Press "Enter" to skip to content

Singapore opens the world’s first AI data center powered by 16 million living human neurons

Singapore has unveiled the world’s first biological data center, using 16 million human neurons grown from stem cells to power AI systems.

The project is a collaboration between NUS Medicine, digital infrastructure company Day One and Australian biotech startup Cortical Labs. The facility consists of 20 biological computers, each containing around 800,000 neurons, which communicate with software through Cortical Labs’ proprietary biOS operating system.

The neurons are created from pluripotent stem cells and grown on silicon platforms equipped with microelectrode arrays. These electrodes send electrical signals to the neurons and record their responses, allowing the biological network to process information much like a conventional processor.

Unlike traditional servers, the neurons must be kept alive. Technicians feed them every three days with sugars, micronutrients and buffering solutions, while a gas system supplies oxygen, nitrogen and carbon dioxide. Each culture can survive for up to six months before being replaced.

A potential alternative for AI data centers

The technology remains experimental, but its developers believe biological computing could offer an energy-efficient alternative for certain AI applications. Neurons can adapt by forming new connections, while requiring far less power than conventional silicon chips.

Each CL1 unit consumes around 30 watts, compared with more than 700 watts for a high-end Nvidia H100 SXM chip. The neurons also do not require conventional cooling systems and can operate at around 37°C.

However, biological computers still face major limitations, particularly their slower processing speeds and lower reproducibility compared with silicon-based systems. For now, they are not capable of replacing the hardware needed to run large language models such as ChatGPT.

Be First to Comment

Leave a Reply

Discover more from Fritz Mag

Subscribe now to keep reading and get access to the full archive.

Continue reading