For two lessons we’ve talked about first-principles thinking in the abstract: strip a problem down to what’s actually true, then rebuild. Now let’s run the machine on a real example — one that turned a “settled fact” into a multi-billion-dollar company. The fact in question: how much a battery costs. The twist: it wasn’t a fact at all. It was an assumption everyone had stopped checking.
”Batteries cost $600 per kWh — they always have”
Rewind to around 2008. If you’d asked any sensible engineer or analyst what a lithium-ion battery pack cost, you’d have gotten a confident answer: about $600 per kWh (per kilowatt-hour, the unit of energy a battery stores). Want a 60 kWh car battery? That’s roughly $36,000 — for the battery alone, before the car around it.
And why did everyone know it cost $600? Because it always had. Last year it was about that. The year before, about that. This is reasoning by analogy — the mental shortcut from lesson 2: the new thing resembles the old thing, so it’ll behave like the old thing. Batteries have always cost ~$600/kWh, therefore batteries cost ~$600/kWh. Tidy. Confident. And, as it turns out, an assumption wearing the costume of a fact.
The first-principles question is rude by comparison. It ignores the price history entirely and asks: what is a battery physically made of, and what would those raw materials cost if I bought them on the open metal market?
That’s a question with an actual answer. A battery isn’t made of “$600.” It’s made of stuff — specific metals, in specific amounts, that all have a price you can look up. So let’s look them up. Don’t take my word for the total — rebuild it yourself.
Before you read — take a guess
Before you peek: if a battery pack 'costs' $600/kWh, roughly what do you think its raw materials — the actual metals — cost on the open market?
Decompose, then rebuild
What does a battery pack really cost?
The 'by analogy' price
$600/kWh
Battery packs had always cost about this much per kWh — so everyone assumed they always would. That's reasoning by analogy.
Click each raw material to add its real cost and rebuild the pack from scratch:
About these numbers
Treat every figure here as illustrative and approximate. Real material prices swing year to year and depend on the exact battery chemistry. We’re not claiming the history happened to the penny — the teaching point is the method and the size of the gap, not the decimals.
Read the result
Add up what you just rebuilt and here’s the picture. The metals and films that physically make up the pack:
| Raw material | What it does | Illustrative cost ($/kWh) |
|---|---|---|
| Nickel | Bulk of the cathode metal | 22 |
| Cobalt | Stabilises the cathode | 18 |
| Lithium & electrolyte salts | Carries the charge | 14 |
| Graphite (carbon anode) | The other electrode | 11 |
| Aluminium | Current collectors & casing | 7 |
| Separator & polymers | Keeps electrodes apart | 5 |
| Steel can & casing | Holds it together | 3 |
| Raw materials total | ~80 |
So the stuff costs about $80/kWh. The “fact” said $600. Where did the other ~$520 go?
| Line item | Illustrative cost ($/kWh) |
|---|---|
| Raw materials (the physics floor) | ~80 |
| Everything else: manufacturing, design, factory scale, margin — and the assumption | ~520 |
| Conventional price | ~600 |
That bottom row is the whole lesson. The “$600 is just what a battery costs” was not a fact about physics — physics only demands the ~$80 of metals. It was a fact about how batteries had been made so far: in small volumes, with old processes, at high margin, by people who’d never seriously asked whether it could be cheaper. Decomposing the pack into raw materials didn’t lower the price by itself — but it revealed that roughly $520 of the price was negotiable. And a negotiable $520, multiplied across millions of cars, is a company.
Fill in the move that did the work:
Pick the right option for each blank, then check.
Reasoning by analogy said batteries cost $600 'because they always had.' First principles instead , found they summed to about $80, and exposed the remaining $520 as an assumption rather than a law of nature.
