Singapore's newest data center is a biological marvel, a living, breathing alternative to traditional silicon chip servers. This cutting-edge facility, housed within the National University of Singapore's Centre for Life Sciences, is a testament to the innovative spirit of Australian biotech startup Cortical Labs. By utilizing human brain cells, the data center offers a unique approach to processing data, one that has the potential to revolutionize the field of artificial intelligence and robotics.
The biological data center, which went live on July 16, is a collaboration between Cortical Labs, NUS, and data center operator DayOne. It comprises 20 units of biological computers (CL1s), each housing at least 200,000 lab-grown neurons on an electrode-fitted silicon chip. These neurons, derived from blood cells reprogrammed into stem cells, exchange electrical signals with a computer, translating their neural activity into raw computing power. The facility's life-support systems, which include a cocktail of sugar, micronutrients, and pH buffers, as well as a gas mixer for carbon dioxide, oxygen, and nitrogen, ensure the cells' survival and optimal performance.
Cortical Labs' founder and CEO, Chong Hon Weng, believes that biological data centers are better suited for applications where datasets may be limited and conditions are highly unpredictable. He cites the use of humanoid robots in public spaces, homes, offices, or industries where conditions are ever-changing as an example. Traditional servers, he argues, lack the necessary data to train robots to perform complex tasks in dynamic environments. Additionally, biological data centers are ideal for cybersecurity applications, as they can spot anomalies without requiring vast training datasets.
One of the key advantages of biological data centers is their low energy footprint. Unlike traditional servers that generate heat and require cooling, brain cells use very little energy. Each CL1 unit consumes only 30 watts, which is less power than a handheld calculator, even with its life-support systems. In contrast, an AI silicon chip like Nvidia's H100 SXM can consume up to 700 watts when running demanding computing tasks, and a typical data center server with eight H100 chips can consume up to 10,200 watts. This significant energy savings is particularly relevant in Singapore, where data centers account for a substantial portion of the country's electricity consumption.
The biological data center in Singapore has the potential to address the country's constraints in terms of electricity and water. By utilizing brain cells, Cortical Labs can help overcome these challenges and contribute to Singapore's goal of greener digital growth. The facility will also aid in determining the manpower requirements and skill sets needed for commercial uses, allowing Cortical Labs to build a scientific case for manufacturing these cell types at scale, purpose-built for data center operations.
However, Chong acknowledges that traditional silicon chips are superior for fast, precise, and repeatable calculations that underpin large language models like ChatGPT. Biological data centers, while offering unique advantages, are not a replacement for traditional servers in all applications. The collaboration between Cortical Labs, NUS, and DayOne is a significant step forward in exploring the potential of biological data centers, and it will be fascinating to see how this technology evolves and impacts the future of data processing and artificial intelligence.
