01
The inversion.
A conventional operator buys power every month, forever, at a price someone else sets. This platform buys its generator once.
That single move turns the dominant operating line into a capital line. Spread across a twenty-five-year asset life, the effective cost of energy is both lower and — far more usefully — knowable.
A platform that earns better than its peers in one good year is interesting. A platform that earns differently than its peers across twenty-five years is investable.
Volume I · Chapter 14
02
Three asset lives,
three kinds
of money.
Separating them is what makes the structure financeable. One pool of capital cannot sensibly fund all three.
| Layer | Life | How it is funded |
|---|---|---|
| Land | Permanent | Contributed as equity; does not depreciate |
| Energy | 25 years | Long-dated capital; one planned mid-life battery augmentation, reserved at construction |
| Compute | 3–5 years | Pre-funded by a sinking-fund provision inside operating cost |
Nothing about the refresh is a surprise expense. It is scheduled, reserved for, and it is also the moment the platform re-chooses its workload.
03
Production,
not treasury.
The distinction that decides whether this is infrastructure or speculation.
- A treasury vehicle buys.
- It acquires digital assets at market price and holds them. Its question is whether the price goes up.
- This platform produces.
- It runs proof-of-work computing on electricity it has already paid to generate, at a marginal energy cost approaching zero. Its question is whether the market price of what it produces stays above its own cost of production.
- Which is a different question.
- One is a bet on a price. The other is a position on a cost curve. Only the second is an infrastructure investment.
The position of being the lowest-cost producer is not a comfort — it is the thesis.
Volume I · Chapter 14
04
The honest
part.
There is a price at which every producer stops, including the cheapest one.
The platform is not insulated from a price collapse. What the design changes is its position in the queue: with a lower cost base, it is built to remain productive after higher-cost operators have curtailed.
And when a workload stops paying, the hardware is re-pointed at one that does. That option exists because the expensive part underneath is already built.
No capital cost, internal rate of return or payback period is published on this site. Indicative ranges across conservative, base and favourable cases — and the model behind them — are released under non-disclosure. Publishing a single number would imply a precision the platform does not claim.