Every five minutes, Victoria’s electricity market sets a wholesale price. On sunny days that price now regularly drops below zero, which means generators are paying to keep producing. In 2019 that happened for about 62 hours across the whole year. In 2025 it happened for about 2,098.

The question: how has the timing of Victoria’s negative wholesale prices changed since 2016, and how does that change line up with rooftop solar output and the demand left for the grid to meet?

62 → 2,098
hours with a negative wholesale price in Victoria, 2019 vs 2025
55%
of the 12pm–1pm hour in 2025 was spent below zero, up from 1% in 2016
3.2×
rooftop solar output compared with the demand met through the grid, at 1:00 pm (market time) on 27 December 2025*

*AEMO’s estimate of rooftop output ÷ AEMO’s scheduled grid demand, both for the end of the same 5-minute interval. See “Methods and data checks” at the end.

Visualisation 1 of 5 · How often

A rare glitch became routine

Each bar shows the share of all hours in a year when the price was below zero, and the coloured segments show what each time of day contributed to that total. In 2019 the figure was 0.7%. In 2025 it was 24%, roughly one hour in every four. The timing changed too: in 2016, 79% of the few negative hours fell overnight, while in 2025 the midday band (amber) alone contributed 45% of them.

Hover over a segment for its hours. Click a time band in the legend to hide it, double-click it to show it alone, and double-click again to bring all bands back. 2026 is a partial year (January–September), so it sits apart from the full years.

Data table for this chart
Hours with a negative wholesale price, Victoria
Year Hours below $0 % of all hours of which midday (10am–3pm)
2016 88 1.0 9
2017 4 <0.1 0
2018 28 0.3 11
2019 62 0.7 32
2020 384 4.4 165
2021 1,320 15.1 612
2022 1,212 13.8 608
2023 1,895 21.6 968
2024 1,855 21.1 942
2025 2,098 24.0 947
2026 (Jan–Sep) 1,357 20.7 536

Source: Australian Energy Market Operator (2026a), aggregated price and demand data, Victoria (VIC1), Jan 2016 – Sep 2026. Shares are time-weighted across 30-minute (pre-Oct 2021) and 5-minute intervals.

Visualisation 2 of 5 · When

It starts at lunchtime, and the window keeps widening

Each square is one hour of one month; darker blue means a larger share of the trading intervals in that hour had a price below $0. Before 2020 the grid is almost entirely pale: in 2019, prices between 10am and 3pm were negative just 1.8% of the time. A dark band then forms around midday, reaching 34% in 2021 and passing half, 53%, in 2023, where it has stayed (52% in 2025). The darkest square of 2025 is November at 12 pm, when the price was below zero 92% of the time. The band is also spreading: at 8am the share rose from 31% in 2024 to 36% in 2025, and at 5pm from 10% to 17%. A fainter band has appeared overnight too. That isn’t solar, and this data can’t show what drives it.

Hover over any square for its month, hour and share. Compare spring (Sep–Nov) across the years: in 2025 that is where the band is darkest. Times are AEMO market time (AEST, no daylight saving), so in summer the solar peak sits an hour earlier than clock time.

Data tables for this chart
Share of time with a negative price, by time window (full years)
Year % of 10am–3pm below $0 % of 8–9am below $0 % of 5–6pm below $0
2016 0.5 0.1 0.0
2017 0.0 0.0 0.0
2018 0.6 0.1 0.0
2019 1.8 0.1 0.0
2020 9.0 3.8 0.8
2021 33.6 18.6 6.5
2022 33.3 20.5 5.2
2023 53.0 28.9 9.9
2024 51.5 31.1 10.4
2025 51.9 36.2 16.8
Share of 10am–3pm with a price below $0 (%), by month: the midday rows of every panel
Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
2016 <1 0 0 0 0 0 0 <1 0 3 2 <1
2017 0 0 0 0 0 0 0 0 0 0 0 0
2018 0 0 0 1 0 0 0 1 2 0 0 3
2019 0 2 0 <1 0 0 2 4 1 4 2 5
2020 8 2 <1 11 3 <1 <1 6 20 13 10 33
2021 35 30 14 21 11 7 17 45 64 77 34 47
2022 22 32 40 19 8 5 17 33 37 48 60 80
2023 58 65 62 38 12 28 43 44 76 71 69 71
2024 65 70 54 37 17 6 12 51 78 92 63 73
2025 65 55 45 49 35 9 26 29 72 76 90 73
2026 (Jan–Sep) 57 48 57 60 25 12 17 27 51 – – –

Source: Australian Energy Market Operator (2026a), aggregated price and demand data, Victoria (VIC1), Jan 2016 – Sep 2026.

