What is the ultimate size limit for transistors?
Engineers have been shrinking transistors for decades to boost computing power, but physical limits are approaching. Experts discuss the fundamental constraints on transistor size and whether smaller is still the primary goal in the age of AI.
Researchers have already demonstrated switching mechanisms at the single-atom scale, though surrounding circuitry remains larger. This marks the fundamental physical boundary for the active switching element itself.
A June study in *Nature Nanotechnology* employed two-dimensional semiconductors, such as tungsten disulfide, to reduce nanoribbon transistor channels to 25 nanometers wide—roughly 0.00025 the width of a human hair. This highlights continued progress in shrinking the active region.
The relentless pursuit of smaller transistors, accelerated by AI demands, could reshape computing economics and energy usage. If atomic-scale limits halt traditional scaling, the industry may pivot to novel architectures or materials, potentially slowing the pace of AI advancement or shifting focus to efficiency. This could affect the cost and environmental footprint of data centers, as well as the performance trajectory of consumer electronics, though the exact timeline remains uncertain.