For decades, engineers have worked to create smaller transistors to fit more of them onto computer chips, aiming to increase processing power. In the age of artificial intelligence (AI), this effort has intensified. Transistors are tiny switches that turn on and off in response to electrical pulses, enabling the operation of devices from smartphones to AI data centers. The smaller the transistors, the more can fit on each chip, increasing speed and functionality. Researchers have built proof-of-concept devices where a single atom controls electron flow, though the rest of the device remains larger. A study published in June in the journal Nature Nanotechnology used two-dimensional semiconductors like tungsten disulfide to shrink nanoribbon transistors to a channel width of 25 nanometers — about 0.00025 the width of a human hair. Two-dimensional semiconductors, which are only one or a few atoms thick, allow for more precise control of electrical current, enabling smaller transistors than traditional silicon. Over the next decade, silicon transistors are expected to shrink, but more slowly than in the past. To go considerably smaller, moving beyond silicon to atomically thin materials may be necessary. Size is not the only consideration; cost, manufacturing, and energy efficiency are also important. Designing and building new chips is expensive, with upfront costs comparable to those of drug development. Manufacturing billions of transistors reliably at scale and at a commercial price is a significant challenge. Energy efficiency is also crucial, as adding more transistors without improving their efficiency can double power consumption. The latest Nvidia data center GPUs consume 1.4 kilowatts of electricity, with the upcoming release expected to consume nearly a kilowatt more. Half of that is lost as waste heat. Engineers are exploring 3D designs to increase power without solely focusing on shrinking transistors. Stacking transistors on top of each other allows more to fit within the same chip area, increasing overall power even if each transistor shrinks only modestly. The quest for faster, more efficient chips is urgent due to the growth of AI. Smaller transistors and more powerful chips benefit both AI data centers and consumer electronics, potentially leading to faster electronics, longer battery life, and lower costs. As long as the physics can be controlled and the economics are viable, efforts to make transistors as small as possible will continue.