Thank you to the University of Technology Sydney and UTS Startups for hosting us. 

I begin by acknowledging the Traditional Custodians of the land on which we gather today, the Gadigal people of the Eora Nation, and pay my respects to their Elders past and present. I extend that respect to all Aboriginal and Torres Strait Islander peoples with us today. 

There's no better place for this conversation. 

Murray Hurps and his team have built UTS Startups into the largest program of its kind in Australia – over a thousand student-founded start-ups, creating more than twelve hundred jobs a year. 

What Murray understood early is that the best way to prepare Australians for the future economy is not just to train people to use the technology that exists – it's to back the entrepreneurs and innovators who want to create the technology that doesn't exist yet. 

That instinct – that Australia should be a country that makes things, not just uses things made elsewhere – is the animating idea of everything I want to say today.

SEIZING A FUTURE MADE IN AUSTRALIA

A core value for the Albanese Government is to ensure that new technology works for Australians, and not the other way around.

This is fundamental to two key pillars of the National AI Plan that Minister Ayres and I outlined in December of last year: to capture the opportunities of AI and ensure the benefits are shared. 

Today, I want to make a simple argument – and then to ask all Australia to come together to do something about it.

The argument is this: Australia is in a critical window in which we will determine whether this country builds a strong domestic AI industry, or whether we become an importer of foreign AI and a permanent renter of intelligence from abroad.

This distinction matters. In the last few months we have seen how dangerous it is to be reliant on foreign imports of a commodity that is critical to our economy. 

You've heard the phrase 'data is the new oil'. This is truer than most people realise – but instead of producing energy, AI will harness data to produce intelligence and services which can be used domestically and exported to the world. 

If we fail to build a strong AI industry around our own data and ingenuity, we will be a more vulnerable, and less prosperous country.

To explain why I believe that we need to build and own our own AI companies, not just adopt foreign AI, let me start with a thought experiment borrowed from the Australian economist Justin Wolfers.

Imagine that it's Christmas day and you walk outside to find a package on your doorstep. Inside the package is a Christmas present for you which is an AI – a piece of software or a robot – that can do your job. When the holidays are over in January, you decide that instead of going back to work full time, you'll send the robot instead. You discover that all the work gets done. The boss is happy. You get paid but you are free to do other things – take on new work, pursue other interests, spend time with your family. In that scenario, AI has made you better off. And it's made the Australian economy better off. This is a scenario where technology makes everyone win.

Now tweak the scenario. Instead of the worker owning the AI, imagine that you come back from the Christmas break to discover that your employer has built a robot over the holidays that can do your job. Now the employer owns the robot and tells you that your job will change or even disappear. You may need to retrain or find another job – and we want employers to take responsibility here. Unlike scenario 1, positive outcomes from the new technology are not automatic.

This matters to the Labor movement and it’s where employers, unions and government must work together to shape the distributional impacts, create opportunities and translate increased productivity into a higher standard of living for all Australians.

But to understand the stakes let me tweak this one more time.  Let's assume the AI isn't built by the Australian employer. Instead it's owned by a foreign company which, because of its scale, has a monopoly over the technology. Because it has a monopoly, it can charge whatever it likes – so it charges the Australian company just a fraction less than your wages. Now you as the worker are worse off. You don't have a job. The Australian company is barely better off. And the Australian taxpayer is much worse off – because the value has flowed offshore.

Same technology. Three completely different outcomes. The difference is not AI itself – it's who owns it, who controls it, and whether the benefits are shared in Australia or concentrated abroad.

The question of Australian AI ownership is not abstract. It is arguably one of the most consequential economic questions of this decade – and why we are working so hard to make sure Australia has its own AI companies.

THE CRITICAL WINDOW TO LEAN AUSTRALIAN

You all know the movie Sliding Doors. Gwyneth Paltrow catches a train, or she doesn't. Two completely different lives unfold from one small moment. Australia is standing at its own sliding doors moment on artificial intelligence.

