Cortical Labs has announced the creation of new data centers operating based on human brain cells. Neurons grown in the lab are placed on silicon chips.
Last year, the Australian company Cortical Labs introduced the world's first biological computer, CL1, consisting of 200,000 living human neurons. Now the company plans to build biological data centers in Melbourne (Australia) and Singapore. Instead of relying on Nvidia chips like artificial intelligence companies, Cortical Labs intends to equip its facilities with racks of biological computers CL1, operating based on a much larger number of human brain cells, writes Focus citing Futurism.
Biological computers CL1 operate based on human brain cells and consume only a fraction of the energy used by conventional artificial intelligence processors, with each unit consuming less energy than a portable calculator.
Cortical Labs announced the creation of two data centers operating based on human brain cells. Neurons grown in the lab are placed on silicon chips, which could potentially compete with chips from companies like Nvidia in the future.
Instead of server racks running on conventional processors, these data centers will house biological computers known as CL1 units, operating based on human brain cells. The data center in Melbourne will host 120 CL1 units, while Singapore will have up to 1,000 units.
The neurons used by Cortical Labs, grown from human stem cells, are placed on a chip that sends electrical signals to the cells and receives them back, then records their response. This allows the software to interact with the cells and interpret their responses as computational results.
In addition to the potential ability to use the brain's capacity to store and process data, there is another advantage to using biology in this way: neurons consume very little energy. Artificial intelligence requires a lot of electricity. However, each CL1 unit consumes less energy than a portable calculator. This is orders of magnitude less energy than what a modern graphics processor requires.
But many questions remain. Most importantly, it is unclear what practical applications these units have. The company has not yet demonstrated that its computers can even remotely compete with the computing power of the current generation of top data center chips.
