Price Taker or Price Maker: Australia's AI Sovereignty Challenge

The 2026 Tom Hughes KC Oration, Shangri-La Sydney, 15 June 2026

Introductory Remarks

Thank you for the warm welcome.

I’m honoured to be here tonight at the invitation of my good friend Julian Leeser.

Julian is a warm spirited man, intellectually curious, and firm in his convictions.

He has a big heart for young Australians, and so I’m really pleased to see him at work in education policy.

We are living through massive change.

Tonight I will speak about one of the main drivers of change: the rise of Artificial Intelligence, or A.I.

Because of the challenge posed by AI, now – more than ever – do we need the Australian education system preparing young hearts and minds for what is to come.

I’m invested in our education system both as a father of three children—aged ten, eight and four—and as a policymaker.

Like any dad, I want the best for my kids. As a policymaker, I want the best for Aussie kids.

And when I look closely at the Australian education system, there is a lot to be concerned about.

Across Australia, the numeracy and literacy levels of our children are sliding.

Our disadvantaged kids are the ones most impacted here. They are the ones who fall behind and struggle to catch up.

This general trend is reflected in international benchmarks where we see a steady decline in the performance of Aussie kids over the last twenty-five years.

We can’t ignore this educational slide with Aussie kids, as it will be accelerated by the widespread adoption of AI.

And if we fail in education, AI will entrench economic and social disadvantage here in Australia, and we will fall behind as a country on the world stage.

Education is a key battleground, and I am confident that we have the right man in Julian driving reform in this area.

Julian, thank you for serving our country at this time of great upheaval. This is not an easy task in education, but you’re the man for the moment.

The Tom Hughes KC Oration

I’m doubly honoured tonight to give the 2026 Tom Hughes Oration.

I never met Tom Hughes myself, although we briefly crossed paths back in the early nineties. That would put me as a young boy, so let me explain.

I grew up in Ashfield—not far from here—and my dad, Peter, sent me out on work experience during the school holidays. He was the Minister of Ashfield Presbyterian Church, so he knew a diverse group of people who worked across the economy.

In 1992, I spent a week in a small van distributing confectionary to corner shops across Western Sydney under the charge of a young salesman called Mike. I remember it well because I was paid for my week’s work with a box of strawberry flavoured Hubba Bubba bubble gum.

In 1993, my father sent me to Wardell Chambers with a barrister by the name of Bill Haffenden. Bill was an adult convert to Christianity, and my dad got to know him through Sydney Anglican church circles.

My dad let Bill preach in the pulpit from time-to-time, where he’d deliver sermons to the flock at Ashfield.

After the service, Bill would greet parishioners in a snappy suit, sometimes cigarette in hand, as they filed out onto Liverpool Rd.

That’s how I came to spend a week with Bill at Wardell Chambers—trailing him to court—wondering how I ended up spending my school holidays with people in wigs and gowns.

At that time, I was a Rugby League fanatic. I loved it. In fact, I was obsessed with it.

I’d scrounge coins together before school, and buy the Daily Telegraph at the Ashfield train station to keep up to date on the weekend rugby league scores.

And that’s why I had a vague understanding that Cronulla Sharks star, Andrew Ettinghausen, was upset about a nude photo taken of him.

Ettinghausen had been photographed naked, without his permission, on a Kangaroos tour of the United Kingdom in 1990.

He’d sued, and won.

During my week at Wardell Chambers, Bill told me that a man called Tom Hughes QC had represented Andrew Ettinghausen in the defamation trial earlier in the year.

That Tom Hughes QC had an office in the building!

Tom had won the case, deploying his wit and dry humour in one of the most memorable pieces of cross-examination in Australian legal history.

As a ten-year-old rugby league fanatic, that was music to my ears.

A friend of Andrew Ettinghausen was a friend of mine.

By then, I’d sworn off ever taking the law, but I banked the name of Tom Hughes QC as one to remember.

A great Australian.

And, indeed, he was.

Tom Hughes served Australia at war in the Royal Australian Air Force, at peace in the Commonwealth Parliament, and many of his fellow Australians through the legal system.

I acknowledge Mrs Christine Hughes here tonight, and other members of the wider Hughes family, and thank them for the opportunity to speak at this forum.

The Two AI Races

Artificial intelligence is reshaping the world, and we can’t hide from it.

We have a choice: shape AI, or be shaped by it.

So, here’s my big idea for the evening:

Last century, Australia missed the opportunity to become a nuclear power.

As a result, we live under the nuclear umbrella of the United States.

Our sovereignty and strategic independence have been constrained as a consequence.

In the age of nuclear weapons, Australia is a price taker, not a price maker.

