The prototype was built by NUS Medicine with Singapore-headquartered data center developer DayOne and Melbourne-based Cortical Labs, and sits at the university's Life Sciences Institute. It went live on July 16 and was demonstrated to more than 80 guests from industry and academia on Aug. 6 before NUS Medicine announced it publicly on Aug. 17.

The three partners describe the installation as the first independently operated biologically integrated server rack in the world, and as a step toward Singapore's first large-scale biological data center. Cortical Labs already runs about 120 of the same units in a small data center in Melbourne, the company told Euronews in April.

Each unit draws about 30 watts including its life support systems, less than a handheld calculator, and the neurons produce so little heat that the cooling equipment dominating conventional data centers becomes largely unnecessary, The Straits Times reported.

Nvidia rates its H100 SXM accelerator at up to 700 watts under load, and the company's DGX H100 server, which carries eight of the chips, has a maximum draw of roughly 10.2 kilowatts.

The International Energy Agency reported in April that data center electricity demand rose 17% in 2025 against 3% growth in global electricity demand overall, and projected that consumption will double by 2030. In Southeast Asia, data center power demand is expected to quadruple from 2.6 gigawatts in 2025 to 10.7 gigawatts by 2035, Computer Weekly reported.

Neither NUS nor Cortical Labs has published a benchmark comparing the units' useful computing output with GPUs on equivalent tasks, TechRepublic reported, and the installation remains a research deployment rather than an enterprise alternative to AI hardware.

Conventional data centers process cloud, web and database workloads on air-cooled silicon processors and general-purpose CPUs.

Biological computers replace those with neurons grown from stem cells, wired to hardware through microelectrode arrays that carry electrical signals in both directions between the cells and the digital system.

The CL1 is the world's first commercial, code-deployable biological computer, and each unit holds roughly 800,000 lab-grown human neurons reprogrammed from the skin or blood of adult donors, IEEE Spectrum reported.

The Straits Times put the count at a minimum of 200,000 per unit, describing cells derived from human blood, converted into stem cells and then cultured onto an electrode-fitted silicon chip.

The Biological Data Center prototype at the NUS Life Sciences Institute. Photo courtesy of the National University of Singapore

Technicians feed the cultures a mixture of sugar, micronutrients and pH buffers every three days while a gas mixer supplies carbon dioxide, oxygen and nitrogen, according to The Straits Times. Cortical Labs says the neurons stay viable for up to six months.

NUS Medicine will culture and maintain them under Rickie Patani, professor of neuroscience at the school and director of the neurobiology program at its Life Sciences Institute. NUS placed 10th worldwide in the QS World University Rankings released in June, ahead of every other university in Southeast Asia, and third in Asia in the Times Higher Education regional table published in April.

Hon Weng Chong, founder and CEO of Cortical Labs, said the prototype "shifts the conversation from research to commercial application." He named drug discovery, humanoid robotics, cybersecurity and fraud detection as the areas where sparse or shifting data favors biological systems over conventional AI.

He told The Straits Times the technology suits robots working in public spaces, homes and offices. Renting a unit's computing resources costs $2,200 a month, the newspaper reported.

The hardware itself sells for about $35,000, or $20,000 apiece in 30-unit racks, according to IEEE Spectrum. Cortical Labs' Melbourne site serves roughly 20 paying customers, mostly university groups and corporate research arms working on drug discovery and robotics, The Straits Times reported.

Chong has framed the technology as a supplement to conventional AI rather than a replacement, aimed at problems where data is scarce rather than at the fast, repeated arithmetic silicon handles.

If the Singapore validation phase succeeds, the work moves into a live DayOne commercial data center, with the partners exploring a phased expansion to as many as 1,000 units, Computer Weekly reported.