The Answer to Australia's Energy Crisis Is Not Where You Think

I grew up watching countries talk about the future. India just built it.

I have spent most of my career inside infrastructure. Not talking about it. Not advising on it from a distance. Inside it. Managing over 2,200 critical Telstra network sites across New South Wales, where the consequence of a single failure was not a bad quarterly report but an entire region losing connectivity. That kind of work teaches you something that no MBA ever will: the gap between planning and execution is where most ambition goes to die.

So when I read that India's 500 MWe Prototype Fast Breeder Reactor had achieved first criticality at Kalpakkam on 6 April 2026, I did not see a headline about nuclear physics. I saw a country that had refused to let that gap defeat them for over two decades.

The PFBR is a sodium-cooled, pool-type reactor running on Uranium-Plutonium Mixed Oxide fuel. It was indigenously designed by the Indira Gandhi Centre for Atomic Research and built by BHAVINI under India's Department of Atomic Energy. Construction began in 2004. The original completion target was 2010. It took sixteen years longer than planned, cost more than twice the original estimate, and required solving engineering problems that had never been solved before at this scale.

None of that matters as much as the fact that they finished.

What a fast breeder reactor actually does, and why founders should care

A conventional nuclear reactor consumes fuel and eventually stops. A fast breeder reactor generates more fuel than it burns. The PFBR's core is surrounded by a Uranium-238 blanket. Fast neutrons convert that fertile material into fissile Plutonium-239 through neutron absorption, creating a self-reinforcing fuel cycle. The reactor is also designed to eventually use Thorium-232, breeding Uranium-233 to power Stage Three of India's nuclear programme.

That is capital efficiency and energy leverage expressed at the atomic level. And the strategic logic behind it is something every founder, investor, and infrastructure builder should understand.

The long arc of category leadership

The PFBR is Stage Two of a three-stage nuclear power programme first conceived by physicist Homi Bhabha in the early 1960s. Stage Three will unlock India's vast thorium reserves, estimated at roughly 25 per cent of the global supply, for centuries of energy independence. This roadmap was drafted decades before most of today's startup founders were born, and India is still executing against it.

True category leadership in any domain, whether AI, energy, or data centres, belongs to people willing to commit to plans that outlive election cycles and quarterly earnings. I think about this often. At Sharktech Global, every platform we build is designed to still matter in ten years, not just next quarter.

Designing around resource asymmetry

India has abundant thorium and constrained uranium. Instead of accepting dependence on foreign fuel and fragile supply chains, they engineered a closed fuel cycle around their actual resource endowment. They made physics and geology work for them.

This is the same mindset I try to apply to everything we build. When you are a technology company operating from Western Sydney, not Sand Hill Road, you do not get to copy the Silicon Valley playbook and hope for the best. You have to design around local constraints. You have to make what you have work harder than what others take for granted.

Sovereign capability as strategy, not slogan

While other nations pivoted in and out of nuclear fashion, India stayed on thesis. With the PFBR now critical, India becomes only the second country in the world after Russia to operate a commercial-scale fast breeder reactor. That is what staying the course looks like in practice. Not a press release. Not a policy paper. A working reactor.

Now bring this back to Australia.

We are entering an era of extreme compute demand. Agentic AI, large-scale model training, and real-time video analytics will require serious, continuous, uninterrupted power. And Australia is not ready.

I do not say that to be provocative. I say it because the numbers are already in front of us and nobody seems to be connecting them.

The Australian Government recognised this trajectory when it released its National Expectations of Data Centres and AI Infrastructure Developers on 23 March 2026, setting out five pillars covering energy transition, water sustainability, workforce investment, national security, and local innovation capability. Proposals that closely align with these expectations will receive priority consideration through Commonwealth regulatory processes.

All of this is extraordinary. And all of it depends on one thing nobody has adequately solved: where does the power come from?

The investment signals are unmistakable. OpenAI has committed to a sovereign AI campus at Eastern Creek through its first Asia-Pacific partnership under the OpenAI for Countries programme. Amazon Web Services announced a $20 billion investment in Australian data centres in 2025, the largest technology investment in the nation's history. Gartner forecasts Australian IT spending will reach $172.3 billion in 2026, with data centre systems growing 22.5 per cent year-on-year to $10.1 billion. Australia is no longer debating whether it will become a major AI infrastructure market. The capital has already decided. The question is whether the energy infrastructure can keep pace.

The bottleneck that nobody wants to name

The bottleneck for AI infrastructure in Australia is no longer silicon, talent, or capital. It is uninterrupted, baseload power and sites that are genuinely "AI-ready": power-dense, grid-integrated, and zoning-approved.