This century, Australia risks missing the opportunity to become an AI power.

And the risk is that our sovereignty and strategic independence will be further constrained by the AI superpowers reshaping the global order.

That will put us in a position of weakness, with less agency over our future.

In the age of AI, will Australia will be a price taker, or a price maker?

We have a choice before us.

I think that Australia should invest deeply in AI to preserve our strategic independence.

This is a decision we need to take now. Because we are running out of time.

Historian Niall Ferguson wrote earlier this month in The Free Press that the unfolding competition in AI has reached a peak moment of danger.

That the AI race is as dangerous as the nuclear arms race that defined the Cold War.

Ferguson helpfully explains that there are in fact two AI races underway.

The first AI race is happening within the United States between five American companies.

You’ll know some of the names: Anthropic, Google, Meta, OpenAI and xAI.

They are the main competitors, and their global power and reach is growing faster than we realise.

Nation-states are dwarfed by their research and development budgets.

In the 2025 fiscal year, the U.S. government invested roughly $3.3 billion in non-defence AI research and development.

But the year prior—2024—the private sector invested more than $109 billion in AI.

In case you missed it, the private sector invested thirty-three times more than the U.S. government did into AI research and development.

This first arms race between the American AI companies is massive.

It’s consuming capital, energy, data and human talent.

And this race is largely unregulated. And will stay that way—at least in the short run.

The Trump White House made that clear through the National AI Legislative Framework, released this past March.

It’s a four-page document, and two points stand out.

First, the White House delegated policy responsibility for AI to the U.S. Congress. But on this condition: that Congress ‘should not create any new federal rulemaking body to regulate AI.’

It appears that the TechBros in Silicon Valley have won this fight over AI regulation, and exert strong influence within the Trump White House.

Second, the White House states that United States’ national strategy is to achieve global AI dominance.

We see that the AI race is not confined to the borders of the United States.

But there is in fact a second AI race—it’s global—and it’s between the United States and the Peoples’ Republic of China.

This second AI race is more intense, more consequential, and with geopolitical implications for us all.

It gravitates around Taiwan, the source of ninety percent of the world’s advanced semi-conductor chips.

Taiwan controls the vital AI infrastructure essential to the ambitions of the U.S. and China.

One might say the keys to victory in the AI race.

Hence this moment of danger.

There is an increased risk of strategic miscalculation, as both the U.S. and China seek to achieve dominance over the other.

I won’t cover all the possible contingencies around war in the Taiwan Strait, except to say that Australia will be involved whether we like it or not.

We’ve seen what upstream supply chain disruption looks like with the closing of the Strait of Hormuz.

I don’t think it’s an exaggeration to say that a hot war between the U.S. and China over AI dominance and advanced chips in Taiwan would be infinitely worse than a hot war in the Middle East.

We won’t be able to escape it, which is why we must work hard to prevent it.

So, what are our strategic options here?

The Strategic Challenges posed by AI

Let’s first focus on the strategic challenges that we face with AI.

The first is that Australia finds itself caught between our closest security partner in the United States, and our biggest trading partner in China.

Both partners are AI superpowers pursuing global AI dominance.

Australia is caught in the middle with limited control over the structural geopolitical competition.

And strategic miscalculation over Taiwan will have devastating impacts on Australia.

The second is that AI development is mostly unregulated.

Australians are adopting AI at speed, but we have little control over its development and the rules governing it.

Anthropic’s new frontier model, Claude Mythos, is a case in point.

Anthropic delayed the release of Claude Mythos because of concerns that it might be used by bad actors and criminals to exploit software vulnerabilities.

Jared Kaplan, the Chief Science Officer at Anthropic, said that Claude Mythos wasn’t dangerous because it was trained to focus on cybersecurity capabilities.

Rather, Kaplan said that Claude Mythos was dangerous because of its

…general intelligence and its general ability with software, it’s particularly good at identifying vulnerabilities in software and how to exploit them.

So dangerous, in fact, that Anthropic delayed the release of Claude Mythos to the public.

Instead, Anthropic initiated PROJECT GLASSWING, and brought together forty leading American corporations to study the model’s dangerous capabilities and develop defensive countermeasures.

The mission, according to Anthropic’s Jared Kaplan, was to ‘…make sure that the defenders get ahead.’

Thankfully, the leadership of Anthropic seem to appreciate the inherent risks and dangers posed by AI.

At least so far as they concern the U.S. national interest.

But even this precaution from Anthropic wasn’t enough.

On Saturday, the U.S. government issued Anthropic an export control directive to suspend access for all foreigners using Fable 5 and Mythos 5.

