AI’s neural networks learn similarly to biological brains by using examples and repetition rather than following the linear logic of conventional computer programming. These networks consist of vast arrays of interconnected nodes, each functioning like a single neuron. Each node can essentially be considered an independent algorithm that shares data with other nodes, allowing the network to develop intelligence and solve problems. This learning process often demands substantial computing power to continuously simulate the behavior of each node during training. Consequently, AI’s machine learning is particularly well-suited for massive parallel processing.

Massive parallelism was once confined to the domain of supercomputers, but recent innovations in chip technology have made it possible to achieve this on a single integrated circuit (IC) substrate that contains billions of transistors. A significant breakthrough in this field is the advent of advanced packaging techniques, which enable multiple logic functions to be placed in extremely close proximity, thereby maximizing speed and minimizing energy consumption.

FormFactor is now producing probe cards capable of testing an entire wafer simultaneously, utilizing over 100,000 individual probes. We have pioneered a distinctive technology known as hybrid MEMS, which combines high-power and high-speed probes on a single probe card, meeting the needs inherent with testing AI devices.

Mike Slessor, FormFactor CEO, discusses semiconductor wafer test and the demands of AI device technology in the Mission Central video below – The Vital Role of Semiconductor Wafer Test in Artificial Intelligence Devices.