IBM has unveiled its new brain-inspired chip ‘NorthPole’

Applications for the new silicon chip could include robotics and driverless cars

When it comes to artificial intelligence tasks, IBM's brain-inspired NorthPole processor is over twenty times quicker and nearly 25 times more efficient in terms of energy than any microchip now available on the market. Applications for the new silicon chip could include robotics and driverless cars, per an IBM study. The goal of brain-inspired computer hardware is to replicate the remarkable speed and energy efficiency of the human brain in computation. Neural networks, which mimic the way a brain functions and learns, are frequently implemented using these machines. One approach that is frequently used in brain-inspired electronics is modeling the data-storage and computation processes of biological neurons.

Combining memory and processors can significantly cut down on the energy and time that computers waste transferring data between those two parts. Developed over the course of eight years, NorthPole expands upon IBM's previous brain-inspired microprocessor, TrueNorth. When TrueNorth made its debut in 2014, its power efficiency was four times lower than that of traditional microprocessors of that time. NorthPole has now outperformed all other chips available, including those manufactured with more sophisticated nodes. Because of NorthPole's efficiency, large liquid-cooling systems are not necessary for it to function; fans and heat sinks are more than sufficient, according to IBM. This implies that it may be used in much smaller areas.