# The Physical Limits of Computing and Future Scaling Exploring the fundamental boundaries of hardware performance beyond traditional transistor scaling laws. - **Source:** [MIT Technology Review](https://www.technologyreview.com) - **Published:** 2026-10-19 - **Topic:** [Software & Engineering](https://brifing.si/c/engineering) - **Original article:** - **On Brifing:** https://brifing.si/s/d0e38574-5de7-4635-9009-8a397299d403 ## Feynman's Visionary Insight In 1959, physicist Richard Feynman proposed that manipulating matter at the atomic scale was physically possible. He argued that laws of physics do not prevent extreme data density, such as storing the entire Encyclopædia Britannica on a pinhead. This lecture established the concept of 'plenty of room at the bottom' for future computing. ## The Era of Transistor Scaling For decades, the industry relied on shrinking transistors to increase density and performance. This trend, famously observed by Gordon Moore, allowed computers to evolve from room-sized machines to portable devices. However, this easy descent has stalled, forcing companies to rely on massive data center buildouts to maintain performance gains through brute force. ## Stagnation of Scaling Laws Traditional scaling laws like Moore’s, Dennard’s, and Koomey’s are flattening or breaking down. Dennard scaling, which kept power consumption stable as chips shrank, ended around 2004 due to heat management issues. Consequently, progress now requires scaling up entire systems with more chips and cooling rather than simply improving individual transistor efficiency. ## Infrastructure Spending Surge Hyperscalers are currently investing over $700 billion in computing infrastructure, doubling previous records. This massive capital expenditure focuses on physical expansion, including concrete, copper, and silicon, to support the intelligence age. This strategy prioritizes immediate capacity over the fundamental efficiency gains that defined the 20th-century information age's rapid technological progress. ## The First Micro-Engineering Milestones Feynman’s 1959 challenge to build a sub-millimeter motor was met by William McLellan using a watchmaker’s lathe and microscope. Later, in 1985, Tom Newman won the second prize by shrinking text by a factor of 25,000. These early experiments proved that micro-scale manipulation was achievable, though they relied on manual skill rather than new technology. ## Quantifying Historical Progress Since the first transistor, computing density has increased by 10 orders of magnitude, while speed has grown by 12 orders. Energy efficiency has improved by 17 orders of magnitude, moving from 100 joules per operation to a single femtojoule. Despite these massive gains, the floor for performance remains significantly lower than current capabilities. ## The Ultimate Laptop Concept Physicist Seth Lloyd calculated the absolute physical limit of computation for a one-kilogram device. Such a machine could theoretically perform 5 times 10 to the 50th operations per second. However, this would require converting all rest energy into computation, creating a plasma state similar to the universe shortly after the Big Bang. ## Thermodynamic Limits of Bits Rolf Landauer established that erasing a single bit releases a minimum amount of heat, roughly 2.9 times 10 to the negative 21 joules at room temperature. Modern chips currently spend several hundred thousand times more energy than this theoretical minimum, indicating significant room for improvement in logic gate energy efficiency and thermal management. ## Information Density and Gravity Jacob Bekenstein determined that the information capacity of any bounded region is limited by its size and energy. For a device the size of a laptop, this limit is approximately 10 to the 42nd bits. This is nearly 30 orders of magnitude beyond the storage capacity of any device currently existing today. ## Communication and Quantum Limits Alexander Holevo defined the limit on how much information a stream of quanta can convey. Current optical links operate at one-millionth of this capacity. Data centers are narrowing this gap by using light for longer distances and reducing electronic overhead, moving closer to the fundamental limits set by quantum mechanics and information theory. ## The Challenge of Quantum Computing Quantum computers aim to utilize every degree of freedom in matter for computation. Currently, they are inefficient, requiring trillions of atoms to create a single reliable qubit. Because qubits are fragile, thousands must be woven together to ensure stability, making the path to reaching fundamental physical limits through quantum systems extremely complex. ## Future Research Trajectories Reaching fundamental limits requires moving beyond conventional logic. Charles Bennett showed that reversible computation could theoretically avoid heat waste by not discarding information. Exploring these boundaries serves a dual purpose: it drives the development of next-generation technologies and deepens our fundamental understanding of the physical laws governing the universe and information. ## Sources covering this story - https://www.technologyreview.com/2026/10/19/1145904/tr-explains-how-far-can-we-push-computing/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._