Australia's existing energy system was built for traditional industrial loads. AEMO estimates that 6,000 kilometres of new transmission and significant investment in storage will be required to keep pace with energy transition and digital-load growth. Without accelerated reform, new grid connections may face delays measured in years, not months.

Meanwhile, data centres draw from the public drinking water supply. Land use conflicts are intensifying in Western Sydney, where data centres compete directly with housing and employment-generating development. A NSW parliamentary inquiry into data centre regulation is underway, with a final report due by September 2026.

And here is the uncomfortable truth: trying to build ten-megawatt GPU clusters on legacy industrial estates without thinking about power, utilities, and approvals from day one is not innovation. It is a failure of first-principles thinking.

The nuclear question Australia keeps avoiding

If Australia had the political will to pursue advanced nuclear energy, the equation changes entirely. Not tomorrow. But the strategic window is open, and the countries making long-term energy commitments now will own the infrastructure advantage for decades.

The Lowy Institute published a piece in March 2026 arguing that nuclear power is a sovereign capability, not merely an energy technology. Their position is direct: energy policy can no longer be separated from geopolitics. In a world where trade is weaponised and supply chains are leveraged for coercion, countries that control their own energy future are countries that control their own destiny.

Australia holds roughly one-third of the world's known uranium reserves. It is the world's fourth-largest uranium producer. And yet it maintains a federal moratorium on civilian nuclear power under the Australian Radiation Protection and Nuclear Safety Act 1998. Most states carry their own bans. This is the same country that signed the AUKUS agreement to build nuclear-powered submarines, a decision that required developing sovereign nuclear expertise for the first time. The industrial and engineering capability required for AUKUS overlaps significantly with what would be needed to develop civilian nuclear energy. The foundations are being laid whether the policy conversation has caught up or not.

I am not writing this as a policy advocate. I am writing it as someone who has spent years inside critical infrastructure and understands what happens when demand outstrips supply. Whether Australia's position on nuclear energy shifts as compute demand compounds and baseload constraints tighten is a question of growing strategic significance. What India demonstrated at Kalpakkam is that the engineering is possible. The question for Australia is whether the political will is.

Capital efficiency in the next decade will be defined by energy sovereignty and infrastructure intelligence. The builders who win will be the ones who own their power strategy, own their site strategy, and own their distribution.

This is the problem I built DivinelabWorx to solve

I started DivinelabWorx because I kept watching the same failure pattern repeat. Serious investors and builders would commit capital to AI infrastructure, data centres, or advanced manufacturing, and then burn twelve to eighteen months discovering that the site they chose could not get power, the zoning was wrong, the utility interfaces had not been designed, or the planning approvals were stuck in a queue behind fifty other applications.

That is not a technology problem. It is an infrastructure intelligence problem. And it is exactly the kind of problem I have spent my career solving.

DivinelabWorx is an infrastructure consultancy. We help builders and investors identify underutilised existing assets: former industrial sites, locations near high-capacity substations and power corridors, zoned land that most people overlook. We validate site-level feasibility across power proximity, grid access, fibre path, water access, and WHS compliance. And we do all of that before capital hits the ground, so serious operators are not burning years on site selection and approvals that should have been resolved from the start.

Our focus is compressing the critical path from investment decision to compute online. In a market where NEXTDC's S4 took two years of early contractor involvement before breaking ground, and where the Mamre Road campus will require staged construction over a decade, the ability to shave months off the front end of a project is worth real money.

When I managed 2,200 Telstra sites across NSW, the discipline was the same: understand the site before you commit to the build, design the hard constraints in from the start, and never let a planning failure become an operational failure. DivinelabWorx applies that same discipline to the next generation of critical infrastructure.

What I actually believe about what comes next

I believe the next decade will sort Australian infrastructure into two categories: builders who understood the energy constraint early enough to design around it, and everyone else.

I believe Western Sydney is going to become one of the most strategically important corridors for AI infrastructure in the Asia-Pacific, not because of hype, but because it has the land supply, the grid proximity, the zoning frameworks, and the political will to make it happen. The Mamre Road Precinct, the Horsley Park corridor, the Western Sydney Employment Area, these are not speculative plays. They are active development zones with real capital flowing in.

I believe that India's achievement at Kalpakkam is not just an energy milestone. It is a signal that the countries willing to think in decades, design around their constraints, and stay on thesis through difficulty will be the countries that own the next century of energy and compute infrastructure.

And I believe Australia has every asset it needs to be one of those countries. The uranium reserves. The land. The institutional stability. The political frameworks. The investment appetite. The only thing missing is the willingness to think from first principles about energy, the way India has been doing since the 1960s.

Build for scale. Build for permanence. The future belongs to those who engineer their own leverage.