Visualisation 3 of 5 · Alongside

Midday demand on the grid fell as the evening peak grew

Each line is one year’s average scheduled grid demand (the demand met by large generators and interconnectors) at each time of day. In 2016 (dark grey, dashed), midday demand sat close to the evening level. By 2025 (blue), average grid demand between 10am and 3pm had fallen 24%, from 5,343 MW to 4,043 MW, over the same years that negative midday prices took hold. The evening average (5pm–9pm) rose from 5,678 MW to 6,099 MW. The midday fall is largest in summer (32%) and smallest in winter (11%), which matches the seasons with the most and least sunshine. The next chart shows a single day, where rooftop output and grid demand can be measured side by side.

Switch between Summer and Winter to compare the size of the midday dip. Hover anywhere to compare every year at the same time of day; click a year in the legend to hide it.

Data tables for this chart
Average scheduled grid demand, Victoria, all seasons
Year Midday 10am–3pm (MW) Evening 5–9pm (MW) Evening minus midday (MW)
2016 5,343 5,678 335
2017 5,178 5,607 429
2018 5,061 5,626 565
2019 4,911 5,660 748
2020 4,609 5,574 965
2021 4,414 5,622 1,207
2022 4,458 5,797 1,339
2023 4,077 5,647 1,569
2024 4,008 5,861 1,853
2025 4,043 6,099 2,056
Average scheduled grid demand by season, 2016 vs 2025 (midday 10am–3pm, evening 5–9pm)
Season Midday 2016 (MW) Midday 2025 (MW) Midday change (%) Evening 2016 (MW) Evening 2025 (MW)
Summer 5,331 3,631 -32 5,325 5,673
Autumn 5,330 3,884 -27 5,633 5,906
Winter 5,696 5,098 -11 6,344 7,181
Spring 5,011 3,543 -29 5,404 5,620

Source: Australian Energy Market Operator (2026a), aggregated price and demand data, Victoria (VIC1), 2016–2025. Scheduled grid demand is AEMO’s TOTALDEMAND, averaged by half-hour.

Visualisation 4 of 5 · One day

27 December 2025: the grid’s quietest moment

This day holds the lowest grid demand in the January 2016 – September 2026 data, whether measured in 5-minute intervals or half-hour averages. Grey is scheduled grid demand; amber is AEMO’s estimate of rooftop solar output at the same moment.

At 1:00 pm market time (2:00 pm on Melbourne’s daylight-saving clock), grid demand fell to 1,225 MW. Rooftops were producing an estimated 3,967 MW, 3.2 times as much, and the wholesale price was −$105/MWh. The price stayed below zero for about 14 hours that day.

The menu adds the next two lowest-demand days that have rooftop estimates, and a weekday with low rooftop output (Fri 19 Dec 2025): with average midday rooftop output of 1,174 MW instead of 3,834 MW, more demand fell on the grid, the price never went below zero, and the day’s lowest grid demand came at 3:25 am, not around midday.

Choose a day from the menu. Complete days of AEMO rooftop estimates run from 4 December 2025 to 3 February 2026, so the first three options are the three lowest-demand days in that window; across 2016–2026 they rank 1st, 2nd and 4th. (1 January 2025, the 3rd-lowest day, falls before the rooftop estimates begin.) The last option is a comparison weekday picked for its low midday rooftop output. The thin dark line is grid demand in 5-minute intervals, and the callout marks each day’s lowest 5-minute value. Hover to read grid demand, rooftop output and price at any time.