The U.S. government cited national security concerns.

At the click of a button, my access to the Claude Fable 5 model was cut by Anthropic.

It was lights out.

But the deeper point should shake us.

At the click of a button, the U.S. government could cut the Australian continent off from the big five AI companies.

You’re probably now thinking:

What’s our fallback plan? Where is our redundancy? These are good questions.

And this is why the issue of Australian sovereign AI capability is so pressing.

The Economic Challenges posed by AI

But AI also poses massive economic implications for the Australian economy.

As Kai-Fu Lee outlined in his book AI Superpowers, AI will replace many jobs across physical and cognitive labour.

Lee, the former CEO of Google China, wrote AI Superpowers almost ten years ago but we are already seeing some of his prophecies come to pass.

We see how advanced robotics and AI have already transformed advanced manufacturing in China, as low-paid workers are replaced by machines.

And we see how Claude and ChatGPT are hollowing out white collar jobs in the services sector.

But Lee highlights two concerning trends in AI.

First, AI naturally tends towards monopolies.

AI consumes massive amounts of data to train and improve models.

Those with access to huge stores of data get a winning edge.

The more data an AI company gets, the better it gets.

Data is the key to market dominance.

That’s why the U.S. and China are the two AI superpowers. Both countries have access to huge amounts of data.

Now if AI tends toward monopoly, it also drives inequality.

The big winner-takes all economics of AI is already cutting across and reshaping many industries.

We are starting to see division in the labour market between those with high paid and low paid jobs, as middle income jobs are replaced by AI.

Downward career mobility is not the stuff of aspiration, so we can expect political ruptures in communities where jobs are replaced by machines.

Lee also predicts that income inequality will reshape the global order, as nations that fail to keep up with the AI revolution ill become poorer.

And those poorer nations will be at the mercy of AI superpowers.

But perhaps Kai-Fu Lee’s greatest insight was identifying the threat that AI poses to meaning.

If AI potentially starves people of work, we can expect great social upheaval.

If we strip from people the meaning that comes from creative and productive work, we can expect a revolt.

Given what is at stake:

Are we surprised by rising anxiety about the future?

Are we surprised by the rising anger towards political and business leaders?

Are we surprised by the dying trust in our institutions?

No wonder executives in Silicon Valley talk about a Universal Basic Income as a solution to job losses.

Anything to keep the unemployed masses happy.

But this is to miss Lee’s point about purpose and meaning.

Lee calls the UBI a ‘painkiller, something to numb and sedate the people who have been hurt by the adoption of AI.’

Lee drives home that the numbing effect goes both ways, as the UBI ‘also assuages the conscience of those who do the displacing.’

The takeaway here is that we really need is to think deeply about AI, and how we can preserve the common good for individuals and our communities.

All Australians must consider the deeper questions involved in AI.

But policymakers have a moral obligation to do the hard work of finding a political and economic settlement that serves all Australians.

Foremost must be the preservation of our Australian democratic traditions, our culture of a fair go and the dignity of work for all.

But to go deep on AI means that you hit theological and philosophical rock pretty quickly.

There are limits to talking about AI in terms of trade-offs. Because there are much deeper moral issues at stake.

Pope Leo XIV and the Moral Challenges Posed by AI

The good news is that Pope Leo XIV has kickstarted a global discussion on AI.

Last month, the Pope released his first encyclical Magnifica Humanitas – a 38,000-word meditation on AI.

Helpfully, it’s written in the natural law tradition and accessible to those beyond the Catholic Church.

Nonetheless, many of us are familiar with the language the Pope uses to make his case. As Tom Holland argued in his book, Dominion, the way we think about things in the West has been shaped by Christianity and its cultural achievements.

Whether we know it or not.

For the record, I did ask Claude to find me an equivalent meditation on AI from Protestant and Jewish faith leaders but there was no equivalent to the Pope’s First Encyclical.

But even as a Protestant Christian, I’m more than happy to say this a profoundly important work on AI.

Happily, the Pope starts with two images from the book of Genesis and Nehemiah that are familiar to both the Christian and Jewish traditions.

As we consider the future of AI, the Pope asks us to consider our motives in embracing this new technology.

First, he gives us the negative story of the Tower of Babel, built to reach the heavens from a place of pride and self-sufficiency.

The result was a breakdown in societal harmony and communication, as we read in Genesis.

Second, he gives us the positive story of the Jewish exiles returning from Babylon and rebuilding the broken walls of Jerusalem.

Together, the tribes of Judah, Benjamin and Levi worked together to rebuild the walls ‘for the people had a mind to work.’ (Nehemiah 4:6).

