Quantum Computers Could Solve AI’s Energy Problem — an Australian Startup Just Put One Inside a Real Data Centre

Artificial intelligence has an energy problem. The giant data centres that train and run AI systems now demand power on the scale of gigawatts — enough to need new power plants, strain electricity grids, and anger the communities that live next door. Now an Australian startup says part of the answer may come from a completely different kind of computer: a quantum machine that sips power instead of gulping it.

Diraq, a quantum computing company based in Sydney, has installed a small quantum computer built on a silicon chip inside a working commercial data centre — the first field test of its kind anywhere in the world, according to the company. The machine is running at Equinix’s data centre in Sydney, side by side with the ordinary servers that power today’s internet and AI services.

The system is tiny by the standards of the machines it hopes to help. It has just eight quantum bits, or qubits — the basic units of information in a quantum computer. But it draws less than 20 kilowatts of power, about as much as a few household kettles, while it runs alongside a normal computer at room temperature. Its founder, Andrew Dzurak, told the news agency AFP that quantum computing is moving out of university laboratories and will be ready to take some of the load off data centres within five years.

Why AI data centres use so much power

To understand why this matters, it helps to understand the problem. Training a modern AI system means running millions of calculations, over and over, on warehouse-sized buildings full of servers. Those servers need enormous amounts of electricity, and they also produce enormous amounts of heat, which takes even more energy to remove.

The result is that large data centres now measure their power needs in gigawatts. Building them can mean building new power plants or putting new pressure on local electricity grids — and that has started to cause pushback from the communities asked to host them.

Some companies are responding by building their own power supply. Anthropic, the AI company behind the Claude chatbot, is spending $3.4 billion on gas generators to power its AI data centres. Others are looking for new kinds of computing altogether. Diraq is betting that quantum computers can do some of the work that AI systems currently do, but far more efficiently.

What is a quantum computer, in simple terms?

An ordinary computer stores information as bits. Each bit is either a 0 or a 1, like a light switch that is either on or off. A quantum computer stores information as qubits, which use the strange rules of quantum mechanics — the physics that governs atoms and the tiny particles inside them.

Qubits can hold more information than ordinary bits, and they can work through certain kinds of calculations much faster. That makes them especially good at a particular class of problems: working out how molecules behave, finding the best answer among a huge number of options, and simulating the physical world. These are exactly the kinds of problems that AI systems chew through vast amounts of energy trying to solve.

“One of the reasons the next generations of AI are looking to consume so much power is that a whole class of the problems they are solving are problems that really are much better suited to quantum calculations,” Dzurak told AFP.

There is a catch. Most quantum computers only work at temperatures colder than outer space, inside huge and expensive cooling systems. That has kept them in laboratories, far away from the real world. Diraq’s approach is different: its system is built on silicon — the same material used to make ordinary computer chips — and it is designed to be self-contained. The machine in Sydney runs inside an enclosure with its own cooling system, drawing less than 20 kilowatts, next to a computer running at normal room temperature.

What Diraq has actually done

The deployment is a test, not a finished product. Diraq describes it as the world’s first field test of a small silicon-chip quantum computer inside a commercial data centre. The point is to find out how a quantum machine behaves in a real working environment, alongside ordinary servers, AI systems and cloud infrastructure — rather than in the controlled conditions of a laboratory.

The machine was announced on 9 November 2026 and is now running at Equinix’s Sydney facility. The two companies announced their partnership in August 2026, with the installation originally expected later in the year. The system takes up about 170 square feet of floor space, and Diraq says it plans to grow the installed system to as many as 256 qubits within the next year.

That is still a long way from the company’s bigger ambitions. Diraq plans to upgrade the system gradually to 150,000 qubits by 2029, and to build a system with two million qubits by 2031. Even then, Dzurak said, the energy use would be only “a couple of hundred kilowatts” — a fraction of what a single AI data centre uses today.

It is important to be clear about what the current machine can and cannot do. With eight qubits, it cannot replace any commercial AI workload. Nobody is claiming it can. What it can do is prove that a silicon quantum computer can live and work inside a real data centre, and help engineers learn how to run one there. That is the step that has to happen before quantum machines can take on real work.

What quantum computers could be used for

If the technology works as its makers hope, the uses go well beyond helping AI. Dzurak pointed to three areas in particular: working out how molecules behave, which could help design new medicines; building much lighter batteries; and improving chemical manufacturing.

The company estimates that using quantum computing for drug trials could raise success rates by 10 percent and cut the cost of bringing a new medicine to market by a third — but only if the system can run for under $100 million a year. “It will only happen if we can build quantum computers that are cost-effective,” Dzurak said.

That word — cost-effective — is the real test. Quantum computing has been “five years away” for a long time. What is different now is that governments and investors are starting to judge it by a harder standard: not whether it can do something clever in a lab, but whether it can solve a real problem at a price that makes sense.

That is exactly what the US Defense Advanced Research Projects Agency, known as DARPA, is testing. Diraq is one of only six companies in the world at the third stage of DARPA’s quantum benchmarking programme, which will test whether any quantum computer can solve real problems at a cost that delivers value by 2033. The company plans to reach DARPA’s definition of a useful machine by 2031, two years ahead of the deadline.

Australia’s quiet quantum boom

The Sydney deployment is also a sign of something bigger: Australia has quietly become one of the world’s leading countries in quantum science. The country has more than 50 quantum-related businesses, according to Cathy Foley, president of the Australian Academy of Technological Sciences and Engineering, who spoke to AFP. Australia has been a global leader in quantum research for decades, she said, partly because of investment in startups that grew out of its universities.

