How small can a transistor actually get

Every computer chip is built from a switch smaller than anything the human eye can resolve, and for decades engineers have made those switches shrink, generation after generation, so more of them fit onto a single sliver of silicon. That race has only intensified with the rise of artificial intelligence, which demands ever more processing power packed into the same physical space. But transistors cannot keep shrinking forever, and researchers who study them say the industry is now running up against limits set by atoms themselves, forcing a rethink of what “smaller” should even mean going forward.

What a Transistor Actually Does Inside a Chip

A transistor is, at its core, a tiny electrical switch etched into a piece of silicon, one of billions packed onto the chips found on a circuit board inside a phone or laptop. “And in that silicon, if you zoom in like a million times … maybe, if you’re lucky, you will start seeing these tiny switches or transistors sitting etched into that piece of silicon,” said Suman Datta, a professor of electrical and computer engineering at Georgia Tech. Each transistor flips on and off in response to pulses of electricity cycling roughly 4 billion times per second, and the precise coordination of billions of these switches is what allows a chip to move and store the 1s and 0s behind everything from a web search to an AI data center.

From Switches People Could See to a Single Controlling Atom

Transistors were once large enough to see with the naked eye; modern versions are now smaller than a human hair, which measures roughly 100,000 nanometers across, by several orders of magnitude. Researchers have already pushed the core switching mechanism about as far as physics allows: proof-of-concept devices exist in which a single atom controls the flow of electrons, Anton Persson and Tara Peña told Live Science. Persson is an assistant professor at Chalmers University of Technology in Sweden, and Peña is an incoming assistant professor at UCLA. The rest of the device surrounding that atom, however, is still far larger, meaning a complete transistor built entirely at that scale does not yet exist.

A 25-Nanometer Channel Built From Atom-Thin Materials

New materials, rather than smaller versions of the same silicon designs, are where researchers now see the most room to shrink. In a study published in Nature Nanotechnology, Persson, Peña and their colleagues built nanoribbon transistors from monolayer two-dimensional semiconductors, including tungsten disulfide, that are only one or a few atoms thick. Using a multipatterning fabrication process and a narrow “dog-bone” channel shape designed to keep the delicate material from tearing, the team shrank working transistor channels down to widths of 25 to 30 nanometers, while also improving current density in the tungsten disulfide devices roughly a hundredfold over earlier demonstrations. Because current can be controlled more precisely in a layer only atoms thick, Persson and Peña said, these materials allow smaller working transistors than conventional silicon can achieve on its own.

The Cost and Manufacturing Limits Nobody Talks About

Even when a smaller design works in a lab, getting it into products is a separate and far more expensive problem. “We are as R&D-intensive as drug companies that spend billions of dollars developing one drug,” Datta said, describing the up-front cost of designing a new chip. Persson and Peña pointed to manufacturing itself as the harder barrier: “The often much harder step is producing billions of those transistors reliably, at scale and at a price that makes sense commercially,” they said. “Something that works once in a laboratory does not necessarily work in a factory.”

Why Energy Efficiency Matters as Much as Size

Packing more transistors onto a chip only helps if each one keeps using roughly the same amount of power, since doubling the transistor count without efficiency gains simply doubles the power draw. Datta pointed to modern AI hardware as an example of what happens when that balance slips: the latest Nvidia data center processors already consume about 1.4 kilowatts of electricity, with the next generation expected to draw nearly a kilowatt more, and roughly half of that energy is lost as waste heat rather than used for computation. “So it’s not just smaller, faster, cheaper, but also more energy efficient,” Datta said. “You have to work on all the four vectors.”

Stacking Chips Upward Instead of Shrinking Them Further

With physical space for additional transistors running out, chip designers are increasingly building upward rather than only inward. “Instead of only making each transistor smaller, the idea is increasingly to stack transistors on top of one another so that more of them fit within the same chip area,” Persson and Peña said, adding that this approach can make electronics considerably more powerful even when individual transistors shrink only modestly from one generation to the next. The payoff extends well beyond AI infrastructure: smaller, more efficient transistors have historically translated into longer battery life, lower costs and smartphones that outperform the room-sized supercomputers of past decades. Whether through new materials, 3D stacking or some combination, researchers expect the underlying push to continue. “We will do everything to make it as small as possible as long as we can control the physics,” Datta said, “and as long as we can, we will find a way to make the economics work.”

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview