3% GDP Data Centre Booms Import Trap?

The Quarter That Grew From One Source

Australia's economy expanded by 0.3 per cent in the March quarter — but the number that demands attention is not the headline figure. It is the 16 per cent surge in machinery and equipment spending, the largest single-quarter jump in nearly three decades, which accounted for the entirety of that growth.

Private investment climbed almost 4 per cent. Westpac economists stated plainly that data centre investment, including spillover effects, drove all of the growth this quarter. Without that single category, the March quarter would not have been a miss — it would have been a contraction.

The ASX 200 closed up 0.8 per cent to 8,534 points. The energy sector led all gains at 1.9 per cent, followed by industrials at 1.3 per cent and consumer cyclicals at 1.2 per cent. The two sectors that finished lower — consumer non-cyclicals down half a per cent and education down 2.6 per cent — were precisely the segments exposed to household income pressure rather than capital expenditure. The market was not celebrating the economy; it was repricing which parts of the economy now matter.

The RBA raised the cash rate to 4.35 per cent in May, reversing a brief easing cycle that markets had treated as the beginning of a cutting path. That reversal now sits against a GDP print that is materially below the prior quarter's 0.6 per cent pace. The 3-year bond yield retreated after the GDP data was released — the bond market read the miss as reducing the probability of further tightening. The equity market reached a different conclusion. Both are responding to the same number. The unstated premise separating them: bond markets are treating the GDP miss as the signal; equity markets are treating the investment composition as the signal. Those two premises cannot both be correct at the same August meeting.

When the Import Bill Arrives

The data centre investment surge that held the March quarter together carries a condition that the headline figure does not show. Australia does not manufacture the servers, the networking equipment, the specialised cooling infrastructure, or the power management hardware that a hyperscale data centre requires. NextDC chief executive Craig Scroggie noted that the top four hyperscalers — Microsoft, Amazon, Google, and Meta — will invest more than US$680 billion in digital infrastructure and AI globally in 2026. The fraction of that directed toward Australia arrives as capital inflow and records as private investment. The machinery purchased records as domestic spending. The payment for that machinery exits as an import.

The 16 per cent machinery and equipment surge is therefore two things simultaneously: a genuine investment signal and a future current account pressure. Capital expenditure today becomes import drag in the trade balance tomorrow. Westpac acknowledged this when they noted that the headwinds would be more fully reflected in the second quarter, with the possibility of a quarterly contraction.

This is the condition that the equity market's current repricing does not yet require an answer to. Infrastructure names and energy companies are pricing in the demand the data centres create — electricity load, construction services, grid connection work. What they are not yet pricing is what happens to the domestic growth signal if the hyperscaler capex cycle decelerates, or if the RBA responds to the import component lifting inflation rather than to the GDP headline suggesting it should pause.

Residents in Melbourne's West Footscray are already living beside a data centre construction site that operates continuously. The Climate Council has flagged that artificial intelligence data centres carry enormous electricity and water requirements that are beginning to stress both grid capacity and municipal water planning. If those physical constraints begin to bind — and the grid pressure is not abstract, given that Australia absorbed an energy shock from the Iran conflict on top of three consecutive rate hikes — then the investment trajectory that carried the March quarter cannot simply be extrapolated into the June quarter.

The market participant who moved first into infrastructure and energy names on the GDP release was pricing the demand side. The bond market, which sold the 3-year yield lower after the same release, was pricing the demand destruction side. One of those two positions will need to confirm or reverse before the August RBA meeting.

August Decision and the Verification Test

The unresolved question from the March quarter data is whether the RBA reads the GDP miss as a reason to pause or reads the import-heavy investment surge as a reason to remain concerned about price pressure. Money markets are currently pricing a 50-50 chance of a rate hike in August, with the December meeting fully priced for a move. That split is not indecision — it reflects the same interpretive gap that produced the divergent bond and equity reactions on the day of the release.

The historical reference point is the 2022-to-2023 RBA cycle, in which the board cut rates, then reversed course and hiked eleven times, and the market repriced its terminal rate assumption at each step. The current position — cash rate at 4.35 per cent after an interrupted easing cycle — rhymes with that episode's mid-cycle confusion, where the growth and inflation signals were pointing in opposite directions and participants who priced the growth signal early were subsequently repriced by the inflation signal.

The condition that would confirm the equity market's current positioning is a Q2 GDP print that holds above zero — demonstrating that data centre investment has enough momentum to absorb the tightening already in place. The condition that would confirm the bond market's positioning is a Q2 print that tips negative, validating Westpac's contraction warning and removing the August hike from the probability table entirely.

The 3-year bond yield's retreat after today's data is the near-term benchmark. If it continues lower ahead of the August meeting, the bond market is accumulating conviction that the GDP miss outweighs the investment composition. If it reverses and yields push back toward pre-release levels, the market is repricing the import-inflation channel as the dominant signal.

The data centre investment thesis does not collapse at either of those outcomes — the hyperscaler spending commitments are multi-year and not sensitive to a single RBA decision. What changes is the domestic growth signal attached to it. Australia may be hosting a structural investment cycle that strengthens its digital economy while simultaneously making the RBA's task harder — and the question of whether that creates a net growth contribution or a net inflation input for 2026 will not be answered before the August meeting.