Curtailable GPU offtake, live on the grid.
Project CURTAIL pairs the FlexSys scheduler with Flipped Energy's wholesale and retail market integrations. When the network operator needs less load, FlexSys drains GPU work without dropping in-flight requests — and bids the freed capacity back into the wholesale spot market.
A regulator-overseen demonstration, not a pitch deck
CURTAIL is being run on the live Transgrid network under the Demand Management Innovation Allowance (DMIA) — real infrastructure, independently measured and verified.
Data centres running AI workloads are, from a grid operator's perspective, a large and growing block of inflexible load. CURTAIL tests whether that load can be made flexible — curtailed on request, without breaking the workloads running on it — and paid for that flexibility the same way any other demand-response resource is.
The scheduler reads AEMO Pre-Dispatch and 5-minute spot price signals directly. Workload placement reflects the price the grid is paying right now. Curtailment events are measured at the data centre's point of common coupling, targeting IEC 61000-4-30 Class A instrumentation — the same standard used for regulatory grade metering, not an engineering estimate.
- For the grid: a new source of fast, verifiable demand response, sized to the AI buildout already underway.
- For the operator: a way to monetise flexibility instead of just absorbing a flat power bill.
- For the workload: nothing in flight gets dropped — flexing happens at the scheduling layer, not by pulling the plug.
Network operator funding, independent verification
CURTAIL isn't a Flipped Energy claim about itself — it's funded by the network operator and checked by parties with no stake in the outcome.
Milestones
Updated as the pilot moves from endorsement to live deployment.
Curtailable capacity, deployable now.
Deployable today via ResetData, with the same operator and control plane described on the architecture page.