AEMO’s PEC Stage 2 data lands in MT PASA as Victoria’s wind lull exposes forecast gaps
AEMO’s new PEC Stage 2 outputs in MT PASA arrive as a Victorian wind drought tests the market’s ability to price scarcity.
AEMO’s Medium-Term PASA weekly dataset has begun carrying PEC Stage 2 outputs, the first unit-level look at projected capability under the new framework that went live in July 2025 for ST PASA. The initial read comes against a sharp real-world test: Victorian wind availability sat near zero at 14:30 on Thursday (2026-05-21), with a broader lull stretching across Sunday (2026-05-17) and Monday (2026-05-18).3
It matters because the ST PASA set, which extends past six days, is the market’s early-warning system for tight supply. When actual availability collapses to nothing, the gap between AEMO’s projections and what generators can physically deliver becomes the number traders and retailers are watching. PEC Stage 2 is meant to narrow that gap by publishing more granular capability data in the weekly MT PASA, not just the short-term window.3
The timing is awkward. This is nearly 10 months since the 31st July 2025 go-live date for unit-level data and other changes in ST PASA, and AEMO has been running workshops through 2026 on the ST PASA Replacement Project, including one on 11th May (2026-05-11). The new PEC Stage 2 fields in MT PASA are the first visible output from that replacement pipeline.4
What the market is looking for is whether the new data captures the kind of multi-day wind drought Victoria just experienced. The ST PASA output for that period showed the low availability clearly, but the question is whether the MT PASA weekly set, with its longer horizon, would have flagged the risk earlier. AEMO’s own software partners have been tracking the rollout since v9.12 was released in June 2025 in anticipation of the ST PASA changes.4
The backdrop is a grid where the margin for error is shrinking. AEMO’s CEO told Australian Energy Week in June (2026-06-11) that 11 data centres totalling 5.4 GW of ultimate load were already working through transmission network connection agreements in the first quarter of the year. Those connections are being factored into demand forecasts, but they are arriving into a system where supply-side transparency is still being rebuilt.6
That demand growth is not hypothetical. Origin Energy reported in April (2026-04-27) that data centres drove a 4% increase in electricity sales volumes in the March 2026 quarter versus the same period a year earlier. The utility’s quarterly report singled out the sector as the primary driver of Australian electricity demand growth.7
The same pressure is visible elsewhere. The EIA’s Annual Energy Outlook 2026 projects data centre server consumption reaching between 446 and 818 billion kilowatt-hours by 2050, with servers alone accounting for an estimated 7% of commercial sector electricity use in 2025 and growing to 22-33% by 2050. US industrial gas consumption already hit a record 23.6 Bcf/d in 2025, 1% above the prior record set in 2023.1
For the NEM, the PEC Stage 2 data is not just an operational tool. It changes how the market prices scarcity. Storage charging and discharging in AEMO’s modelling runs are decided by the model rather than assumed, and scarcity prices emerge from demand-response and value-of-lost-load tiers when supply tightens, according to analysis published in June (2026-06-02). Better capability data in MT PASA should mean those scarcity tiers trigger with more accurate timing.5
The limits are worth stating. A non-zero loss-of-load expectation is not a prediction of outage, as ENTSO-E notes in its seasonal adequacy work. Weekly LOLP is a probability that supply falls short for at least one hour in a given scenario, with the caveat that exceptional measures can always be taken. The PEC Stage 2 outputs should be read the same way.2
The first test will come with the next wind event. Traders will be comparing AEMO’s MT PASA projections for the coming weeks against what the ST PASA set actually showed during the Victoria lull, looking for how far ahead the new unit-level data can flag a collapse. If the weekly set would have shown the risk five to seven days earlier, the framework is working. If it missed it, the gap between the two datasets becomes the problem.3
The more immediate signal is whether AEMO’s next ST PASA Replacement Project workshops address how PEC Stage 2 data handles correlated multi-day renewable lulls, which is exactly the failure mode that matters for winter peaking. Victoria’s near-zero wind availability on Thursday (2026-05-21) is the benchmark case.3