Opinion: AEMO's Governance Review Inherits a Market Its Models Cannot See

AEMO's Governance Review Inherits a Market Its Models Cannot See Victoria's wholesale electricity price closed Friday at $10.50 per megawatt-hour, down 84.8% on the day, while South Australia settled at $44.08, off 71.7%. These are not anomalies requiring explanation. They are the new baseline arithmetic of a grid where rooftop solar capacity has surpassed total coal-fired generation capacity, and a market operator is preparing to review its own institutional mandate without reliable dispatch data on its largest generation source. The governance review AEMO announced Friday arrives framed by a comfortable headline: renewables reached 42.1% of NEM generation in the June quarter, wholesale prices fell to their lowest Q2 average since 2020, and the tender pipeline shows 32.3 gigawatts of committed capacity. The framing invites a self-congratulatory reading. The self-congratulation is premature. The problem with the 32.3 GW pipeline figure is that financial close on new projects collapsed 46% in 2025 relative to prior years, according to Clean Energy Council project tracking, while the tender pipeline itself grew. These two numbers move in opposite directions, which means the pipeline is not a leading indicator of build, it is an inventory of projects that cannot secure financing at current price levels. At approximately 2.3 GW of annual financial close, the committed capacity additions entering service each year have been outpaced by the ambition of developers who cannot get projects across the line. Into that gap walk 17 data centre projects seeking 9 GW of firm connection capacity, per AEMO's published connection queue. At 2.3 GW of annual financial close, servicing those requests alone, assuming they all proceeded and consumed the entire committed-capacity pipeline, would absorb roughly four years of supply additions before a single new renewable megawatt served any other customer. The low-price environment that looks like a success story for consumers becomes, under this arithmetic, a financing environment actively preventing the supply additions needed to serve the demand wave building behind it. The counter-argument here is worth confronting directly rather than dismissing. Current low prices may be transitory, the product of coal fleet outages compressing thermal generation, mild Q2 weather softening demand, or simply the seasonal pattern of an autumn quarter with high solar yield and modest industrial activity. Moreover, not all 17 data centre projects will reach financial close; several may revise their connection requests as their own financing becomes complicated by the same forward price uncertainty affecting renewable developers. The 9 GW figure is an expression of developer appetite, not contracted load. If the price signal normalises as coal exits and storage fills the gap, the argument runs, the financing market recovers and the data centre demand overhang gets absorbed over a longer build cycle. This argument has some merit for the near term. It has almost none for the structural trajectory. Gas-fired generation fell 30% in the June quarter to its lowest June output since 2003, before most east-coast LNG infrastructure existed. Gas peakers earn revenue in the hours when renewables cannot set price. Rooftop solar, which AEMO cannot dispatch and therefore cannot model in its dispatch stack, is compressing exactly those hours. This is not a temporary pattern driven by coal outages. It is the logical consequence of behind-the-meter capacity exceeding total coal capacity while AEMO's reliability calculations treat it as an absence rather than a presence. Every adequacy assessment AEMO publishes is structurally blind to its single largest generation source. The governance review cannot address a problem it cannot see. The data centre demand question also cuts differently depending on the load profile. Reduced or speculative volumes are a reasonable hedge against overstating the demand wave. What is not debatable is the type of load being requested: data centres require 24/7 firm connection, not variable industrial demand that can be curtailed in tight periods. A governance framework calibrated to variable demand will struggle with continuous firm load regardless of whether 9 GW materialises or 5 GW does. The modelling problem runs deeper still. WattClarity's review of Project Energy Connect Stage 2 flags that the NSW1-SA1 interconnector is now appearing in MT PASA data, forcing NEMDE, historically a hub-and-spoke linear programming model, to process loop flows for the first time. Traders pricing long-dated renewable contracts against a four-year-low spot baseline are calibrating against a dispatch engine whose behaviour under this new topology is genuinely unknown. Publishing forward reliability assessments against an unvalidated model is not governance. There is also a competitive dimension no analyst in this discussion has quantified. Wallumbilla gas settled at $11.05 per gigajoule on Friday. JKM, the Asian LNG benchmark, closed at $22.94 per MMBtu. At a standard gas-to-power heat rate, JKM implies fuel costs for Asian gas-fired generation that dwarf NEM wholesale prices. Australian aluminium smelters, chemical producers, and miners operate at a structural cost discount to Asian competitors that predates any governance review and is widening. That wedge does not appear in AEMO's framing of what this review is for. So what should the review actually prescribe? Three things, specifically. First, AEMO needs a mandatory real-time reporting framework for behind-the-meter generation that creates a dispatchable proxy for rooftop solar. Treating it as negative demand in aggregate is not sufficient when it constitutes the largest generation source in the NEM. Second, financial close rate must become a headline reliability metric alongside tender pipeline counts. A 46% collapse in financial close while the pipeline grows is a structural warning signal that should trigger automatic review of connection charging and contract terms, not a footnote in an annual report. Third, AEMO must publish an independently verified confidence interval for MT PASA loop-flow modelling under PEC Stage 2 topology before using those outputs to inform capacity adequacy decisions. Traders cannot price long-dated contracts against numbers derived from a model whose behaviour is acknowledged to be unknown. The AEMO chief executive's comments Friday about expanded workload are accurate. The workload has expanded. What the governance review must answer is whether the institution's modelling tools, reporting frameworks, and reliability metrics have expanded commensurately. Friday's Victorian spot price, $10.50, generated largely by capacity AEMO cannot dispatch, suggests the answer is no.