Nanoscale 3D transistors made from ultrathin semiconductor materials can operate more efficiently than silicon-based devices, leveraging quantum mechanical properties to potentially enable ultra-low-power AI applications.
The disadvantage seems to be that these transistors can only control very tiny currents. They currently lack enough ass to control much else.
Not familiar with this particular paper but transistors are good at acting as switches for higher powered loads so I wonder if these ultra tiny ones can be used to power on slightly less tiny versions which can handle larger circuits.
That’s basically how IGBTs in power electronics work, in stuff like trains and electric cars. It’s a sensitive, easily activated voltage-driven MOSFET driving a larger BJT transistor in a chain.
Also how Darlington pairs work. So, yeah, maybe they could do all the computation at that level and then cascade the output through larger transistors to talk to the outside world.
Not familiar with this particular paper but transistors are good at acting as switches for higher powered loads so I wonder if these ultra tiny ones can be used to power on slightly less tiny versions which can handle larger circuits.
That’s basically how IGBTs in power electronics work, in stuff like trains and electric cars. It’s a sensitive, easily activated voltage-driven MOSFET driving a larger BJT transistor in a chain.
Also how Darlington pairs work. So, yeah, maybe they could do all the computation at that level and then cascade the output through larger transistors to talk to the outside world.