Treasury Offers Chalmers An Ai Productivity Prize — With Strings Attached

CANBERRA — Treasury has identified artificial intelligence as the first credible accelerator of global productivity growth in almost two decades, giving Treasurer Jim Chalmers a potential economic dividend to build into next month’s Intergenerational Report.

AI will “push out the global productivity frontier” and could boost Australian productivity growth to between 1.5 and 2 percent per year over the next decade, according to a policy paper, seen by POLITICO, addressed to Chalmers.

But Treasury is hedging its bets. It’s keeping productivity growth forecasts at the current expectation of 1.2 percent per year, and says AI impacts will be “uneven.”

Productivity growth rates are a major concern among federal politicians and policymakers. Without higher productivity, real wage growth and living standards stall or fall.

For the Albanese Labor government that increases the threat posed by One Nation and other advocates of populist economic policies.

Capturing a productivity dividend from AI requires more than ensuring access to leading AI models, the submission says.

Diffusion of AI is the central economic issue

Firms in sectors such as construction, health care or aged care need access to AI tools, but also skilled staff, reliable data, affordable electricity and the capacity to reorganize their work.

Chalmers said AI was “shaping up to be the biggest economic transformation in our lifetime,” which would “impact every part of our economy and society, in every corner of our country.”

“We can’t just sit around and hope the benefits of AI fall into our lap,” he said.

Chalmers and his state colleagues are given homework by the paper’s authors.

Federal and state governments will need to determine how existing laws apply to AI systems and address long-standing barriers to competition, infrastructure and business growth.

The Treasury paper is official advice to Chalmers and will underpin a substantial part of an artificial intelligence chapter in Chalmers’ upcoming intergenerational report.

The three-year gap between intergenerational reports (the legal requirement is a report every five years) indicates the urgency of both Australia’s productivity and AI challenges.

Stephen King, a productivity commissioner who specializes in business dynamism, told POLITICO that the government needs to conduct a “legal gap analysis” to ensure “all our laws and regulatory structures are fit-for-purpose for AI.” Regulators should also issue guidance to businesses on AI use and compliance, he said.

King identified algorithmic collusion as one risk: AI agents could coordinate in ways that breach Australia’s Competition and Consumer Act.

The Productivity Commission has opposed a standalone AI Act modeled on the EU’s approach, he said, because new legislation could create uncertainty while courts tested its application. Australia’s data and privacy laws also lag global best practice, limiting the country’s ability to build valuable AI applications around trusted data sets.

Two-speed economy looms 

About two-thirds of Australian businesses report using AI in some form, while fewer than one-in-10 describe their adoption as significant, per the Treasury report.

The risk of a two-speed economy stems from larger firms being likely to move faster because they have more data, greater access to skilled workers and more capacity to pay for the organizational changes required to make AI useful.

The report notes that geopolitical tensions could restrict access to capital, technology or data.

AI is also “unlike traditional technology shocks” where productivity gains were concentrated in high-trade sectors, because it may also have major impacts on how localized services are delivered, including health, aged care, education, tourism and construction. Together those sectors account for around half of Australian value-added.

Public sector capability is also seen as a defining factor in whatever productivity is squeezed out of AI. That capability is starting to emerge: Federal public servants now have access to a central AI portal and Microsoft Copilot. Departments have chief AI officers and “AI champions” to promote experimentation.

Data center boom, but for who?

Data centers are on track to be the defining infrastructure and energy question of the next decade.

Treasury says the pipeline of AI-related investment could reach $150 billion by 2030, equivalent to about 5 percent of GDP.

Australia had 162 operational data centers by March 31, with about 130 more proposed.

The U.S. has at least 4,700 data centers — nearly triple the rate of data centers when taking account of population difference.

The government has made clear its interest in attracting data center and AI training investments, including because of long-term economic benefits. The Treasury notes that the immediate economic impact could mostly flow overseas: about two-thirds of data center supplies such as servers, storage systems, networking equipment and other technology would be imported, would flow overseas.

While data center investment is spreading beyond NSW and Victoria into Queensland, Western Australia and South Australia, it is competing with public infrastructure projects for labor, concrete, copper and other construction inputs.

The Australian Energy Market Operator expects data centers to account for about 7 percent of National Electricity Market demand by 2029-30, up from about 2 percent today.

The problem for the Treasury’s productivity equation is that those financial risks and energy crunches could arrive before AI’s productivity gains.

Labor market disruption is real but hard to measure

AI is expected to augment many roles, but distribution of the gains will depend on who has access to training and who can adapt as firms reorganize.

Treasury’s advice leaves Chalmers with a substantial policy challenge.

For AI to deliver broad benefits and a substantial productivity boost, AI policy will have to expand beyond regulating chatbots or attracting data centers. It will need to help firms adopt the tools and help them move their workers into higher-value work.