Invest1 publisher3 min readPublished
Australia's AI forecast puts A$2.2m of 2036 output behind each net new job
Jim Chalmers released the 2026 Intergenerational Report on September 20 crediting AI with A$95bn to A$116bn of GDP by 2036 and 36,000 to 44,000 jobs, against A$460m of committed government spending.
The Investor · Invest desk
What happened
- Treasurer Jim Chalmers released Australia's 2026 Intergenerational Report on September 20, a 40-year projection in which the economy more than doubles in size by the mid-2060s with AI as the primary engine.
- The report's economic models put AI's contribution to Australian GDP at between A$95 billion and A$116 billion by 2036.
- That contribution is described as a 2.6% to 3.2% uplift in national output within the next decade.
- The same models estimate that AI integration creates between 36,000 and 44,000 new jobs by 2036.
- Government spending commitments for AI-related initiatives, covering infrastructure through to skills training, already exceed A$460 million, while the data centre investment pipeline is projected at about A$150 billion by 2030.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- cost Public money covers about 0.3% of the projected buildout, so the capital behind the GDP band has to be raised by whoever ends up owning the racks and the generation to run them.
- constraint With the working-age share falling, the projected uplift can only arrive through output per existing worker, and a shortfall would show up as slower growth rather than in the unemployment figures.
- precedent Putting an AI number inside a sovereign fiscal projection gives every budget between now and 2036 a benchmark to be measured against.
- exposure Hardware suppliers, cloud providers, energy companies and construction firms are named as the pipeline's counterparties before any of that spending has been appropriated.
Take the low end of the GDP band with the high end of the jobs band and the report implies A$2.16m of annual output for each net new job; reverse the pairing and it is A$3.22m [2]. Hiring forecasts do not produce numbers like that. Whatever the models are doing to get from here to A$95bn or A$116bn, they are doing almost none of it through headcount.
The demographics in the same document point the same way. Girls born in 2065-66 are expected to live nearly 90 years and boys 86, about four years longer than current figures, while fewer children are born and a shrinking working-age population supports a growing cohort of retirees [8]. If the working-age share falls, output per worker has to rise for the GDP band to hold.
The band itself is one forecast, not two. A$95bn at a 2.6% uplift implies about A$3.65 trillion of Australian output in 2036, and A$116bn at 3.2% implies about A$3.63 trillion [1], so the low and high cases sit on the same macro base and differ only in how much of it is credited to AI.
Committed money is A$460m across infrastructure and skills [6], about 0.3% of the A$150bn data centre pipeline projected for 2030 [3]. Crypto Briefing wrote that the pipeline will require "hardware suppliers, cloud infrastructure providers, energy companies, and construction firms" [9]. The article does not say who funds the A$150bn or on what schedule. The report also projects a transition to cleaner energy sources and a halving of government debt against current forecasts [10], and the National AI Plan underneath it has three pillars: infrastructure development, workforce skills and risk management [11].
A$150bn of capex against A$95bn to A$116bn of annual uplift is 1.3 to 1.6 years of that uplift spent up front [4]. For infrastructure that is an ordinary ratio. It does mean the output line and the buildout line move together, and only one of them has an appropriation behind it.
I would follow the second one. Crypto Briefing's stated view is that the metric to track is near-term execution on infrastructure spending, skills programs and regulatory frameworks, not the 40-year GDP projection [13], and on the funding evidence here I agree. The counter-case is straightforward. The A$150bn is a projection of investment, not a commitment, so the modelled uplift may not depend on those projects at all, and the AI numbers could hold on software adoption with far less concrete poured.
What would break the productivity reading is the jobs band. Crypto Briefing wrote that the 36,000 to 44,000 net estimates "depend heavily on assumptions about how quickly displaced workers can retrain and transition to new roles" [12]. If large displacement sits underneath that net figure, A$2.16m per job is an artefact of netting.
What to watch
- Named financing and completion dates for the Australian data centre projects counted toward the A$150bn 2030 figure.
- Whether the National AI Plan's infrastructure pillar turns any of the A$460m into tenders with published values.
- Whether Treasury publishes the displacement assumptions sitting under the 36,000 to 44,000 net jobs band.