Power prices move. Everyone in hyperscale infrastructure knows this. What is less understood is how much the value of a single operational decision changes depending on when it is made.
If your facility can shift 10 MW of workload during an hour of peak demand, that decision is worth very different amounts depending on your power contract, your grid market, and the time of day. At off-peak rates, shifting that workload may recover modest value. During a peak demand event in a volatile grid market, the same decision can be worth an order of magnitude more.
The consequence chain that triggers the workload shift decision does not care about the power market. It fires when the thermal or compute anomaly crosses a threshold. But the value that gets captured depends entirely on what the power market is doing at that moment.
The decision does not change. The value captured changes by 10x depending on when the system makes it.
Most hyperscale operators are on long-term power purchase agreements. The base rate is fixed. But that base rate almost never covers the full picture.
Demand charges are typically calculated on the peak 15-minute interval in a billing period. One bad 15-minute window sets your demand charge for the month. Reducing peak draw during that window is worth real money — not the base rate per kWh, but the demand charge rate, which can be three to five times higher.
Ancillary service markets — frequency response, spinning reserve, demand response — pay for the ability to shed load on short notice. A facility that can reliably demonstrate 10 MW of dispatchable demand reduction is worth a premium to grid operators. That premium is only accessible if the facility can act fast enough to participate.
The difference between 0.8 seconds and 22 minutes is the difference between participating in a demand response event and missing the response window entirely. The value of that participation is not captured in the thermal domain. It is captured in the power billing domain. But you only get there through the thermal event chain.
Here is what makes this interesting. The most valuable power market opportunities are correlated with exactly the conditions that produce the most consequence events.
High-demand periods drive grid price spikes. High-demand periods also produce the most thermal stress on cooling infrastructure. A hot summer afternoon that pushes spot power prices to $300 per MWh is the same afternoon that your CRAC units are working hardest and most likely to produce a consequence event requiring a workload shift.
If you are running a system that can trace that consequence chain in under a second, the thermal event is your early warning signal for a power market opportunity. You shift the workload for thermal reasons. You also capture demand response value that would have been inaccessible without the speed.
If you are running manual operations, you resolve the thermal event after the market window has closed.
The gain-share math that most operators think about is based on a flat $/kW value. That is a reasonable starting point. It is also a significant underestimate for facilities in markets with demand charges and ancillary service participation opportunities.
A more complete model weights recovered kW by when they were recovered. Off-peak recovery gets the base rate. Recovery during peak demand windows gets a premium that reflects demand charge avoidance. Recovery during grid stress events with active demand response participation gets the full market value of that event.
Synestra's EA tracking captures this timing data. We know not just how many kW were recovered but when they were recovered and what market conditions were active. That makes the gain-share calculation accurate rather than approximate. And it means the value you see in year two of a gain-share agreement reflects the full value of the intelligence, not just a flat-rate estimate.
Facilities that are built with real-time consequence intelligence from day one have access to a class of economic value that facilities without it cannot reach.
This is not about recovering value after a problem occurs. It is about having the speed and cross-domain visibility to participate in market events that conventional operations cannot respond to in time.
The power price window is narrow. It opens and closes in minutes. The consequence chain that creates the opportunity can be traced in under a second. What you do with those 59 seconds determines whether you capture the value or watch it expire.