Data tables for this chart
The four days in the menu
Day Rank, 2016–2026 Lowest 5-min grid demand (MW) Time (market time) Price then ($/MWh) Mean rooftop output 10am–3pm (MW) Hours below $0
Sat 27 Dec 2025 1st 1,225 1:00 pm -105 3,834 14
Thu 25 Dec 2025 2nd 1,289 1:00 pm -207 3,171 24
Thu 1 Jan 2026 4th 1,409 1:35 pm -110 3,663 12
Fri 19 Dec 2025 comparison day 4,522 3:25 am 80 1,174 0
Rooftop solar, scheduled grid demand and 5-minute price, 27 December 2025, 10am–4pm
Time (market time) Rooftop solar (MW) Grid demand (MW) Price ($/MWh) Rooftop share (%)
10:00 am 3,485 1,733 -100 67
10:30 am 3,680 1,597 -100 70
11:00 am 3,844 1,441 -101 73
11:30 am 3,932 1,370 -132 74
12:00 pm 3,966 1,342 -178 75
12:30 pm 3,945 1,273 -178 76
1:00 pm 3,967 1,225 -105 76
1:30 pm 3,895 1,297 -101 75
2:00 pm 3,835 1,357 -105 74
2:30 pm 3,733 1,529 -135 71
3:00 pm 3,542 1,743 -95 67
3:30 pm 3,262 1,990 -66 62
4:00 pm 2,928 2,401 -60 55

Sources: Australian Energy Market Operator (2026a), aggregated price and demand data, VIC1; Australian Energy Market Operator (2026b), rooftop PV actual estimates (MEASUREMENT type), VIC1, Dec 2025 – Feb 2026. Each rooftop estimate is matched to the 5-minute interval ending at the same time.

Visualisation 5 of 5 · Where

Where the rooftops are

By 30 September 2026, Clean Energy Regulator records show 6,091 MW of small-scale solar installed in Victoria since 2001; the figure at the end of December 2025 was 5,661 MW. That is installed capacity, not output, and the two are measured differently: at the low point on 27 December 2025, AEMO estimated rooftop output across Victoria at 3,967 MW. Postcode 3029 (Hoppers Crossing, Tarneit and Truganina) has more than any other: 194.1 MW across 28,370 systems. The biggest totals sit in Melbourne’s growth corridors: 3029 and 3030 in the west, 3064 in the north, and 3977 and 3978 in the south-east. Measured by density instead, the picture shifts inward: the most tightly packed postcodes are 3070 (Northcote) and 3056 (Brunswick), with 2,469 and 2,400 kW per km² of rooftop solar.

The two layers answer different questions. Colour shows density (how much solar is packed into each square kilometre); circles show each postcode’s total, drawn only for the 53 postcodes with at least 25 MW so that smaller ones don’t crowd the map. Hover or tap a postcode for its figures, switch layers in the box at top right, and zoom out to see regional Victoria.

Data tables for this chart
Ten Victorian postcodes with the most installed small-scale solar (circles layer)
Postcode Installed (MW) Systems Area (km²) kW per km²
3029 194.1 28,370 117.8 1,648
3064 146.0 20,895 132.2 1,105
3977 125.9 20,595 142.2 886
3030 123.5 19,120 240.1 514
3978 88.4 12,718 96.5 916
3175 64.4 5,683 73.8 872
3350 58.2 8,547 86.7 671
3810 57.5 8,695 142.7 403
3806 56.3 8,996 45.8 1,231
3551 55.4 8,252 1,767.5 31
Ten Victorian postcodes with the most solar per square kilometre (colour layer)
Postcode kW per km² Installed (MW) Systems Area (km²)
3070 2,469 14.8 2,795 6.0
3056 2,400 12.0 1,839 5.0
3027 2,289 8.8 1,437 3.8
3066 2,144 2.8 284 1.3
3057 2,101 4.7 810 2.3
3204 2,082 17.4 2,740 8.4
3058 2,043 23.9 3,786 11.7
3071 2,007 10.3 1,706 5.1
3065 2,001 2.8 410 1.4
3068 1,965 8.0 1,451 4.1

Sources: Clean Energy Regulator (2026), small-scale installation postcode data (SGU-Solar capacity and installations, 2001 – Sep 2026); Australian Bureau of Statistics (2021), ASGS Edition 3 Postal Areas. Capacity counts systems up to 100 kW, including upgrades; it measures what is installed, not what is generated.