In one future, we back Australian AI capability. We build an industry. We keep our data, our expertise, and the economic value of this technology onshore. We become a country that exports AI solutions to the world.

In the other future, we adopt enthusiastically. We buy foreign. We rent our intelligence from overseas. And we spend the next generation watching the wealth flow to someone else's country.

The doors are open right now – but they will not stay open forever.

The second point I want to make is that this is a national effort. 

We all need to lean into buying Australian AI technology and services. The government has an important role to play, but so do Australian businesses.

Every week, leaders and managers make purchasing decisions about AI. Business executives sit through pitch meetings. Secretaries and Chief Financial Officers of public agencies review procurement briefs. Chief information and technology officers weigh up vendor proposals. 

In the Commonwealth Government alone, technology accounts for roughly a fifth of procurement.

The question on every desk is: How do we bring AI into our operations? 

But there is a second question not enough people are asking: where is the AI I'm buying coming from?

In this noisy global marketplace, the question of whether there is an Australian option –  and whether it deserves serious consideration – is too often being drowned out.

Let me make this concrete with what I call a tale of two procurements. Both involve AI in healthcare. They led to radically different futures.

The first was in breast cancer, where AI offers the potential for dramatic diagnostic and cost improvements. BreastScreen NSW had to choose an AI partner. They had to make an important and consequential procurement decision about which technology they would go with.

In one corner was an Australian option – the Breast Cancer Screening with Artificial Intelligence program (BRAIx) led by Associate Professor Helen Frazer. Australia has one of the highest-quality datasets in the world. Our national screening program has been running for over thirty years and was digitised more than a decade ago. The taxpayer has invested over three billion dollars. The data it has generated is a national asset of extraordinary value. Frazer's team has developed an AI mammography reader that detects cancer better than ninety per cent of specialist radiologists. This is Australian capability, built on Australian data, with the potential to save thousands of lives – and to be exported globally.

BreastScreen NSW didn't choose BRAIx or any other Australian capability. They went with Lunit – a Korean-listed company whose algorithm was trained primarily on women from Korea, UK and the USA, with no peer-reviewed results in an Australian setting, no randomised controlled trial, and limited transparency. I have no doubt the Korean option had some compelling features – but when you outsource that capability to a foreign vendor, you don't just buy a product. You create a path dependency. You hollow out local expertise. You hand over globally unique data with no value captured for the nation. And you foreclose the possibility of an Australian export.

I'm not suggesting BreastScreen NSW decided in bad faith, or even for the wrong reasons. But decisions like this need a wider lens – one that considers not just which product is most mature today, but which path creates the most value for Australia over the next decade. This is a sliding doors moment. One door leads to sovereign capability; the other to permanent dependency.

Let me contrast this procurement in breast cancer screening with another procurement in radiology which had a very different outcome. When it was considering its AI options, the I-MED Radiology Network – one of the world's largest private radiology companies – took a chance on Harrison.ai, an Australian company founded in Sydney in 2018. Today, Harrison.ai is deployed in over a thousand healthcare facilities across more than forty countries. Half of all Australian radiologists use its tools. It is live in over forty NHS Trusts in the UK and powers every public emergency department CT scan in Hong Kong. The company has raised over three hundred million dollars, and has been backed with a $32 million equity investment from the National Reconstruction Fund. 

Harrison.ai is what happens when an Australian company is given a chance. Australia could have hundreds more – if our local capability is backed with the same conviction.

Now let me be clear, my intention is not to criticise the procurement decision of BreastScreen NSW or any other organisation. I am sure they made their decision in the best interests of their organisation, managing risk and cost and time and many other factors.

My argument is not that Australian companies should always be buying Australian AI. In some cases, Australia won't have a competitive product.

So what am I asking for? I'm asking Australian businesses: when you have a choice between a foreign AI product and an Australian one, and the choice is close – and often it is close – lean Australian. 

The compounding benefits of that choice, repeated across thousands of organisations, will determine which side of the sliding doors we end up on.

