Australia's electricity transition is not a contest between one falling line and one rising line. It is a reorganisation of an entire market: what generates power, when it arrives, how much carbon it carries and what the system charges when supply becomes scarce. Twenty-five years of National Electricity Market data, summarised in the chart above, show these forces moving together, but not always in the same direction, and not at the same speed.
The central argument is that renewable growth has already changed the composition of the grid. The next phase is harder because it must change the grid's timing, networks and market incentives as well. Generation can move faster than the infrastructure built to support it, and the five views that follow are really one question asked five ways: has Australia built a cleaner grid, or only a grid with cleaner ingredients waiting to be coordinated.
The inheritance still shapes the transition
The NEM was built around large coal generators supplying predictable output through transmission designed to carry power from a small number of central locations. That architecture influenced dispatch, regional employment and price formation for decades. Coal's legacy therefore remains physical even when its share declines: wires, operating practices and market rules do not disappear when a new wind or solar project connects.
The chart opening this piece places that inheritance beside the emerging system. Smoothed monthly output shows coal remaining the largest source while black-coal generation trends down and wind and solar become material contributors. The important point is not that one technology instantly replaces another. It is that the distance between the old mix and the new one is narrowing persistently across the record, which is the trend the rest of this piece tries to explain rather than restate.
Annual totals conceal the operating problem
Figure 2. Monthly distributions reveal different seasonal signatures for coal, wind, hydro, solar and other generation sources.
Each fuel's monthly generation was grouped by calendar month across the full record and rendered as a boxplot, so the spread within a month is visible alongside its median rather than collapsed into a single annual average. That distinction is the entire point of the chart: an annual total can be identical for two very different portfolios.
Every technology arrives with a different rhythm. Solar follows daylight and the seasons. Wind has a broader but still variable pattern. Hydro responds to water availability, while thermal generators historically supplied a steadier floor. The monthly boxplots make clear why annual renewable totals cannot describe reliability: two portfolios can produce the same yearly energy and behave completely differently on a hot evening or a still winter morning.
This is where transmission, storage and flexible demand stop being supporting characters. Transmission shares diversity between regions, storage moves electricity between hours, and flexible demand moves consumption toward periods of abundant supply. The variability is not evidence that renewables cannot work; it is the design brief for the system around them.
The mix determines the carbon result
Figure 3. Carbon intensity differs sharply across generation sources, so a change in total output is not equivalent to a change in emissions.
Each fuel carries a fixed emissions intensity, in tonnes of CO2 per megawatt-hour; this chart weights that fixed figure by each fuel's dispatched share of generation across the record, so what's shown is the blended intensity actually delivered to the grid rather than a static per-fuel reference table. That is what lets the chart speak to composition, not just capacity.
The carbon-intensity series separates two ideas often bundled together. Meeting demand is a quantity problem; lowering emissions is a composition problem. Coal and gas can generate the same unit of electricity as wind, hydro or solar while producing a very different emissions result. That is why renewable capacity matters only when it displaces higher-intensity output in actual dispatch.
The chart also shows that the benefit is structural rather than cosmetic. A cleaner average grid emerges from repeated changes in which generators run, not from relabelling the same supply. The transition succeeds when low-carbon sources become dependable enough (through networks, storage and firming) to change that dispatch order consistently.
Cleaner does not automatically mean cheaper
Figure 4. Emissions and average market price have a weak, changing relationship, with several recent price spikes occurring away from the highest-emissions observations.
This scatter plots average settlement price against average emissions intensity for each period in the record, letting the two series be compared directly rather than argued about in the abstract. A weak or shifting correlation here is itself the finding: it means price and carbon intensity have started answering to different pressures.
The price scatter resists a comforting shortcut. Lower emissions do not mechanically produce a lower market price. Recent high-price observations sit within a system affected by scarcity, outages, fuel costs, transmission constraints and the mechanics of replacing old capacity. Price is responding to when supply is available and where it can travel, not only to how carbon-intensive it is.
This distinction matters politically. A transition that cuts emissions but leaves households exposed to volatility will lose trust. The evidence points toward a dual test: reduce carbon intensity while improving the system's ability to deliver affordable power through stressed hours.
Growth rates reveal where pressure will build next
Figure 5. Rolling growth rates show mature thermal generation changing slowly while newer renewable sources expand from a smaller base.
Each fuel's month-over-month generation change was calculated as a rolling percentage growth rate rather than a raw output difference, so a mature, high-base fuel like coal can be read on the same axis as a smaller, faster-compounding one like wind without the comparison being swamped by scale alone.
The growth-rate view is the forward pressure in the story. Coal moves slowly because it begins with a huge installed base; wind and solar can compound rapidly because they are newer and continue to attract investment. High percentage growth from a small base should not be confused with instant dominance, but repeated compounding eventually becomes system-shaping.
That creates a race between generation and integration. If renewable output grows faster than transmission, storage and flexible demand, the market can experience curtailment in abundant periods and scarcity in others. The bottleneck moves from building energy to coordinating it.
Reading the transition as one system
Figure 6. A large-format synthesis board that brings the article's five principal pieces of evidence into one readable argument.
Seen together, the charts replace the idea of a simple fuel swap with a systems problem. The generation mix is changing, but seasonality, carbon intensity, prices and growth rates show that each part of the market is moving on a different clock. The board makes the dependency visible: new supply has value only when the rest of the grid can move and firm it.
That is why the next phase will be judged less by another record for installed renewable capacity than by coordination. Transmission, storage, flexible demand and market design determine whether rapid construction becomes dependable, affordable decarbonisation. The evidence points to progress, but also to the infrastructure now required to preserve it.
The whole system has to move together
Taken together, the five views describe one transition from different angles. The generation chart shows a changing mix. Seasonality shows why timing matters. Carbon intensity shows why composition matters. The price relationship shows why affordability cannot be assumed. Growth rates show why the surrounding infrastructure is now under pressure to accelerate.
The deeper conclusion is therefore not simply that renewables are replacing coal. Australia is moving from a system organised around controllable central generation to one organised around diverse, lower-carbon and more variable resources. Success depends on whether networks, storage, demand and market design can make that new portfolio operate as one system.
Coal's decline and renewable growth are the visible transition. Coordination is the decisive one, and it's the part I find myself watching most closely: wind and solar compounding at their current rate is, on this evidence, no longer really in question. Whether the transmission lines, storage buildout and market rules keep pace with them is the part nobody can read off a generation chart, and it's the number I'd want updated first if I revisited this analysis in another two years.
Methods and original sources
The analysis combines long-run, aggregate NEM generation, rolling growth, seasonal output, carbon-intensity and market-price views, built in Python with pandas handling the groupby, rolling-window and weighted-intensity calculations and matplotlib rendering the boxplot, scatter and line charts. These reveal structural patterns but do not model every regional constraint, dispatch interval, outage or transmission bottleneck.