From sand to software.
18 walk-throughs, each a scroll of chapters with a scene you can turn and controls you can change. Start with a bare wafer and follow it through the building, then follow the money, and end with what a chip actually does.
From sand to chip.
From sand to wafer.
Before the fab: silicon counted in nines, grown as one perfect crystal from a pot of melt, sawn with wire and polished to a mirror.
Anatomy of a fab.
The building as a machine: six layers, falling air, a railway on the ceiling, and a route a wafer takes a thousand times.
Writing with light.
How a lithography scanner prints a chip, from the projector principle to EUV tin plasma and double patterning.
The impossible machine.
Precision times speed: stages at 32 g, overlay under a nanometre, and a wafer every twenty seconds.
Sculpting in atoms.
Etch and deposition: plasma chambers, the deepest holes in industry, and films grown one atomic layer at a time.
Wrapping the gate.
How the transistor went 3D: planar, FinFET, gate-all-around nanosheets, backside power and what comes after.
What “2 nm” means.
Nothing on a 2 nm chip is 2 nanometres: node names against real gate and metal pitches, density, and a roadmap that grows upward.
Finding the defect.
Inspection and metrology: spotting a 10 nm flaw on a 300 mm wafer, the speed and resolution trade-off, and why every lot is sampled.
Chips on chips.
How AI chips are packaged: HBM memory towers, copper roads through silicon, hybrid bonding and giant interposers.
Follow the money.
What a chip costs.
Dies per wafer, killer defects and yield: the cost of one good chip, calculated live from numbers you set.
What a fab buys.
Where the money in a new fab goes: building against tools, which tools, and how depreciation lands on every wafer.
How fast can you build a fab?.
Terafab against the builders who never announce anything: its stated size and goal beside real buildings, real fabs and the machines on the critical path. A dated edition, updated each quarter.
Only three left.
Why only a handful of companies can make the newest chips: the shrinking count per node and the costs that did it.
The memory cycle.
Why memory prices boom and bust: the pig cycle, a two-year build lag, and a model of the cycle you can break yourself.
Where a chip gets made.
One chip, many countries: who adds what value where, the route a chip travels, and the chokepoints on the way.
Power and water.
What a fab consumes: electricity, ultrapure water and potent gases, per wafer and per country, and what that means for where fabs go.
What a chip does.
From switch to software.
How a switch becomes software: gates from transistors, an adder you can drive, memory cells, and a 32-bit instruction.
A billion ticks a second.
Clock speed explained: how short a tick is, why clocks stalled in 2005, and how wider cores, more cores and an Apple system on a chip made up for it.
Each explainer names the publishers its figures come from. Scenes are illustrations; readouts say when a number is arithmetic rather than a published figure.