Growth forecasts have been revised up ~4x in two years: the 2024 vintage (3.22%) is now a museum piece
Growth forecasts for Australian data centres have been revised up roughly fourfold in two years, occupancy has grown about fortyfold since 2005, and investment is on an A$20bn to A$60bn trajectory this decade. Each figure with its publisher and vintage.
Growth, four ways: realised and forecast, revenue and physical
Forecasts of capacity, hyperscale revenue and investment published after the December 2024 AirTrunk transaction and the 2025 AI-contract wave sit at double-digit growth (Mordor 15.25% and 20.88%; Arizton 13.52% on an investment basis). Broad revenue forecasts have stayed single-digit: IMARC's 2026 edition projects 5.47% a year for 2026-34, and Mordor's own revenue series 4.25% for 2026-31. The only pre-transaction forecast in the set, Arizton's May 2024 edition, sat at 3.22%. The upgrade is concentrated in the measures the AI contracts drive: capacity and investment.
How to use the dispersion
- For revenue planning. we run 8 / 12 / 16% on the services-revenue basis (Section 10 scenarios). The published revenue-basis rates bracket it, from IMARC's 5.47% to Mordor's 20.88%; Arizton's 13.52% counts investment spend and sets no revenue rate here.
- For capacity planning. we anchor on AEMO's demand-derived path (~6 GW by 2030) rather than any research-house CAGR, because the grid is the binding constraint (Section 06).
- Mordor's 20.9% hyperscale CAGR. is the AI-bull-case proxy and is treated as an upper bound; the base case does not use it.
inashanu sizing bridge
2026 core revenue: IBISWorld J5921 A$4.1bn (2026 forecast) + OD5323 A$1.5bn (2026) = A$5.6bn core hosting/colocation revenue, 2026. Broad-market alternative: IMARC US$4.8bn (2025) ~ A$7.3bn at the source-implied 0.66 USD/AUD. We carry A$5.6bn as 'core' and A$7.3bn as 'broad' throughout; nothing in this report depends on choosing between them.
Published CAGR estimates, %
| Item | Value (%) |
|---|---|
| Arizton 2024 vintage, pre-AI boom (to 2029, US$ investment) | 3.22 |
| IMARC solution+services (2026–34, US$) IMARC (2026) | 5.47 |
| IBISWorld J5921 historic (2020–25, A$) IBISWorld (Oct 2025, Mar 2026) | 7.6 |
| IBISWorld OD5323 historic (2021–26, A$) IBISWorld (Oct 2025, Mar 2026) | 9.2 |
| Arizton via R&M (2025–31, US$ investment) | 13.52 |
| Mordor, capacity (2025–30, MW) Mordor Intelligence (Jan-Mar 2026) | 15.25 |
| Mordor, hyperscale revenue (2026–31, US$) Mordor Intelligence (Jan-Mar 2026) | 20.88 |
Published CAGR estimates differ 6x on scope, metric and vintage: the report treats growth as a range, not a point
Historic growth confirms durability
IBISWorld's realised (not forecast) CAGRs, 7.6% (2020-25) for cloud hosting/data processing and 9.2% (2021-26) for internet hosting, were achieved before the AI capex wave. The floor under this market was set by cloud migration; AI adds a second engine on top.
Occupancy grew ~40-fold in 20 years: two-thirds of it since 2020, with AI demand up ~27x
M3 Property's demand-side series is the cleanest long-run evidence: 37 MW of national occupancy in 2005 became 1,315 MW in 2025. M3 frames this as a 40-fold expansion; recomputing from the published endpoints gives ~36x; we show the endpoints and let readers judge. AI-workload-specific demand rose ~27x over the same window.
M3 Property occupancy series, MW
| Item | Value |
|---|---|
| 2005 | |
| 2020* | |
| 2025 |
National occupancy, MW taken up by users (demand side): 19.5% p.a. 2005-25 (inashanu recompute)
Operating capacity is ~1.35 GW: the '2 GW headline' is a different metric, not a bigger market
Australian capacity estimates, GW
Knight Frank 2025 'built IT capacity' (UNVERIFIED)
M3 Property built-out capacity, 2025 (broader basis)
CBRE expected live within three years
CBRE live capacity, end-2025
ITK Research operating IT load, end-2025
The '2 GW headline' confuses operating IT load with gross site MW: metrics differ by definition, not just by source
Which number should a committee use?