The Pope makes clear that the story of the exiles rebuilding the wall of Jerusalem is the one that should guide our approach to AI.

He calls on all people to ‘abandon the construction of yet another Tower of Babel and to join forces in building up the common good.’

Instead, we should take on the leadership posture of a Nehemiah, building up people, communities and institutions ‘even when a technocratic mentality or partisan interests seem to prevail.’

I think it’s also important to note that the Pope states that AI and technology is not inherently evil.

But neither is it neutral.

AI is a tool.

It merely reflects the assumptions, preferences and culture of its designers.

And so, we must be wise in the way we interact with it. Never uncritically.

All of this may seem quite remote to someone outside of the Christian faith.

I get that, and I respect that.

But if a Presbyterian can praise the Pope, maybe there is something here for us all.

Indeed, the Pope emphasises things common to all of us: the equal dignity of all human beings, the supreme value of human rights, the principle of the common good and the importance of subsidiarity—upholding and preserving all the little institutions that make up the fabric of Australian society.

What Edmund Burke called ‘the little platoons’.

The Encyclical also outlines the risks and consequences of AI to truth and democracy, the way we communicate with one another, the role of schools, the value of work and the problem of unemployment, and the future of institutions like the family.

After all, who wants to listen to a speech written by a machine?

(For the record: none of this has been generated by Claude).

We should welcome this invitation to a moral debate. And the Pope has given us a universal framework to consider the big questions posed by AI.

Nuclear Weapons and Australian Sovereignty

But, in the end, we are people of action.

We live in a world where we make imperfect decisions. And we need Australia to thrive in a world increasingly shaped and governed by AI.

We cannot sit by, passively waiting for the answers. We must strive to compete in the AI race today.

History reminds us that we have been here before, when Australian Prime Ministers considered what to do in the age of nuclear weapons after Hiroshima and Nagasaki.

Like the challenge of AI, the nuclear age forced Prime Ministers Robert Menzies and John Gorton to wrestle with the problem of Australian sovereignty and strategic independence.

The central question was stark: should Australia pursue a nuclear weapon of its own?

Most Australians aren’t aware that the Australian government only gave up on a nuclear weapon in the early 1970s, after signing the Non-Proliferation Treaty.

Robert Menzies spoke in Parliament in 1957 voicing Australia’s desire that nuclear weapons were to remain only in ‘the hands of the United States, the United Kingdom and the Soviet Union and no others.’

He said that Australia’s immediate defence plans should be conventional, only to clarify the following day he did not mean to ‘exclude the possibility of future procurement.’

This view—that Australia might one day need a nuclear weapon—prevailed in Canberra over the following decade, with the Department of Defence maintaining its option for future procurement by not explicitly ruling out the bomb.

But, in 1968, the Non-Proliferation Treaty brought the issue to a head.

Prime Minister John Gorton, new in office after Harold Holt’s death, received a Top Secret (Australian Eyes Only) briefing paper from the Australian Atomic Energy Commission on Nuclear Weapons Policy.

It made the case for a civil nuclear program with the strategic rationale of transitioning it to a weapons program, if required.

The briefing paper of March 1968 was clear:

‘Australia has the information, the nuclear skills and the industrial capacity to set up and operate facilities to produce fissionable material (i.e. plutonium or enriched uranium) of weapons grade, either alone or with overseas assistance.’

Then came the stern warning:

‘If Australia accedes to the Non-Proliferation Treaty, work on the development of knowledge relevant to the design of nuclear weapons in Australia will be prohibited.’

If Australia remained out of the treaty, it would take a minimum lead time of seven years to build a nuclear weapon, assuming a civil industry was in place.

Gorton wanted to keep his options open. He didn’t fully trust the U.S. or U.K. nuclear deterrent.

Would they really trade one of their cities for one of ours?

Gorton wanted Australia to have its own nuclear deterrent. And that’s why he was reluctant to sign the Non-Proliferation Treaty.

In April 1968, Gorton met with U.S. Secretary of State, Dean Rusk, in Canberra to discuss the Treaty.

Rusk wrote in a cable to Washington that in speaking with Gorton he ‘ran into a full battery of reservations about the Non-Proliferation Treaty.’

He wrote that Gorton:

…sounded almost like De Gaulle in saying that Australia could not rely upon the United States for nuclear weapons under ANZUS in the event of nuclear blackmail or attack on Australia.

I think it was a big compliment for an American Secretary of State to compare our Prime Minister to the French President Charles De Gaulle.

De Gaulle was man known for his belligerent pursuit of French interests.

The De Gaulle quip shows that Gorton did his job, pushing hard for Australian interests—including the option of a nuclear weapon.

