Demand flexibility changes when or how much electricity people use in response to the needs of the power system. A lower peak can be valuable even when total electricity consumption barely changes. That distinction is essential when reading claims about future grid savings.
This is a retrospective analysis of the IEA report Scaling Up Demand Flexibility, published on 26 June 2026. Its Thailand case examines 2030. The results describe potential future outcomes, not savings already achieved across Thailand.
What does the Thailand scenario show?
The IEA’s executive summary says flexibility, mainly from industry, could reduce Thailand’s national peak demand by up to 13% in 2030. It also identifies the possibility of freeing up to 15% of transmission capacity on many corridors. Some lines would carry more electricity, however, so activation needs coordination.
The same assessment highlights the importance of smart controls, participation and market arrangements. A technical opportunity is not automatically a dependable service: consumers have to be able and willing to respond.
Read “up to” as a condition attached to the estimate. The result is not a promise that every business will reduce its bill by 13%, nor a forecast that Thailand will consume 13% less electricity over the year. The variable being discussed is national peak demand.
A worked example: a smaller peak without less energy
Consider a hypothetical facility that normally runs a 2 MW process for two hours during a busy evening period. It moves the same process to two hours when the grid has more spare capacity, with no change in operating power or duration.
- Electricity removed from the original window: 2 MW × 2 hours = 4 MWh.
- Electricity added in the new window: 2 MW × 2 hours = 4 MWh.
- Change in total energy in this simplified example: zero.
The facility’s load is 2 MW lower during the original window. Whether that reduces the system’s highest demand depends on when other users consume electricity. If many facilities move to the same new window, the combined profile must be checked for a new peak.
These invented numbers explain the arithmetic; they are not IEA model inputs or measurements from a Thai factory. They also exclude changes in losses and process efficiency. Our guide to megawatts and megawatt-hours explains the distinction between power and energy in more detail.
Singapore shows why programme design matters
In its 29 October 2025 announcement, Singapore’s Energy Market Authority described a Demand Response programme in which businesses reduce electricity demand when wholesale prices are high. It also described a separate Interruptible Load programme, activated by the power system operator to address supply-demand imbalances.
EMA noted that factories’ operating constraints can limit how consistently they offer interruptible demand. It also announced a 13-month sandbox with ComfortDelGro involving EV charging stations and environmental control systems on the SBS Transit Rail network.
These dated announcements illustrate different reasons for requesting a response. They do not establish the sandbox’s eventual results. A programme responding to prices should not be assumed to have identical obligations or availability to one providing a reliability service.
What a credible flexibility claim should report
The IEA’s 2025 assessment of demand flexibility presents adjusting electricity use as a way to improve the use of power-system assets. For a reader evaluating a specific project, the useful next step is to ask what was actually delivered.
- Timing: Which hours changed, and did they coincide with the problem the programme was designed to address?
- Duration: How long could the response continue?
- Comparison: How was consumption without the intervention estimated?
- After-effects: Was electricity use shifted to another period, reduced overall or supplied from storage?
- Limits: Which operating requirements restricted participation?
Demand flexibility belongs alongside investment in generation, storage and networks. It does not establish that every grid-connection constraint has disappeared. The strongest claim names the service, the period and the evidence, then distinguishes modelled potential from measured delivery.