CHECK AGAINST DELIVERY

The macroeconomics of AI: Building an Australian AI stack

OPENING AND INTRODUCTION

I would like to acknowledge the Gadigal of the Eora Nation, the traditional custodians of this land, and pay my respects to the Elders both past and present.

I would also like to acknowledge the many esteemed people here today.

In the mid-18th century, the steam engine reorganised production.

Come the early 20th century, electrification transformed the factory and the home.

And in the late 20th century, the internet upended communication and commerce.

Today, artificial intelligence is the next great general-purpose technology.

We are still discovering its possibilities, but its direction is clear: AI converts data, water and energy into intelligence—and with it, the capacity to solve problems at scale.

The question before us is not whether AI will change the economy—it will.

The question is how we maximise the benefits to Australia and the Australian people.

For a Labor government committed to fairness, shared prosperity, democratic accountability, dignity of work and environmental sustainability, AI is not an abstract debate.

Its potential distributional consequences for working people mean it must be a values-driven project—and, crucially, a practical one executed at speed.

Today, I want to explore four economic themes as they might relate to the future of AI –

Productivity; labour markets; capital concentration, and finally opportunities for Australia.

I. AI as a General-Purpose Technology: Productivity and Growth

Economists describe certain innovations as general-purpose technologies.

They are not single-use tools; but rather platforms that enable waves of innovation and new forms of production across the economy.

At its core, AI lowers the cost of prediction, pattern recognition, and certain forms of decision-making. In more precise economic terms, it reduces the marginal cost of producing information-intensive services, including synthesis and generation.

That matters because modern economies, particularly the service-oriented economies of western democracies, are saturated with information-dense tasks.

When we augment people’s skills and talents with AI, we open the door to faster, better and more inclusive growth.

But there are two cautions here based on what we have learnt over the past 200 years.

First, productivity gains are not automatic.

Historically, it has taken decades for general-purpose technologies to translate into measurable growth. The so-called productivity paradox of the digital age reminds us that complementary investments—in skills, organisational change, infrastructure—are essential.

Second, even when productivity rises, the distribution of growth is uncertain.

Fairness, as I have argued in recent weeks, will not be predetermined or self-executing. It is a choice. Something that has to be designed, not discovered.

II. Labour Markets and the Future of Work

One of the most important macroeconomic concerns around AI is its impact on employment. How will AI impact jobs?

The forecasts for AI could not be further apart.

At one end is sweeping optimism: productivity surges, incomes climb, and people are freed from drudgery to focus on creativity, care and higher pursuits.

At the other is deep pessimism: wealth concentrates, inequality widens, jobs are displaced faster than they are created, and democratic institutions strain under the weight of technological power.

Between those poles lies the real task of policy.

To analyse these scenarios, economists distinguish between tasks and jobs.

A job is a bundle of tasks. AI does not typically eliminate entire occupations overnight; it automates or augments specific tasks within them.

This tends to produce three types of economic effects including substitution, complementarity or creation of new tasks.

Substitution reduces demand for workers, complementarity increases productivity and demand for workers who use AI effectively, and creation enables new tasks and new even potentially new industries that did not previously exist.

If substitution dominates and creation of new tasks is slow to emerge, we will undoubtedly experience disruption, particularly for routine cognitive work.

If complementarity and creation are strong, we may instead see higher productivity and rising wages for many.

As many distinguished economists have pointed out, the effects will be uneven across the economy and depend on the type of work being performed.

But to understand the distributional effects and macroeconomic outcomes for Australia you cannot focus on the substitution or augmentation effects alone.

There is a deeper lever we must keep in view which I will focus on today: who owns the technology. Ownership and access—more than in any previous wave—will heavily shape how gains are captured and with whom they will be shared. That is why Australia’s approach must be pro-adoption and pro-creation.

III. Capital, Concentration, and the Distribution of Growth

The second point I want to make is that market structure and ownership matters.

AI is capital-intensive. It runs on vast datasets, compute infrastructure and scarce technical talent. That means high fixed costs, powerful economies of scale and strong network effects.

Left unchecked, that creates a flywheel: more data, better models, more users, more data again. Markets concentrate. Rents accrue to capital. Labour’s share shrinks. Smaller firms are squeezed. And in a global digital economy, value flows to the nations that build the systems, not those that merely use them.

The question is not whether AI lifts productivity. It will. The question is who captures the gains.

If Australians own the tools, we share in the upside, higher incomes, stronger national wealth, broader opportunity.