That’s why Gorton pushed through Cabinet a plan in 1969 for a 500 Mega-Watt nuclear reactor in Jervis Bay.

None of this came to pass, however.

Gorton faced leadership issues within the party, and reluctantly signed the NPT in 1970.

Prime Minister William McMahon, a favourite of Tom Hughes, killed off the Jervis Bay nuclear reactor along with Australia’s strategic option for a weapons program.

Ever since, our strategic interests have been overwhelmingly shaped by our alliance with the United States.

You either have your own nuclear weapons, or you build a relationship with an ally who does.

We chose the relationship with the United States over an Australian nuclear deterrent.

That partly explains Naval Communication Station in Exmouth, Pine Gap, the Marine presence in Darwin and the AUKUS submarine base in Perth.

And why we have been so passive in developing our own defence capabilities.

The Way Forward for Australia: AI Sovereign or AI Supplicant?

We now face a similar moment in the global AI race.

Once again, we are faced with the problem of transformative dual-use technology controlled by great powers.

And I feel the weight of Australian history upon us.

I can hear Menzies and Gorton asking:

How hard are we working to preserve Australia’s sovereignty and strategic options?

Will we build our own sovereign AI capability and preserve our strategic independence?

Or will we take the path of AI non-proliferation and hope that the Americans look after us?

Tonight, I put on my De Gaullist kepi cap, and I say that we can’t afford to make the same mistake with AI, as we did with nuclear power.

What must we do?

It’s a stark choice, but the answer is clear.

We must build a sovereign AI capability.

Or accept that Australia’s future will be shaped by other nations. A future with limited strategic options.

Without sovereign AI, we will be a supplicant state tethered to the United States by submarine cables.

I don’t want that.

I’m a patriot, and I want Australia to be strong, free, prosperous and secure.

So, here are three closing thoughts:

First, we must play to our strengths.

We are too far behind the big AI companies to build frontier models.

At least for the moment. We should never say never.

But we do have an abundance of what frontier AI needs: land, energy, critical minerals, stable government and, most importantly, Australian talent.

Australia could become the AI compute hub of the Southern Hemisphere.

That’s why the energy debate is so tightly woven with AI.

AI is energy intensive – a ChatGPT search is ten times more energy intensive than a Google search.

AI companies are looking for reliable sources of energy, and Australia has a natural abundance.

Gas-fired power generation is critical to the rollout of data centres, and there is now a three-to-five- year backlog for orders on gas turbines.

Australia can meet that AI demand if we unlock our energy abundance, and let capital markets do the work.

That means abandoning Net Zero, removing the moratorium on nuclear power, and unlocking more of our gas and uranium.

It means deeper bilateral partnerships on energy and critical minerals with close partners in our region.

In short: we need to create the conditions for the AI hyperscalers like Anthropic to invest here, and partner with Australians to build a sovereign AI capability.

Getting the incentives and tax settings right will be important, so that we serve the interests of the Australian people—not the interests of American Big tech.

So will the governance framework for the deployment of AI in Australia.

But this is the play we must make to build sovereign AI here at home.

Second, we need to double our diplomatic efforts in AI.

Australia has an Ambassador to the Holy See in Rome. The Vatican is a city-state within Italy’s borders.

The five major American AI companies are starting the wield power and influence similar to that of city-states.

Elon Musk became the world’s first trillionaire over the weekend.

We should have an Ambassador for AI—based in Silicon Valley—with a mandate for engagement with the key AI leaders.

If the Pope is important enough to have an Ambassador, shouldn’t AI leaders—given their massive global financial and cultural power—have the same engagement from the Australian government?

We also need to make sure that we are engaged in all the global forums that deal with AI.

I note we have an Ambassador for Cyber Affairs and Critical Technology. She has a big job acting for our interests in global forums.

Yes, those forums can be frustrating.

But Australia will need to work through international institutions to secure our interests in respect of AI.

Finally, our education system needs an overhaul.

People are our greatest strength, and hope.

We must fix our education system, so we can unleash Australian hearts and minds on AI.

At the moment, we are unleashing AI on our kids. And they aren’t ready for it.

We need to get back to basics.

Maths. Science. Literature. History.

The Western Canon.

Writing by hand. Time in thought. Reasoning with others.

Why?

Innovation is driven by people, not machines.

We want our kids to build AI models rather than be slaves to them.

That’s why we need to invest in our young.

If we are going to survive the AI disruption ahead, we need young Australians who can think clearly, who can act justly and who can lead wisely.

There is much to done.

But I am confident that we thrive in the change ahead.

With that, I take off my De Gaullist kepi cap, and thank you for listening to me.

[ENDS]