Midday versus evening

The five charts show the same shift from different angles. Measured over the same windows and at the same half-hourly resolution, midday and evening have moved apart:

Victoria, midday (10am–3pm) versus evening (5pm–9pm). All values use half-hourly average prices and demand.
Window Median price 2016 ($/MWh) Median price 2025 ($/MWh) Half-hours below $0, 2016 (%) Half-hours below $0, 2025 (%) Mean scheduled grid demand 2016 (MW) Mean scheduled grid demand 2025 (MW)
Midday (10am–3pm) 37 -4 <1 52 5,343 4,043
Evening (5pm–9pm) 50 124 0 7 5,678 6,099

Source: Australian Energy Market Operator (2026a), aggregated price and demand data, Victoria (VIC1), 2016 and 2025, averaged to half-hours.

In 2016 the two windows looked alike: the median midday price was $37/MWh against $50 in the evening, and both were below zero less than 1% of the time. By 2025 the median midday price was −$4/MWh, below zero in 52% of half-hours, while the evening median was $124/MWh. Over the same period, grid demand fell at midday and rose in the evening. The data shows these changes happening together and lining up with rooftop output on the days we can measure; it does not isolate rooftop solar as the only cause. Wind output, interconnector flows, generator bidding and how flexibly coal plants can turn down all affect prices too.

These are wholesale prices, the price generators are paid in the market. They are not household bills, which also cover network, retail and environmental costs and are mostly set in advance. The point is timing. In 2025 the typical midday wholesale price sat just below zero while the typical evening price was $124/MWh, and the evening, not the middle of the day, is now when grid demand is highest.

Methods and data checks

Data and periods. AEMO aggregated price and demand data for Victoria (VIC1): 626,688 trading intervals from January 2016 to September 2026, downloaded as 129 monthly files on 27 October 2026. AEMO rooftop PV actual estimates (MEASUREMENT type) for 4 December 2025 to 4 February 2026, 3,024 half-hourly values; where AEMO revised an estimate, the latest revision is used. Clean Energy Regulator small-scale solar capacity and installations by postcode, January 2001 to September 2026. ABS 2021 Postal Area boundaries (GDA2020).

Definitions. Scheduled grid demand is AEMO’s TOTALDEMAND: the demand met by scheduled and semi-scheduled generators and interconnectors. It excludes smaller non-scheduled generators and rooftop solar, so it is narrower than AEMO’s “operational demand”. Rooftop solar is AEMO’s modelled estimate of output at the end of each half-hour, not metered generation. Rooftop share = rooftop ÷ (rooftop + scheduled grid demand). Because small non-scheduled generators are left out of the denominator, this share slightly overstates rooftop solar’s share of all consumption.

Time conventions. All timestamps are AEMO market time (AEST, UTC+10, no daylight saving) and mark the end of each interval. Hours of the day refer to the interval’s start. The market settled every 30 minutes until 1 October 2021 and every 5 minutes after.

Alignment. AEMO’s rooftop figure is an estimate of output at the end of each half-hour, and its 5-minute scheduled demand is the demand target for the end of each dispatch interval. For the single-day chart, each rooftop estimate is therefore matched to the 5-minute interval ending at the same moment; all 3,024 rooftop values found a match. This is why the record is reported as 1,225 MW at 1:00 pm (a single 5-minute interval), while half-hour averages give 1,244 MW for the half-hour to 1:00 pm and an average price of −$152/MWh. Either way the rooftop share is about 76% (76.4% vs 76.1%). Elsewhere, 5-minute data is averaged into half-hours.

Negative-price shares are weighted by interval length. As a check, the same yearly shares were recalculated from half-hourly average prices (a common resolution for every year): the largest difference in any year is 1.6 percentage points (2024: 21.1% from interval-weighted prices vs 22.7% from half-hourly averages), so the trend does not depend on the change in settlement interval.

Interval lengths are taken from the gap between consecutive timestamps rather than a fixed changeover time, because AEMO’s files switch to 5-minute intervals at 12:05 am on 1 October 2021. As a check, every full year adds up to exactly 8,760 hours (8,784 in leap years).