If the tools are owned elsewhere, and we simply rent them, the dividends flow offshore. Productivity may rise, but national income does not necessarily follow. We will gain some benefits from adoption, but relative to other countries we may fall behind.

These two scenarios are very similar in terms of the implementation of AI, but very different in terms of its distributional consequences.

This simple illustration shows that AI can have very different impacts on people and countries depending on how the technology is implemented and who owns it.

That is the fork in the road. AI can broaden prosperity, or concentrate it. The difference lies in ownership, market structure and policy.

The conclusion for Australia is clear: we need AI that augments our labour force and is owned by Australians.

IV. Australia’s Opportunities in the AI Stack

But Australia must be smart and realistic in terms of where we play in the various industries and business models that will make up the AI economy.

The AI supply chain is often conceptualised as a “stack” built in layers, each with distinct technical functions, economic characteristics, and trade implications.

The key layers include energy, semiconductor chips, data centre infrastructure, foundation GPT models, and software and services.

Global AI opportunities are rapidly accelerating. Worldwide spending on AI is forecast by Gartner to grow from $1.76 trillion in US dollars today to $3.34 trillion in 2027. On this trajectory the global AI economy could be worth around $5 trillion USD by 2030, which is over $7 trillion in Australian dollars.

Between now and 2027 the forecast growth across different layers of the stack include:

AI infrastructure (chips and data centres) – $965 billion to $1.7 trillion

AI models - $14 billion to $43 billion

AI software - $283 billion to $636 billion

AI services - $439 billion to $761 billion

AI cybersecurity - $26 billion to $86 billion

Source: Gartner, Worldwide AI Spending will Total $2.5 Trillion in 2026, January 2026^[1]

This staggering growth presents a once in a generation opportunity for Australia to compete to generate new jobs and prosperity.

With ready access to 4.8 billion people in the Asia-Pacific^[2], and a range of other advantages with energy, connectivity and stability, we can be a launching pad in the region.

Now let’s say as a thought experiment Australia lifts its share of the global AI economy by 0.5% heading into 2030. What might that mean for our prosperity?

This would imply an additional $25 billion per annum relative to our baseline performance.

This means hundreds of thousands of new high-value jobs, productivity spillovers, higher taxation bases, and a greater depth of talent that will create more prosperity for Australia.

To give Australia the best chance of success, we need public policy and private investment working together — directed toward the parts of the AI ecosystem where our strengths are real and our returns are likely to be highest.

Across the AI stack, some layers demand extraordinary capital and highly specialised engineering—for example, manufacturing advanced semiconductor chips or building frontier scale foundation models.

These are fields where only a handful of nations and firms have the scale to compete.

But the picture looks very different when we shift our focus to AI applications, services and software. This is where Australia can genuinely compete on the world stage. Our entrepreneurs, universities and businesses are already producing globally recognised technology. This is the part of the stack where Australia can build, export and lead.

The case for Australia to pursue the next generation of AI software—and to develop our own platforms—is straightforward.

Digital platforms thrive on network effects: each new user makes the service more valuable for the next. This creates a powerful flywheel of adoption, data and growth. We see this in companies like Uber and Amazon.

But AI is even more consequential. Unlike ridesharing or online retail, AI is a general-purpose technology that will reshape almost every part of the economy. The software platforms built on top of it will become enduring engines of economic value.

And perhaps the most important rule of digital platforms is this: it’s not always the best service that wins—it’s often the first service that reaches scale with a good enough product. That means speed is not just an advantage; it is decisive.

Australia has the talent, the research base and the commercial capability. The fact that we are early adopters will also be an advantage. What we need now is urgency.

V. The Albanese Government’s Whole of Government Approach to Prosperity in AI age

The Albanese government is bringing a whole of government approach to prosperity in the AI age.

That approach is based on the recognition that we need to urgently build an AI ecosystem that works for Australians.

We cannot sit back and lease someone else’s brainpower. We must build it here, back Australian founders, equip Australian workers, and ensure that AI is something we design, deploy and lead, not something we merely import and consume.

More lazy Liberal laissez-faire will mean the continued Uberisation of our economy - a future where we rent the platforms, import the intelligence and export the profits.

That is why we are delivering a bold Labor plan to make AI work for Australians.

We are using the levers of government to make sure we are makers not just takers, creators not just consumers; builders not just bystanders.

Our plan to do this stretches across government.

First, at the base of the AI stack, we are working to generate the foundational input of Artificial Intelligence: electricity.

Australia receives 58 million petajoules (PJ) of sunlight every year^[3]. We have the highest solar radiation per square metre of any country on earth - an enormous foundational competitive advantage in the global AI race.

