IBM has unveiled a prototype AI chip called NorthPole. The chip is designed to be more energy-efficient and faster than existing AI chips.

NorthPole is based on a new type of transistor called a memristor. Memristors are able to store information in a more energy-efficient way than traditional transistors. This makes NorthPole ideal for running AI applications, which can be very computationally expensive.

IBM claims that NorthPole is 25 times more energy-efficient than existing AI chips and 22 times faster. This could lead to significant reductions in the cost of running AI applications.

NorthPole is still in the prototype stage, but IBM is hoping to commercialize the chip within the next few years. If successful, NorthPole could have a major impact on the AI industry.