A beginner-friendly explanation of Moore's Law, why shrinking transistors made chips faster and cheaper for decades, and why that shrinking has slowed down — without assuming any prior electronics background.
If you've ever wondered why a phone from a few years ago feels slow compared to a new one, or why computers seem to get both faster and cheaper over time, the explanation traces back to one of the most famous observations in engineering: Moore's Law.
What Gordon Moore Actually Observed
In 1965, Intel co-founder Gordon Moore noticed that the number of transistors engineers could fit onto a chip, at a cost that still made economic sense, was roughly doubling every one to two years. This wasn't a law of physics — it was an observation about the pace of manufacturing improvement — but it turned out to hold remarkably well for decades, and it became something of a self-fulfilling target: the entire semiconductor industry planned its research and factory investments around keeping pace with it.
Why Doubling Transistors Actually Makes Chips Better
As covered in the companion beginner article on what a computer chip is, more transistors in the same space generally means more computing logic packed into the same chip. But the benefit compounds in a second way too: smaller transistors also switch faster and use less energy per switch, because there's less physical distance and less electrical charge involved in each switching event. So shrinking transistors historically delivered three benefits simultaneously — more computing power, higher speed, and lower power consumption — which is an unusually good combination and part of why the pace of improvement felt so dramatic for so long.
Why the Shrinking Has Slowed Down
Transistors can't shrink forever. As features approach the scale of just a few dozen atoms across, engineers run into real physical limits: at these dimensions, electrons can occasionally "leak" through parts of the transistor that should be blocking them (a quantum-mechanical effect called tunneling), and manufacturing a working chip with that little margin for error becomes exponentially more expensive. Each new generation of manufacturing tools — especially the extreme ultraviolet (EUV) lithography machines covered in the companion manufacturing article — costs more than the last, and only a handful of companies worldwide can afford to build and operate the most advanced fabs at all.
The result is that the historical doubling pace has clearly slowed. This doesn't mean chip progress has stopped — it means the industry has shifted much of its innovation toward other techniques that don't rely purely on making a single transistor smaller.
How Chips Keep Improving Anyway
- 3D stacking: instead of only spreading transistors across a flat, two-dimensional surface, manufacturers increasingly stack multiple layers of circuitry vertically, effectively adding a third dimension to how densely a chip can pack computing power.
- Specialized cores: rather than relying solely on making one general-purpose processor faster, modern chips increasingly include specialized circuitry built for specific workloads — AI/machine-learning acceleration being the most prominent current example — which can be far more efficient at that specific task than general-purpose logic running the same job.
- Advanced packaging: techniques that connect several smaller chips together, physically close and with very high-speed connections between them, so they behave functionally like one larger chip, without needing to manufacture that entire large chip as a single piece of silicon.
Why This History Matters Going Forward
Understanding Moore's Law — and honestly, its real slowdown — helps make sense of a lot of current industry news: why AI chip companies talk so much about specialized accelerator architectures rather than just raw transistor counts, why chip packaging technology has become such a major competitive battleground, and why leading-edge fabs cost tens of billions of dollars to build. If you want to see these design tradeoffs applied hands-on, the Build Your Own Silicon Chip simulator walks through a real, simplified RTL-to-silicon design flow, and Semiconductor Manufacturing: How Computer Chips Are Actually Made covers the fabrication process referenced throughout this article in full technical depth.