Our government is determined to capitalise on that advantage. We are using the levers of government to create the renewable energy we will need and the transmission to get it where we need it. And we are using data centres to underwrite new renewable projects and transmission across the country. The Liberals missed this opportunity. In their decade of government they generated negligible new solar capacity. The Albanese government under Chris Bowen has changed gear. Australia is now turning sunlight into electrons by the gigawatt.

The second part of our plan is to turn electrons into intelligence, the raw output of artificial intelligence. The economy has seen massive investment in data centres and digital infrastructure. In 2024 Australia was the second highest recipient of data centre investment in the world^[4].

Australian companies, NextDC, Firmus, Airtrunk, Sharon AI, Vault, CDC and others are leading the way on supporting an industry that is substantially Australian built and Australian owned - truly a digital Future Made in Australia. Through our ongoing work on data centre principles we will ensure these deployments work for people and communities, and support our national interests.

The third part of our plan is to turn intelligence into productivity. We are using the levers of government to support Australian AI models and applications through investments in science, universities, backing Aussie companies through the NRF, R&D tax incentive, and new AI Accelerator funding. We have more than 1,500 Australian AI companies. The National Artificial Intelligence Centre is supporting Australian businesses to adopt AI. And many of our biggest businesses are leading the way in AI innovation.

And while much is being done we also know there is much more to do. For example, Australia contributes 1.88 percent of global AI publications, yet accounts for only 0.18 percent of global AI patents^[5], highlighting commercialisation challenges that we must continue to confront and overcome.

Fourth, we’re helping to turn that productivity into higher profits and wages. Because we are a Labor Government we’re ensuring that all Australians participate in this transformation. Through the work of Jason Clare and Andrew Giles and Jobs and Skills Australia we’ve been charting a course on AI education and skilling.

Michelle Rowland has stood up for Australian artists, journalists and publishers. We want to grow the value of Aussie creatives, not give it away. Amanda Rishworth is bringing businesses and workers together to collaboratively work through workplace AI implementation. We have shown that AI driven productivity can and will be a win-win for Australian business and workers. And Dr Andrew Leigh is leading work on competition policy.

Fifth and finally, we’re determined to use AI to improve the levers of government themselves. Katy Gallagher is making groundbreaking investments in AI innovation the public service to ensure all public servants have access to AI tools and the capability to use them to deliver better and more efficient services and policy work. We can offer our citizens better and more accessible health care through targeted investments directed at those who need it most. We can provide education at scale throughout our citizens lives … breaking down barriers of location age and circumstance.

So that’s the plan. We will turn sunlight into electrons, electrons into intelligence, intelligence into productivity, productivity into wages and profits, and all this into higher living standards for working people.

We can build a prosperous future for our people in the digital economy. A future filled with shared opportunity.

CONCLUSION

Let me close with this thought.

AI is a tool. It can increase productivity, expand knowledge, and solve complex problems. It can also concentrate wealth, displace workers, and entrench inequality.

The macroeconomic outcomes are not preordained.

If we commit to a vision of shared benefits, and match that with ingenuity and enterprise—we can build an Australian economy that has more jobs and prosperity to share.

We stand at an important juncture.

Down one path lies the destiny of a skilful technology taker – where Australia’s growth depends on its ability to adopt AI and drive productivity across the economy.

Down the other path lies the destiny of a world class adopter and creator and exporter of AI technology. Our economy will be larger and more productive. There will be new and higher paying jobs. There will be more capacity to reinvest in society and ensure opportunity, access and equity.

Macroeconomics gives us the tools to understand the stakes. Labor values provide us the compass to choose wisely to ensure the benefits are shared broadly.

What is clear is when it comes to seizing the opportunities, we cannot wait too long. The flywheel is starting to turn, and the next five years will be critical for Australia.

The central task of Labor is to ensure that technology works for the Australian people, and not the other way around.

END NOTES

[1] Gartner values in USD, www.gartner.com/en/newsroom/press-releases/2026-1-15-gartner-says-worldwide-ai-spending-will-total-2-point-5-trillion-dollars-in-2026 
[2] Updated, based on United Nations Asia Pacific Population and Development Report data sheet, 2025, https://repository.unescap.org/server/api/core/bitstreams/aadca0c1-9249-447e-8bfb-241187ee64c0/content  
[3] Geoscience Australia, 2015. 
[4] Knight Frank, Global Data Centres Report, 2025
[5] National AI Centre, AI Ecosystem Report, 2025.