A trillion is a very big number. That might sound as though I’m trying to be cute by understating things, but the truth is that there are far, far bigger numbers than a mere trillion.
Nevertheless, it is very, very big by our everyday standards of measurement. We don’t go a trillion miles or talk about a trillion minutes passing by, although we do sometimes talk about a trillion dollars nowadays.
A trillion is a thousand billion, and we might be able to grasp that in a comparative manner by remembering that there are about 8 billion humans, so a trillion is more than a hundred times bigger than that.
Now, turn your thoughts to a computer transistor. This humble invention might end up changing the world more than any other single idea we’ve ever had, but that remains to be seen. In spite of its world-changing nature, the transistor is incredibly tiny.
How tiny? Let me answer that the long way, by describing how transistors used to be.
The very first transistor was made by John Bardeen and Walter Brattain at the utterly iconic Bell Labs in 1947. Bardeen and Brattain’s design was the size of a matchbox, and it was made of germanium, an element the pair found to be a semiconductor—it could operate as either a conductor or an insulator of electricity. The semiconducting nature was important since transistors would replace the much larger vacuum tubes, like those that made up ENIAC.
ENIAC and other machines like it took up several rooms of a large office building, so miniaturizing was very important.
Bill Shockley vastly improved this design, for which work he received the Nobel Prize in physics in 1956, and silicon replaced germanium due to both cost and effectiveness. Things got smaller over time, and the self-fulfilling prophecy known as Moore’s Law took over from there.
Now, we’re back to today, and to our original question: how tiny are today’s transistors? I can give you a number, but let me just bring this piece full circle by telling you that you can fit trillions of transistors on a Triscuit.
They’re small enough to be invisible—by a long stretch—and they keep getting smaller every couple of years. The wavelength of visible light is dozens of times bigger than the size of a transistor, and the very smallest ones are more than a hundred times too small to be seen.
The leading-edge transistors are already smaller than all of the viruses we know about. Soon enough, they’ll be so small that electrons will try to tunnel from one atom to another, making them all but impossible to use. We’ll hit that wall within a few years, according to researchers.
At this small of a scale, a transistor will be several atoms across. As far as we know, you can’t build a transistor with anything smaller than atoms, so this will be as small as transistors get, at least with the way we’re currently making them. For more on Moore (and Moore’s Law), read about how this self-fulfilling prophecy came true:
Think about how incredible this is. We’ve built trillions of these little on/off switches, like turning a water faucet on or off, but with electrical current. That’s why silicon is such a great material to use. It’s both abundant and effective, allowing easy switching between those two modes.
That’s all there is to a transistor, yet when you put trillions of them together, you end up with emergent properties beyond what simple yes and no answers would seem to give you. With this complexity, the modern world has been created. The personal computer revolution of the 70s was made possible by the humble transistor, and computers in homes paved the way for the internet, social media, memes, and generative AI.
Trillions of transistors can fit on a Triscuit. Without these tiny building blocks, none of this would be here. It’s worth considering just how tiny they are from time to time, and how many of them there are in the world. They’re already embedded in our homes and follow us everywhere we go.
Forget their size! Why isn't anyone talking about how transistors are tiny and cute humanoid creatures? Where can I adopt one?!
I wrote a paper early on in my electronics education in 1988. I proposed that we would soon have 1,000,000 transistors on a chip-- an (almost) outrageous claim. They have reached the limits many times now and surpassed them. 3D transistors next?