From Transistors to Chips, and Moore's Law
On this page 5 headings
Start here A modern phone chip has billions of transistors. That’s more tiny switches than there are people on Earth, packed into something smaller than your fingernail. No human can design billions of switches one by one. So how is it done?
Engineers never design billions of switches one at a time. They build in levels, like LEGO:
| Level | Made of | Roughly how many inside |
|---|---|---|
| Transistor | One tiny switch | 1 |
| Gate (AND, OR, NOT…) | A few transistors | 2–8 transistors |
| Cell / small block | A group of gates | 2–16 gates |
| Module (adder, memory block…) | Many cells | thousands of transistors |
| Chip (IC) | Many modules | millions to billions of transistors |
| Circuit board (PCB) | Several chips | about 8–16 chips |
| Device (phone) | Boards, screen, battery | about 8–10 chips/boards |
At each level, you only need to know what a block does, not every switch inside it. This trick is called abstraction: hide the details, use the block.
Putting billions of transistors on one chip is called VLSI (Very Large-Scale Integration). In simple words: managing billions of transistors.
In 1965, Gordon Moore (who later co-founded Intel) noticed a pattern:
The number of transistors that fit on a chip keeps doubling at a steady rate.
- In his 1965 paper, “Cramming more components onto integrated circuits”, he said it doubled every year.
- In 1975 he changed this to about every two years, and that’s the version people remember.
How fast is “doubling every two years”?
- After 2 years: 2×
- After 10 years: 2⁵ = 32×
- After 20 years: 2¹⁰ = 1,024×, about a thousand times more
That’s why your phone today is far more powerful than a room-sized computer from the 1960s.
| Year | Milestone |
|---|---|
| 1947 | First transistor (Bell Labs; it later won the Nobel Prize) |
| 1954 | First silicon transistor |
| 1963 | CMOS logic invented (the kind used in almost all chips today) |
| Late 1960s–1970s | First commercial CMOS chips; used in digital watches because they use very little power |
| 1971 | Intel 4004, the first single-chip microprocessor: about 2,300 transistors |
| 2020s | The biggest chips have tens of billions of transistors, more than the number of people on Earth |
In one line: Engineers build huge chips level by level (transistor → gate → module → chip → board → device), and Moore’s Law says the number of transistors on a chip has doubled about every two years. Exam words: transistor, gate, module, integrated circuit (IC), PCB, abstraction, VLSI, Moore’s Law, Gordon Moore, Intel, doubling, CMOS, microprocessor.
Spot it: “A chip has X transistors today. According to Moore’s Law, will it have Y after T years?”
Steps:
- How many times bigger? Y ÷ X.
- How many doublings is that? Find n with ÷ X.
- Each doubling takes 2 years, so the time = 2n years.
- Compare with T.
Worked example: 2 million today → 32 million in 7 years? True or false?
- 32 ÷ 2 = 16
- 16 = 2⁴ → 4 doublings
- 4 × 2 = 8 years
- The question says 7 → FALSE. (With 8 years it would be TRUE; with 4 years, FALSE.)
Trap 1: Moore’s Law is about the number of transistors doubling. “The size of transistors doubles every two years” is FALSE: transistors get smaller, which is why more of them fit. Trap 2: use 2 years per doubling (that’s what this course and the exams use), not 1 year.