According to Science and Technology Daily, a team from the Massachusetts Institute of Technology recently introduced an innovative electronic stacking technology in the latest issue of Nature magazine. This technology can significantly increase the number of transistors on a chip, thereby making the development of artificial intelligence (AI) hardware more efficient. With this new method, the team succeeded in creating multilayer chips in which alternate layers of high-quality semiconductor material are grown directly on top of each other.
It is reported that engineers have developed a new multi-layer chip design that eliminates reliance on silicon substrates and ensures that operating temperatures are kept low to protect underlying circuits. This approach allows for high-performance crystal designs that eliminate reliance on silicon substrates and ensures operating temperatures are kept low to protect underlying circuitry. This approach allows high-performance transistors, memory and logic components to be built on any random crystal surface, rather than being limited to traditional silicon substrates. This enables more direct contact between semiconductor layers, thereby improving the quality and speed of inter-layer communication and improving computing performance.
This technology is expected to be used to manufacture AI hardware in laptops and wearable devices. Its speed and functionality will be comparable to current supercomputers, and it will have data storage capabilities that match physical data centers. This breakthrough brings huge potential to the semiconductor industry, allowing chips to be stacked beyond traditional limitations, greatly improving the computing power of artificial intelligence, logic operations and memory applications.
Tags：