To procurement decision-makers specifically: I am asking you to actively seek out Australian AI options before you sign. Put them on the shortlist. Give them a fair hearing. If the foreign product is clearly superior, fine. But don't let the Australian option lose by default – because it wasn't in the room. 

Plenty of Australian AI companies are competing on the global stage. An example in customer support AI is Lorikeet – a Sydney company now ranked eighth globally for start-up AI spend, ahead of Perplexity, ahead of Canva, competing head-to-head with Silicon Valley’s Intercom and Sierra. 

In AI-powered logistics and fulfilment, there is Skutopia. For building AI agents and workflows, there is Relevance AI. 

For hospitals or GP clinic choosing an AI scribe, look at Heidi Health – a Melbourne company now processing a million consultations a week across five countries. 

These are examples of real Australian companies, building world-class products, right now. They just need to be in the room when the decision is made.

Australia has more than 1,500 AI focused companies across sectors. We may not build the foundation models – that requires hundreds of billions we don't currently command. 

But that is only one part of the story. Competitive advantage in AI increasingly sits higher up the stack. Between the base model and the end product are the layers that turn general‑purpose models into applied and domain specific AI solutions that are reliable, deployable and trusted.

Across Medical imaging, mining automation, agricultural AI, climate modelling, financial services intelligence – we have researchers, firms, and datasets that are genuinely world-leading and can compete at a world class level

AIML at the University of Adelaide is one of the most respected machine learning centres in the world. CSIRO's Data61 has been at the frontier for years.

The good news is that there are many Australian corporates willing to back local capability. In 2024, Commonwealth Bank entered a five-year partnership with AIML to establish the CommBank Centre for Foundational AI Research at Lot Fourteen in Adelaide. CBA's Chief Information Officer said the partnership paid for itself within the first three weeks. 

That is the model we need: investment driven not by sentiment but by strategic advantage – proximity, co-creation, continuity of service, the ability to shape the product, and the knowledge that sensitive data stays under Australian jurisdiction and Australian law.

We need more Australian corporates and government organisations giving Australian AI companies the consideration and look-in they deserve and recognising the long-term benefits to them of having a local partner and a strong local AI industry.

When procurement teams default to the foreign option, a path dependency is created that lasts a decade or more. That instinct is rational. But at scale it is ruinous – because when every organisation follows that logic simultaneously, we hollow out our own capability and make it impossible for Australian firms to reach the scale they need.

And if you need evidence of what happens when a country fails to build its own technology industry during a wave of transformation, we don't need to look overseas. We just need to look at our own recent history.

We've done this before. The 2010s were the decade of the software platform. Australia became the most Uber-dense market on earth. We embraced Google and Facebook faster than nearly any comparable economy.

The political scientist Jeffrey Ding argues persuasively in Technology and the Rise of Great Powers that what determines a nation's trajectory is not who invents a general-purpose technology but who diffuses it most effectively. Adoption matters. Global companies in Australia bring capability, develop talent, generate spin-outs. But Ding's argument cuts both ways. A country that only adopts – without building the firms and skills to shape the technology – will always be working to someone else's playbook.

In the 2010s, we were overwhelmingly users, not builders. The ACCC documented that seventy-one cents of every digital advertising dollar in this country goes to Google or Facebook – a fifteen-billion-dollar market, with the vast bulk flowing offshore. Uber commands nearly eighty per cent of our rideshare market and generated nearly four billion in Australian revenue in 2024. There were exceptions – Canva, Atlassian, Afterpay captured enormous value and anchored real ecosystems. But the rule was clear: Australia was a net importer of technology and a net exporter of value.

I call this the Uberisation of the Australian economy. A new wave of technology arrives, we adopt it eagerly, and because we have too few local champions, the economic benefit accrues to someone else's country. We get the convenience. They get the wealth.