The general recipe this illustrates
The battery story isn’t a battery trick. It’s a four-step recipe you can run on almost any “that’s just what it costs / how long it takes / how it’s done” claim. We met the pieces in lessons 2 and 3 — here they are assembled:
- State the apparent cost or limit. Write down the conventional figure, plainly. (“A battery pack costs $600/kWh.”) This is usually an analogy in disguise — it’s been this way, so it is this way.
- Decompose into irreducible parts. Break the thing into pieces that are physically or factually real and can’t be argued away. (Nickel. Cobalt. Lithium. A steel can.)
- Cost or bound each part from what you actually know. Look up the real price — or the real minimum — of each piece, independent of the conventional total. ($22 of nickel, $18 of cobalt…)
- Rebuild, then measure the gap. Add the parts back up to get a floor. The distance between that floor and the conventional figure is where the opportunity (or the overpaying) lives. ($80 floor vs. $600 price → a $520 gap.)
This works far away from batteries:
- A restaurant meal. Menu price: $28. Decompose: the ingredients on the plate might be $6 of food. The other $22 is the cook, the rent, the waiter, the dishwasher, the margin — real costs, but now you can see what you’re paying for, and you know a $9 home version is physically possible.
- A scheduled task. “This report takes a week.” Decompose: pulling the data (2 hours), writing it up (3 hours), one round of review (2 hours). The irreducible work is maybe a day. The other four days are queueing, context-switching, and “it’s always taken a week.” Same shape, no metals involved.
A founder hears 'launching a basic software app costs $500,000 — that's just the going rate.' Applying the battery recipe, what's her FIRST move?
Pitfall: the rebuilt floor is a lower bound, not a price tag
Here’s where eager first-principles thinkers faceplant. The rebuilt $80 is what’s physically necessary, not what you can actually buy a pack for. Manufacturing, yield losses, R&D, equipment, and risk are real costs, not pure waste sitting there for the taking. A battery factory that produces zero defects, hires no engineers, and pays for no machines does not exist.
So first principles tells you what’s possible, not what’s free. The $80 floor said “a pack could cost far less than $600” — and over the following years, packs did fall dramatically toward that floor, but it took enormous factories and years of engineering to close the gap. Nobody walked into a metals market with $80 and walked out with a battery pack.
Don't confuse the floor with the price
“The materials cost $80” is not the same as “I can buy a pack for $80.” The floor tells you the gap is worth attacking. Closing it is the actual work — and that work has real, irreducible costs of its own.
Sort each cost into the irreducible physics floor (the ~$80) versus the everything-else that first principles flags as negotiable but not free.
Place each item in the right group.
- R&D and engineering salaries
- Margin and 'it has always cost this much'
- Nickel for the cathode
- Lithium charge-carrying salts
- Factory machinery and assembly labour
- The steel can
Recap: the battery move at a glance
Big picture
First principles on a battery pack
- Battery pack
- Analogy says
- $600/kWh
- 'always has cost this'
- First principles asks
- What is it made of?
- What do those metals cost?
- Decompose & rebuild
- Nickel, cobalt, lithium…
- Sums to ~$80 floor
- The gap
- ~$520 = the opportunity
- Floor ≠ purchase price
- Analogy says
Run the recipe
In the example, the raw materials sum to about $80/kWh against a conventional $600/kWh. Roughly what fraction of the price was materials?
Check your answer to continue.
Where this goes next
You’ve now seen first-principles thinking earn its reputation: it found roughly $520 of opportunity that an entire industry had walked past, simply by refusing to accept “it’s always cost this much.” Decompose, price the parts, rebuild, measure the gap.
But notice how much work that was — looking up seven materials, costing each, rebuilding the total, then carefully not over-claiming. First principles is powerful, and it is slow. You can’t run it on every decision; you’d never get out of bed. Reasoning by analogy, the very shortcut we just dunked on, is fast and usually good enough — which is exactly why your brain defaults to it.
So the real skill isn’t “always use first principles.” It’s knowing when. In the next lesson, we flip the script: when is analogy actually the smarter move, and how do you tell a question that deserves a full decomposition from one that doesn’t?