Melbourne-based Cortical Labs and the Yong Loo Lin School of Medicine at National University of Singapore.

The prototype incorporates a 20-unit deployment of Cortical Labs' CL1 biological computing system in a live research environment at NUS Medicine. Cortical Labs describes the CL1 as a programmable biological computer in which living neurons are cultivated on a silicon interface and communicate bidirectionally with software.

That makes the project fundamentally different from a conventional AI server rack — and also much more experimental.

What is biological computing?

Biological computing uses living neural cells as part of the computational system rather than merely trying to imitate neurons with mathematical models.

In the CL1, neurons are cultivated in a nutrient-rich environment on a silicon chip. Electrical signals are delivered to the cells, while their activity is measured and fed back into a software-controlled environment. Cortical Labs calls the resulting architecture a biological intelligence operating system, or biOS.

The underlying scientific idea is not entirely new. Cortical Labs attracted significant attention after researchers demonstrated that cultured neurons could learn to play a simplified version of Pong. Nature reported in 2022 that hundreds of thousands of human neurons grown on electrodes had been trained to interact with the game.

What is new in Singapore is the infrastructure proposition.

Rather than treating biological computing purely as a laboratory experiment, DayOne is putting the technology into a data-center context and testing how such systems could operate alongside conventional digital infrastructure.

That distinction matters because biological computing is unlikely to replace GPUs or CPUs across general-purpose enterprise workloads anytime soon.

A complement to GPUs, not a replacement

The more credible near-term opportunity is hybrid computing.

Modern AI systems depend heavily on silicon accelerators from companies such as NVIDIA, while cloud platforms operated by Microsoft, Google and Amazon continue to scale conventional AI infrastructure.

Biological computing approaches the problem from another direction. Living neural networks naturally exhibit adaptation, parallel activity and learning dynamics. Recent research into in-vitro neuronal networks has found evidence that biological networks can perform computation with low metabolic energy and that stimulation patterns can influence both energy consumption and information throughput.

A 2026 review in Current Biology likewise argues that biological principles could inform more energy-efficient AI architectures, particularly in learning, adaptation and signalling.

That does not establish that a biological computer is more energy efficient than a production GPU for every AI workload. The full energy cost of maintaining living cells, laboratory infrastructure, interfaces and supporting systems has to be included in any serious comparison.

The relevant question is narrower: are there workloads where biological systems deliver useful intelligence with substantially less data or energy?

Cortical Labs believes the answer could emerge in areas including drug discovery, robotics, cybersecurity and fraud detection.

Singapore is a strategic test bed

The choice of Singapore is also significant.

The city-state is already a major regional data-center hub, with more than 1.4 GW of capacity and more than 70 cloud, enterprise and colocation facilities, according to Singapore's Infocomm Media Development Authority.

But Singapore's physical constraints make additional compute capacity unusually sensitive to energy efficiency.

IMDA says data centers are power- and resource-intensive and has established a Green Data Centre Roadmap aimed at adding at least 300 MW of capacity through efficiency improvements and green-energy deployment.

The government has also introduced SS 715:2025, a data-center IT energy-efficiency standard intended to support at least 30% energy savings in IT equipment through better hardware selection and optimization.

Biological computing therefore arrives in a market already treating energy efficiency as an infrastructure constraint rather than a corporate sustainability add-on.

DayOne's involvement is important for another reason. The company operates digital infrastructure across Asia Pacific and Europe and says it has secured approximately 2.1 GW of bookings since inception. The prototype gives it an opportunity to investigate whether emerging computing architectures eventually require different infrastructure models from traditional hyperscale facilities.

From neuroscience to enterprise computing

The NUS partnership also gives the project a research dimension that conventional data-center operators generally do not possess.

Researchers led by Professor Rickie Patani are using the biological platform to investigate neurobiology, learning and adaptation while exploring applications such as biomedical modelling, drug discovery and neurological disease research.

That dual-use model could prove more important than raw compute benchmarks.

A biological computer capable of interacting with living neurons provides researchers with an experimental system for studying how biological intelligence learns and responds. At the same time, the same platform could eventually become a specialized compute substrate for tasks where adaptability and sample efficiency matter.

The commercial opportunity remains unproven. Biological systems introduce challenges that silicon infrastructure largely avoids, including maintaining viable cells, managing biological variability, establishing reproducible benchmarks and determining how these systems scale.

There are also governance and ethical questions as biological computing becomes more sophisticated.

For enterprise technology leaders, the immediate lesson is therefore not to replace GPU clusters with wetware. It is to watch the emergence of heterogeneous AI infrastructure in which GPUs, CPUs, neuromorphic processors and potentially biological processors perform different classes of work.

DayOne's Singapore prototype is an early attempt to move that idea from the laboratory toward infrastructure.

If the experiments produce commercially useful results, the data center of the future may not be defined by one processor architecture. It may be a collection of radically different computing substrates, with each selected according to the energy, latency, learning and adaptability requirements of the workload.

Market Landscape

The biological-computing experiment comes as data-center operators face a convergence of AI growth, electricity constraints and sustainability requirements.

Singapore is particularly relevant because its government is simultaneously encouraging additional data-center capacity and imposing stronger efficiency requirements. IMDA's second Data Centre Call for Application made at least 200 MW of additional capacity available while requiring applicants to meet stringent sustainability criteria, including a target PUE of 1.25 or better and at least 50% green-energy pathways for proposed capacity.

DayOne's approach differs from mainstream infrastructure strategies pursued by hyperscalers.

NVIDIA continues to improve GPU performance and energy efficiency at the accelerator level. Google, Microsoft and Amazon are investing in increasingly sophisticated data-center architectures, renewable energy and cooling technologies. Neuromorphic computing companies are meanwhile attempting to reproduce brain-inspired efficiency using electronic hardware rather than living cells.

Cortical Labs sits at a more experimental end of this spectrum. Its CL1 combines biological neurons with silicon interfaces and software, creating a hybrid system rather than an alternative to electronics altogether.

The commercial test will be whether this architecture can deliver measurable advantages on useful workloads after accounting for the complete operational requirements of biological infrastructure.

Top Insights

• DayOne, Cortical Labs and NUS Medicine are testing biological computing as a specialized AI infrastructure layer for adaptive workloads and scientific research.
• The 20-unit CL1 deployment brings living-neuron computing into a data-center environment, potentially creating a new category between laboratory equipment and enterprise compute.
• Singapore’s energy constraints and Green Data Centre Roadmap make the city-state a natural test market for lower-power alternatives to conventional AI infrastructure.
• Biological computing could complement NVIDIA GPUs rather than replace them, targeting workloads involving adaptive learning, sparse data, biomedical modelling and complex real-world signals.
• Enterprise adoption remains experimental because biological systems introduce maintenance, reproducibility, scalability and governance challenges absent from conventional silicon infrastructure.

Get in touch with our MarTech Experts