This didn't happen by accident. It happened because of a decade of policy neglect. The previous government's philosophy was simple: it doesn't matter where the technology comes from, as long as Australians can use it. Don't pick winners. Let the market sort it out. Trust the invisible hand. What that philosophy delivered was an economy increasingly dependent on foreign platforms for critical digital infrastructure – with no serious strategy to develop Australian capability and alternatives. 

It clearly does matter where the technology comes from. It matters for jobs, for tax revenue, for sovereign capability, for the long-term structure of our economy. And developments in both Europe and Canada show there is a global awakening on this front.

It also matters for workers.

Labor Governments also understand that Australian workers and unions will also benefit from more sovereign AI. Australian businesses create more Australian jobs. And if the systems transforming Australian workplaces are all built offshore, workers will have no seat at the table when those systems are designed. If they are built here, we will have a more influential say.

And I want to acknowledge the work of Minister Rishworth and the Artificial Intelligence and Workplace Employment forum which the Minister convened earlier this week in Adelaide, bringing together unions, business and government to progress discussions on AI workplace focus areas of trust, capability, safety, transparency and productivity.

The wave of digital platforms was one sliding doors moment and we did not rise to the occasion. We cannot afford to do that again.

Now the new wave is here. AI is the substrate on which every other technology will be built. It is a general-purpose technology that will permeate the entire economy – not just advertising or rideshare.

The Australian Government is not standing still. 

The National AI Plan is the most comprehensive roadmap this country has ever had for building an AI-enabled economy. We have consolidated more than $460 million in AI-related funding, launched the AI Accelerator initiative, and backed companies through the National Reconstruction Fund and the Industry Growth Program.

That includes a $32 million equity stake through the National Reconstruction Fund in Harrison.ai, as well $20 million investment in Sydney‑based Silicon Quantum Computing, a world‑leading company operating at the frontier of AI‑enabling semiconductor manufacturing. 

We are building the skills and infrastructure and incentives to make it easier for Australian AI to compete. 

Through our Buy Australian Plan, we have already made meaningful changes to the Commonwealth Procurement Rules to back Australian businesses which include:

• Requiring that government agencies approach Australian businesses first, before all others, for procurements that fall below our Free Trade Agreement thresholds of $125,000.
• Raising the threshold for streamlined procurement contracts from an Australian or New Zealand small or medium business to $500,000
• An explicit requirement that economic benefit to the Australian economy is considered as part of the value for money assessment for all procurements valued at $1 million or more. 

And we have released the Australian Government’s expectations for data centres and AI infrastructure developers - the first articulation of Australian values and our national interests as it applies to AI infrastructure, particularly energy and water security.

CLOSING

In closing, we will continue to pull every lever available to us to ensure that government settings actively support Australian AI capability.

This is not about directing individual procurement outcomes. 

It is about making sure Australian capability is visible, competitive and capable of getting a fair go.  

But Government policy settings alone will not get us there – 

Government can set the conditions, send signals, and de-risk early-stage companies. What it cannot do is replace the procurement decisions of thousands of private-sector organisations. That is where the real leverage lies.

Some will say this is all protectionism dressed up as patriotism. It is not. I am not asking anyone to buy an inferior product. 

What I am enlivening is the same core principle I articulated at the beginning of this speech – that technology must work for Australians, and not the other way around.

And in that context I am asking people to recognise a dynamic calculation: the Australian product, given support, will improve faster, integrate more deeply, keep your data under your jurisdiction, and generate jobs and intellectual property that stay in this country. 

Artificial intelligence is the most consequential technology of our generation. The countries that build AI will capture its value. The countries that merely use AI risk being its subjects.

Australia is standing at the sliding doors. We have the researchers, the data, the talent, and companies proving every day that Australian AI can compete with the best in the world.

What we need is for Australian organisations – public and private – to give them a chance. Not forever. Not unconditionally. Just a chance.

Because the door we walk through now – this year, next year – will determine whether we spend the next generation building intelligence or being dependent on renting it. 

Our Government intends to make sure we walk through the right one.

Thank you.