How Many Transistors Are in a Modern CPU?

Tens of billions — a mainstream laptop chip like Apple's M1 holds about 16 billion, and the current consumer record, the M3 Ultra, packs 184 billion. The interesting part isn't the number, it's the climb: the first CPU, in 1971, had just 2,250.

Five decades of doubling

The transistor timeline

Pick a chip and watch the count — and how far it has run ahead of the 1971 original.

Apple M1SoC2020 · 5 nm

16 billion

transistors

vs the 1971 Intel 4004

7 million×

more transistors

That’s about

23 doublings

≈ one every 2.2 years

Apple’s first desktop-class ARM chip — 16 billion.

Step from the 4004 up to the M3 Ultra and the “doublings” counter climbs to around 26 — a near-perfect confirmation of a rule written down sixty years ago.

“Billions” is true — but which chip?

There's no single number, because a CPU's transistor count depends entirely on the chip. A modern desktop or laptop processor sits in the range of roughly 10 to 30 billion transistors. Apple's M1 (2020) has 16 billion; the M4 (2024) around 28 billion. Push to the biggest consumer silicon and Apple's dual-die M3 Ultra reaches 184 billion — the highest in any consumer microprocessor as of 2025.

So the honest answer to "how many transistors in a modern CPU" is a range with a date attached, not a fixed figure — and it climbs every year.

The rule that predicted all of this: Moore’s Law

In 1965, Intel co-founder Gordon Moore noticed that the number of transistors on a chip was doubling at a steady cadence, and predicted it would keep going. Revised in 1975 to a doubling roughly every two years, "Moore's Law" has held with uncanny accuracy ever since.

Run the arithmetic in the tool above: from the Intel 4004's 2,250 transistors in 1971 to 184 billion in 2025 is about 26 doublings in 54 years — one roughly every two years, exactly as predicted. Few forecasts in technology have survived half a century this well.

Each one is smaller than a virus

Fitting tens of billions of switches onto a fingernail-sized die means the switches are almost unimaginably small. Today's leading chips are built on a 3-nanometre-class process; the smallest features are a few dozen atoms across. For comparison, a typical virus is 20 to 300 nanometres wide — so a modern transistor's critical dimensions are smaller than a virus, and a silicon atom itself is only about 0.2 nm.

That's also why the count keeps rising even as physical chips stay roughly the same size: each generation shrinks the transistor, so more of them fit in the same square centimetre.

The chips that dwarf a CPU

If you widen the question beyond CPUs, the numbers get stranger. Nvidia's B200 GPU (2024) carries 208 billion transistors across two linked dies. And the outright record belongs to something that isn't really a chip at all: Cerebras's WSE-3 keeps an entire silicon wafer intact as one processor — 4 trillion transistors, built for training large AI models.

A CPU is just one design point on that spectrum. The trend driving all of them is the same one Moore spotted in 1965 — it simply runs hardest now in the chips built for AI.

Game: guess the count

Five chips across five decades. The jumps are bigger than your intuition expects.

Round 1 of 5Score 0