Moore's Law is the observation, made by Intel co-founder Gordon Moore in 1965 and revised in 1975, that the number of transistors on an integrated circuit doubles roughly every two years. It is not a law of physics but an economic and engineering trend that drove the cost per transistor down and computing power up for decades.

It depends on how you define it. Raw transistor counts are still rising, helped by larger dies, chiplets and 3D stacking, so by that measure Moore's Law is slowing but not dead. However, the doubling has slowed beyond two years, and the cost per transistor has stopped falling, so the economic version of Moore's Law has effectively ended. What truly died, around 2006, was Dennard scaling.

What is Dennard scaling and why did it end?

Dennard scaling was the companion rule that as transistors shrank, their power density stayed constant, so each new generation ran faster at the same power. It broke down around 2005 to 2006 because voltages could no longer be lowered safely and leakage current rose, causing the power wall. This is why CPU clock speeds stalled around 3 to 4 GHz and the industry pivoted to multi-core designs.

What comes after Moore's Law?

The industry has shifted from pure shrinking to other levers: new transistor structures like FinFET and gate-all-around nanosheets, chiplets that combine smaller dies, 3D stacking of silicon, advanced packaging, and specialized accelerators such as GPUs, NPUs and AI chips that do more work per transistor. This broader approach is often called More than Moore.

Is Moore's Law Dead? The Honest Answer — With an Animated Chart

Start here

What Moore's Law actually says

In 1965, Intel co-founder Gordon Moore noticed the number of transistors on a chip was doubling every year. In 1975 he revised it to doubling roughly every two years. That's it. That's the whole law:

💡 The one-sentence version

The number of transistors on an integrated circuit doubles about every two years. Note what it does not say: nothing about speed, nothing about clock frequency. People conflate it with "computers get twice as fast" — and that confusion is the source of half the "Moore's Law is dead" arguments.

Crucially, Moore's Law is not a law of physics. It's an economic and engineering observation — a self-fulfilling industry target that drove the cost per transistor down relentlessly for 50 years. That economic engine is exactly what's now sputtering.

See it move

50 years of transistors — animated

This chart plots real chips on a logarithmic scale (each gridline is 10× the one below). The dashed line is "ideal" Moore's Law — a perfect doubling every 2 years from the 1971 Intel 4004. Watch how the real chips track it for decades, then begin to need bigger dies, more cores and chiplets to keep climbing:

Transistor count per chip, 1971–2023 (log scale)

Ideal Moore's Law (2-yr doubling) Real chips Dennard scaling ends (~2006)

Figure 1 — Transistor counts still rise, but increasingly via larger/stacked silicon rather than pure density. Values are approximate, for illustration.

The real death

What actually died: Dennard scaling (2006)

Here's the part most headlines miss. The magic of the 1990s wasn't just more transistors — it was that each new generation ran faster at the same power. That bonus had a name: Dennard scaling (Robert Dennard, 1974). It said as transistors shrink, their power density stays constant, so you can clock them faster "for free."

Around 2005–2006, Dennard scaling broke:

Voltage stopped dropping — you can't keep lowering the supply voltage without the transistor failing to switch reliably.
Leakage current exploded — tiny transistors leak power even when "off."
Result: the power wall 🧱 — packing more, faster transistors meant heat you couldn't remove.

Figure 2 — After ~2006 transistor counts kept climbing, but clock speed flattened. The industry spent the extra transistors on more cores, not more GHz.

⚠ This is the real story

When people feel like "Moore's Law is dead," what they actually feel is the end of Dennard scaling: single-core speed stopped rocketing around 2006. Your 2024 laptop's cores aren't dramatically faster than a 2012 core — there are just more of them, plus smarter design.

The verdict

So — dead, or not?

The honest answer is: it depends which version you mean.

Version of "Moore's Law"	Status
Transistor count per chip still rising	Alive — but slowing (now >2-yr doubling, via chiplets/3D)
Cost per transistor keeps falling	Effectively over — costs flattened around 28nm/16nm
Dennard scaling (free speed/power)	Dead since ~2006
"Computers keep getting faster"	Yes — but via parallelism & specialization, not clock speed

So "Moore's Law is dead" is a great headline and a sloppy statement. More precise: pure geometric shrinking is hitting physical and economic limits — so the industry changed the game.

What's next

Life after Moore — "More than Moore"

Progress didn't stop; it diversified. Instead of only shrinking, engineers now pull several other levers at once:

Figure 3 — The post-Moore playbook: better transistors, chiplets, 3D, packaging, and specialized silicon.

Better transistors — planar → FinFET → gate-all-around (GAA) nanosheets reduce leakage and keep scaling viable.
Chiplets — instead of one huge, low-yield die, combine several small dies in one package (AMD's approach). Cheaper and more flexible.
3D stacking — stack silicon vertically (logic on logic, or cache on logic) to add transistors without growing area.
Advanced packaging — interposers and CoWoS put compute and memory (HBM) microns apart for huge bandwidth.
Specialization — a transistor doing the right job (in a GPU/NPU/AI chip) beats a faster general-purpose one. This is where most gains now come from.

✅ The whole page in three sentences

Moore's Law says transistor count doubles ~every 2 years — counts are still rising but slowing, and the cost-per-transistor version has effectively ended. What people really mourn is Dennard scaling, which died around 2006 and stalled clock speeds at ~3–4 GHz. Computing keeps improving anyway — through multicore, chiplets, 3D stacking, packaging and specialization ("More than Moore").

Reference

FAQ

What is Moore's Law?

Gordon Moore's 1965/1975 observation that transistor count per chip doubles about every two years. It's an economic/engineering trend, not a physical law.

Is Moore's Law dead?

Transistor counts still rise (slowing), but cost-per-transistor scaling has effectively ended. The bigger break was Dennard scaling, gone since ~2006.

What is Dennard scaling?

The rule that shrinking transistors kept power density constant — giving "free" speed. It broke around 2006 (the power wall), stalling clock speeds.

What replaces Moore's Law?

GAA transistors, chiplets, 3D stacking, advanced packaging, and specialized accelerators — collectively "More than Moore."