Selected days. The single-day chart can only show days with a complete set of rooftop estimates (4 December 2025 to 3 February 2026). The three lowest-demand days in that window rank 1st, 2nd and 4th across 2016–2026. (1 January 2025, the 3rd-lowest day, falls before the rooftop estimates begin.) Callouts mark each day’s lowest 5-minute grid demand; where that moment falls between two half-hour marks, the callout gives the rooftop estimate for the nearest half-hour mark and says which one.

Checks. Duplicate timestamps were removed; 0 gaps were found in the interval sequence and 0 half-hours were missing a 5-minute interval. The CER monthly figures were summed and matched the CER’s own totals column (national installations exactly; capacity to within 1 kW of rounding). Of 694 Victorian postcode boundaries, 693 matched a CER record and one with no record was excluded. Postcode area is the ABS figure (AREASQKM21); boundaries are simplified to a 50 m tolerance for display only. 2026 is a partial year and is excluded from seasonal averages and full-year comparisons. The lowest-demand day is the same at 5-minute and half-hourly resolution. CER records count systems when certificates are created, which can be up to 12 months after installation, so recent months may be incomplete.

Reproducibility. The three download scripts and this R Markdown file reproduce every number and chart from the original sources.

References

Australian Bureau of Statistics. (2021). Australian Statistical Geography Standard (ASGS) Edition 3: Postal Areas – 2021 – Shapefile (GDA2020) [Data set]. Retrieved October 27, 2026, from https://www.abs.gov.au/statistics/standards/australian-statistical-geography-standard-asgs-edition-3/jul2021-jun2026/access-and-downloads/digital-boundary-files

Australian Energy Market Operator. (2026a). Aggregated price and demand data: Victoria (VIC1), January 2016 – September 2026 [Data set]. Retrieved October 27, 2026, from https://aemo.com.au/energy-systems/electricity/national-electricity-market-nem/data-nem/aggregated-data

Australian Energy Market Operator. (2026b). Rooftop PV actual estimates (ROOFTOP_PV/ACTUAL), December 2025 – February 2026 [Data set]. NEMweb. Retrieved October 27, 2026, from https://nemweb.com.au/Reports/ARCHIVE/ROOFTOP_PV/ACTUAL/

Cheng, J., Schloerke, B., Karambelkar, B., Xie, Y., & Aden-Buie, G. (2025). leaflet: Create Interactive Web Maps with the JavaScript ‘Leaflet’ Library (Version 2.2.3) [Computer software]. https://doi.org/10.32614/CRAN.package.leaflet

Clean Energy Regulator. (2026). Small-scale installation postcode data: SGU-Solar capacity and installations, 2011 to present and totals [Data set]. Retrieved October 27, 2026, from https://cer.gov.au/markets/reports-and-data/small-scale-installation-postcode-data

Coverdale, C. (2026). aemo: Download Australian Energy Market Operator Data (Version 0.4.2) [Computer software]. https://CRAN.R-project.org/package=aemo

Pebesma, E. (2018). Simple features for R: Standardized support for spatial vector data. The R Journal, 10(1), 439–446. https://doi.org/10.32614/RJ-2018-009

R Core Team. (2025). R: A language and environment for statistical computing (Version 4.5.2) [Computer software]. R Foundation for Statistical Computing. https://www.R-project.org/

Sievert, C. (2020). Interactive web-based data visualization with R, plotly, and shiny. Chapman and Hall/CRC. https://plotly-r.com

Wickham, H., Averick, M., Bryan, J., Chang, W., McGowan, L. D., François, R., Grolemund, G., Hayes, A., Henry, L., Hester, J., Kuhn, M., Pedersen, T. L., Miller, E., Bache, S. M., Müller, K., Ooms, J., Robinson, D., Seidel, D. P., Spinu, V., … Yutani, H. (2019). Welcome to the tidyverse. Journal of Open Source Software, 4(43), 1686. https://doi.org/10.21105/joss.01686

Acknowledgements

This project draws on publicly available datasets provided by the Australian Energy Market Operator (AEMO), the Clean Energy Regulator (CER), and the Australian Bureau of Statistics (ABS). These sources provided the electricity market, rooftop solar installation, and geographical data used in the analysis. Data processing, statistical analysis, and visualisations were carried out using R and relevant open-source packages, including tidyverse, ggplot2, plotly, leaflet, and sf. The contributions of these organisations and software developers are acknowledged in the reference list.