RBA staff estimate Australian data-centre investment grew tenfold in three years
Imported equipment absorbs much of the spending, while grid connections and the pace of new power supply shape the buildout’s wider economic effects.
Listen to this story
The audio brief
Story brief
3 key pointsThe buildout’s economic footprint depends on what is imported and whether power and grid capacity can keep pace. RBA staff estimate data-centres added up to 0.2 percentage points to annual GDP growth over the past year after imports were excluded, while warning that supply-chain estimates rely on assumptions and spending may displace other activity in a capacity-constrained economy. Electricity is a key bottleneck: operators may outpace new renewable generation, and network upgrades can delay projects.
- 01
Servers, storage and networking equipment account for roughly three-quarters of investment; much is imported, often from Taiwan and South Korea.
- 02
AEMO estimates data-centres used about 3% of National Electricity Market operational electricity in 2025/26; Oxford Economics projects 7.5–17.1% by 2035/36.
- 03
Long-term renewable power contracts can help generators secure financing, but operators pursuing advanced projects may compete for existing supply.
Australia’s data-centre investment has grown around tenfold in three years, but the domestic economic boost is smaller than the spending suggests. A Reserve Bank of Australia staff note published November 9 estimates that imports absorb about 70% of investment, while the buildout creates new demands on electricity networks, land and construction resources.
The study uses the capital expenditure survey to estimate the investment surge. Its authors expect spending to remain significant, with its scale and timing depending on demand for computing capacity and constraints on development. The findings are staff analysis, not the stated views of the central bank or its boards.
The building is only a quarter of the bill
Constructing the warehouse-like building accounts for around one-quarter of data-centre investment. The remaining three-quarters goes into servers, storage and networking equipment. Much of that equipment comes from overseas, typically Asian economies including Taiwan and South Korea. Those purchases help explain why a large spending figure does not translate directly into an equally large increase in Australian production.
After subtracting imports, the authors estimate data-centres directly contributed up to around 0.2 percentage points of annual GDP growth over the past year. Their 70% import assumption sits within an external estimate range of 50–80%. Equipment deliveries also make investment volatile between quarters: a small number of companies import it at different stages of construction.
The mix varies by business model. Investment has grown across two types of facilities:
- Colocation centres lease space and supporting infrastructure to customers, who typically provide their own computing hardware. Much Australian investment to date has followed this model, largely serving cloud computing and storage.
- Other operators supply computing capacity directly, typically using specialised chips to run AI models. More expensive advanced chips mean imports account for a larger share of these facilities’ investment.
Local spending can create work—or move it
Domestic benefits extend beyond the building contractor. The study traces spending through suppliers, finding substantial activity in engineering and other professional services, materials manufacturing, finance, transport and wholesale trade. Together, those supplying industries generate a similarly large share of activity to building and civil engineering contractors and their construction subcontractors.
But tracing where spending lands does not prove it creates additional output. When the economy is close to capacity, workers and materials may be pulled from other projects. The authors also caution that national industry tables do not identify data-centres separately, so their supply-chain analysis requires assumptions about imports and the industries receiving domestic spending.
AEMO estimates data-centres accounted for around 3% of operational electricity consumption in the National Electricity Market.
Oxford Economics analysis for AEMO suggests data-centres could reach this share of the market’s operational electricity consumption.
New demand does not guarantee new supply
Electricity brings a similar contrast between opportunity and pressure. The note’s utilities analysis describes rising demand for long-term renewable power purchase agreements. These contracts offer generators revenue certainty and can help secure finance. RBA business contacts say some renewable projects might not otherwise have proceeded. But operators seeking projects already well advanced may instead compete for existing renewable supply.
Timing compounds that risk. Business contacts say data-centres can typically be built faster than renewable generation can come online. Some therefore expect rapid demand growth could temporarily push up wholesale electricity prices. Access to electricity networks, particularly in metropolitan areas, is already the main development constraint reported by those contacts; required network upgrades can delay connections and operations.
Sources
- rba.gov.auData Centre Investment in Australia | Insight 2026-11
Reader comments
Newest comments first. Replies stay oldest first.