Power has a shape across the year and across the day. What that does to an indexed hosting rate, and how to plan production around it.
A fleet owner with hardware at two sites on the same rate per kilowatt hour can still see one site cost noticeably more than the other across a given stretch of the year. Neither number was wrong. What changed is the shape underneath it (the pattern of highs and lows a power price traces across the day and across the year), and an average rate does not tell you where you sit inside that pattern at any given moment.
Why power has a shape at all
Electricity is not stored at scale the way most other commodities are, so the price at any moment reflects what is being generated and consumed right then, not what was generated last week. Demand rises and falls through the day as homes, industry and transport draw more or less power, and generation rises and falls with it, whether from dispatchable plant coming on and off or from weather-dependent sources that produce more in some conditions than others. The result is a price curve rather than a flat line, with recurring patterns across the day and, layered on top of that, recurring patterns across the year as heating, cooling and daylight hours shift what is being demanded and what is available to meet it.
The shape is not specific to any one grid or any one operator. It is a structural feature of how power markets work, and it shows up in different shapes in different regions depending on the generation mix and the demand pattern local to that grid. On many grids winter is where that shape peaks, as heating demand rises and daylight hours fall, though the timing differs with the generation mix. The point to take from it is that the shape exists at all, and that a single average rate is a summary of a curve, not a description of it.
Where curtailment and load fit into the shape
The same conditions that push a price curve up or down are often the conditions that make a grid tighter or looser at a given moment. When demand is high relative to available generation, prices tend to rise, and that is also when a grid operator is more likely to ask large flexible loads, including a hosting site, to reduce draw. When demand is low relative to generation, prices tend to fall, sometimes towards levels that make running flexible load attractive rather than costly. A hosting agreement that includes curtailment provisions is, in effect, built around this same shape: the periods most likely to trigger a curtailment request are correlated with the periods where an indexed rate is also likely to be at its least favourable, and the periods least likely to see any request are correlated with the more favourable end of the same curve.
Check this directly rather than assuming it works in your favour by default. Ask how curtailment events have been distributed across the day and across the year at a given site, in pattern rather than in specific figures, and ask whether the contract's indexation and its curtailment terms are describing the same underlying shape or two unrelated ones.
What an annual average conceals
An average rate is arithmetically correct and can still be a poor guide to any individual month or week, because it collapses a curve with real highs and lows into one number that describes none of the periods inside it. A fleet owner who budgets against the annual average alone will find that some stretches run comfortably under it and others sit well above it, and if the higher stretches happen to land on a period where you were also planning peak production, the two effects compound rather than cancel.
This is the practical difference between an indexed rate and a fixed one, seen from where a customer sits rather than from the contract language describing either. A fixed rate removes the shape from your own experience of the bill, even though the underlying market shape has not gone anywhere; someone in the chain is still carrying that variability. An indexed rate passes the shape through to you directly, in exchange for typically tracking the underlying market more closely on average. Neither structure is simply better. What matters is whether you know which one you signed, and whether your production planning already accounts for the shape it exposes you to.
Planning production against a curve, not a point estimate
Treating a quoted rate as a single planning input is where the average becomes a liability rather than a convenience. A more useful approach starts from the shape itself: ask what the site's historical price curve has looked like across a full year, not a single average, and use that curve, alongside whatever forward visibility the operator can offer, to plan where higher-draw activity sits relative to it. This matters most for anything genuinely discretionary (work that can be timed rather than work that has to run continuously regardless of price), since that is the only category where the shape of the curve changes your options.
The same applies to budgeting. A monthly figure based on the annual average, with no allowance for the spread around it, tends to look fine on a spreadsheet and then produce a surprising invoice in whichever month happens to sit furthest from that average. Building a range into the budget instead of a single figure, and asking the operator what the historical spread around their own average has looked like, gives you something closer to what you will see across a real year of billing.
What to ask before you sign against an average
Before treating a quoted average rate as the number that describes your year, get clear on the following:
- Whether the rate is indexed or fixed, and if indexed, to what reference and on what notice.
- What the site's historical price curve looks like across a full year, not only the average.
- How curtailment events at the site have tended to line up with the higher-price part of that curve.
- Whether your own planned production, or anything discretionary in it, can be timed around the shape rather than run flat against it.
- Whether you may switch off in peak-price hours, and whether hours you choose not to run are still billed.
We go through the components that make up a hosting rate, and which of them are fixed against which are indexed, in how industrial power pricing works for mining. The mechanics of curtailment itself, including what it costs and how caps are typically structured, are covered separately in curtailment cost and caps. Our hosting rate page sets out how the energy part of a rate can be fixed, indexed or capped, and why each of those shapes removes a different risk.
