In Singapore, a collaborative team from NUS Medicine, DayOne, and Melbourne-based Cortical Labs has switched on a prototype data centre rack that operates with living neurons instead of traditional silicon chips. This breakthrough aims to transform computing efficiency by integrating biological systems into digital infrastructure.
- Neuron-powered rack integrates 20 CL1 biological computers
- Prototype targets AI scalability with potentially lower neuron energy use
- Life support limits neuron viability, posing operational challenges
What happened
NUS Medicine, DayOne, and Cortical Labs jointly introduced a 20-unit biological data centre rack in Singapore powered by living human neurons. This demonstration took place on August 6 and showcased the first independently operated server rack combining biological neurons with computational infrastructure.
The neurons used in the CL1 computers are cultured from stem cells and maintained to keep them alive during operation. The rack consumes up to 20 kilowatts, similar to traditional data centre racks, because much of the power sustains cellular life rather than computational processes alone.
Why it matters
This development represents a pioneering step in transitioning from silicon-based chips to neuron-based biological computing, which proponents believe could drastically reduce the energy cost of intensive AI workloads. Biological computing could potentially scale AI capacity with far less overall power demand if the life support systems become more efficient.
However, the current prototype lacks published efficiency data and faces biological constraints, like neurons needing replacement or maintenance every six months. This limitation contrasts with silicon servers, which do not require such life-support upkeep, reflecting the challenges of merging living systems with technology in commercial environments.
What to watch next
Future developments will focus on improving the life support systems and expanding the operational lifespan of neuronal data centres to make them commercially viable. Observers should monitor efforts to provide clear metrics on energy efficiency per task to better compare biological versus traditional computing systems.
Interest is also growing in parallel technologies such as neuromorphic chips that mimic neurons but do not require biological life support, exemplified by projects in Europe. Meanwhile, applications named by Cortical Labs’ founder include drug discovery, humanoid robotics, and cybersecurity, signaling potential markets that might first adopt these biologically integrated computing solutions.