For revenue underwriting: ~1.35 GW operating IT load (ITK, converging with CEFC/Baringa, USSC and Oxford Economics). For development pipeline: M3's ~2,000 MW built-out. Treat Knight Frank's 4.5 GW as unverified (poor OCR retrieval, data-gap register item 3).
CBRE expects live capacity to reach ~1.8 GW within three years. M3 projects:
Capital is arriving faster than megawatts: investment is on an A$20bn-to-A$60bn trajectory this decade
US$6.7bn (~A$10bn) was invested in Australian data centres in 2024, second only to the US globally (Knight Frank, cited in the National AI Plan). CBRE forecasts investment growing ~50% over four years to A$46bn in 2029 on committed projects; Oxford Economics expects A$60bn by 2030 from A$20bn in 2026. The National AI Plan cites A$100bn+ of future investment.
Investment trajectory indicators, A$bn
| Item | Value |
|---|---|
| 2024 actual Knight Frank Knight Frank via National AI Plan / Quartz (23 Apr 2026) | |
| 2026 Oxford Economics Oxford Economics via API Magazine (8 Sep 2026) | |
| 2029 CBRE committed CBRE via de Valence (Dec 2025) | |
| 2030 Oxford Economics Oxford Economics via API Magazine (8 Sep 2026) |
Bases differ: Knight Frank is an actual annual flow; CBRE is committed-project based; Oxford Economics is a run-rate trajectory. Figures are presented side by side and not blended.
Global position
| Market | Operating / trajectory | Note |
|---|---|---|
| United States | 24,000 to 31,000 MW (early 2030s) | ~50 GW operating on ITK's basis; world's largest |
| India | 2,000 to 10,000 MW (early 2030s) | Fastest large-market build-out |
| Australia | ~1,350 to 5,000-6,220 MW (2030) | Top-10 now; potentially top-3 early 2030s (M3) |
Sovereign-AI tailwind
Australia is one of only four APAC nations exempt from US AI-chip export restrictions: a recurring, cited driver for locating GPU clusters here, and a structural advantage none of the demand forecasts fully price.
Facility economics of scale
- ~2,000 MW Built-out capacity
- ~9,500 MW Expanding toward, by the early 2030s
- ~A$26bn Implied construction investment to 2030 (M3)
Sources
- IBISWorld, as at Oct 2025, Mar 2026
- IMARC, as at 2026
- Arizton via ResearchAndMarkets, as at Feb 2026
- Mordor Intelligence, as at Jan-Mar 2026
- Arizton via PR Newswire, as at 9 May 2024
- M3 Property, as at Nov 2025
- AEMO/Oxford Economics, as at 2025 IASR
- M3 Property, Data Centre Growth in Australia, as at 18 Nov 2025
- The Industrialist, as at 24 Nov 2025
- ITK Research, as at 11 May 2026
- CBRE via LinkedIn/Naicker, as at 2026
- Knight Frank, as at 9 Apr 2025, unverified
- Knight Frank via National AI Plan / Quartz, as at 23 Apr 2026
- de Valence/Substack, as at 27 Dec 2025
- CBRE via de Valence, as at Dec 2025
- Oxford Economics via API Magazine, as at 8 Sep 2026
- Gadens via EcoGPT, as at 4 Dec 2025
- IMARC/openPR, as at Apr 2026
- ITK Research, as at May 2026
Related
- Why do Australian data centre market size estimates differ so much?
- What is in Australia's data centre development pipeline?
- How much data centre capacity will Australia have in 2030?
- How much have the hyperscalers committed to Australian data centres?
The full primer is published as an 85-page PDF (download) with the underlying market model as a workbook (download).
This page is one of 30 in Australian data centres: the research, indexed by the question each one answers, with the four pages where the published sources disagree set first.
The research is published by inashanu, a studio that produces fixed-scope market research for Australian private equity, priced and dated before work starts. The sample engagement is this primer in full.
This standard, on your question. Six fixed-price deliverables, A$14,000 to A$45,000 excluding GST, opening at the A$14,000 market map in 10 business days. State the decision and the date it has to be made; the written assessment of whether the evidence base will carry it follows within two business days and carries no charge. Request a brief assessment · the catalogue and its prices.