But turning discoveries into businesses is the hard part. “Australia is very good at making discoveries, but turning them into successful businesses requires much more than great science,” Foley said. “Commercialisation is certainly one of our biggest challenges.”

Diraq is not the only Australian company in the race. Another startup, PsiQuantum, is also on DARPA’s list and has received one billion Australian dollars in government investment to build a different kind of quantum computer in the state of Queensland. Construction started in June. And Australian physicist Michelle Simmons’ startup, Silicon Quantum Computing, was named by Time magazine as having one of the best inventions of 2026, for a quantum-enhanced AI chip.

Dzurak described the moment as Australia’s “Silicon Valley moment” in quantum. The Australian government has made quantum a priority through AUKUS, its defence technology partnership with Britain and the United States. Foley said AUKUS has reduced the security barriers for sharing quantum technology between the countries — something Dzurak said has been “essential” for Diraq to work across Australia and its offices in the larger US market.

Why this matters for the AI energy crunch

The timing is no accident. AI is growing fast, and so is the argument about its energy use. Data centres are getting bigger, power grids are under strain, and the companies building AI systems are under pressure to show they can grow without consuming ever more electricity.

Quantum computing will not solve that problem on its own, and it will not solve it soon. Even Diraq’s own timetable puts useful machines five years away. But the Sydney test is a concrete step towards a future in which some of the hardest calculations — the ones AI currently brute-forces with huge amounts of power — are handed to machines built for the job.

It also matters because of how the machine is built. Diraq’s quantum processors use silicon technology and manufacturing processes similar to those used for ordinary computer chips. That means, in theory, they could one day be made in the same factories that already produce billions of chips a year — rather than in specialist labs. That is the path from a laboratory curiosity to a technology that can be built at scale.

The same week Diraq’s machine went live, the world’s climate negotiators were meeting in Antalya, Turkey, for the COP31 summit, where the energy demands of AI and the pressure on power grids are part of a much wider argument about how the world powers its future. 2026 is on track to be one of the hottest years ever recorded, and the question of how to run the technology of the future without overheating the planet is no longer abstract.

For now, the Sydney machine is small, experimental and a long way from doing real work. But it is running in a real data centre, using real power, in the real world. For a technology that has spent decades in the laboratory, that is the point.

What happens next

Diraq will spend the coming months testing how its machine performs inside the Equinix data centre, and upgrading it towards 256 qubits. The bigger milestones are further out: 150,000 qubits by 2029, and a two-million-qubit system by 2031. DARPA’s benchmarking programme will deliver its verdict by 2033, testing whether any quantum computer can solve real problems at a price that makes sense.

In the meantime, the company is also leading an Australian government-backed group — including the energy market operator AEMO, UNSW Sydney, Macquarie University and OPAL-RT — looking at whether quantum computing can improve how increasingly complex electricity networks are managed. The project received $500,000 through the Critical Technologies Challenge Program.

If the Sydney experiment works, the next step will be bigger machines in more data centres. If it does not, it will join a long list of quantum promises that arrived too early. Either way, the answer will now come from a working data centre, not a laboratory.

Frequently asked questions

What is Diraq?

Diraq is an Australian quantum computing startup based in Sydney. It builds quantum computers using silicon chips, similar in design to ordinary computer processors.

What has Diraq just done?

It has installed an eight-qubit silicon quantum computer inside Equinix’s commercial data centre in Sydney — the first time a small silicon-chip quantum computer has been field-tested in a working commercial data centre.

How much power does it use?

Less than 20 kilowatts, including its cooling and control equipment. Diraq says a future two-million-qubit system would use only a couple of hundred kilowatts.

Can it replace AI computers today?

No. With eight qubits, the machine is far too small for commercial work. Its purpose is to test how quantum computers operate alongside normal computing infrastructure in a real data centre.

When will quantum computers be useful?

Diraq says its machines could start easing the load on data centres within five years. It aims to reach 150,000 qubits by 2029 and two million qubits by 2031. DARPA’s benchmarking programme will test whether quantum computers can solve real problems at a useful cost by 2033.

Why is Australia important in quantum computing?

Australia has more than 50 quantum-related businesses and has led quantum research for decades. The government has made quantum a priority through the AUKUS partnership with Britain and the United States.

Conclusion

AI’s hunger for electricity is one of the biggest challenges facing the technology industry. Diraq’s small machine in a Sydney data centre does not solve it — not yet. But it is the first time a silicon quantum computer has been put to work in the place where the problem actually lives, and that is a meaningful step. The next five years will show whether quantum computing can grow from a laboratory promise into a practical tool — and whether the machines that help AI think can also help it stop consuming so much power.

Sources

• AFP via Phys.org, “Quantum could solve AI energy drain, says tech founder”, 10 November 2026: https://phys.org/news/2026-11-quantum-ai-energy-tech-founder.html
• Kuwait Times, “Quantum computing could curb AI’s energy drain, says Australian founder”: https://kuwaittimes.com/article/53354/world/quantum-computing-could-curb-ais-energy-drain-says-australian-founder/amp/
• The Economic Times, “Quantum could solve AI energy drain, says tech founder”: https://economictimes.indiatimes.com/tech/artificial-intelligence/quantum-could-solve-ai-energy-drain-says-tech-founder/articleshow/135699105.cms
• Lapaas Voice, “Diraq Unveils World’s First Commercial 8-Qubit Quantum Computer Field Deployment in Sydney”: https://lapaasvoice.com/diraq-unveils-worlds-first-commercial-8-qubit-quantum-computer-field-deployment-in-sydney/
• AltEnergy, “Diraq quantum technology targets AI’s growing energy appetite”: https://altenergy.com.au/newsandviews/show/181/diraq-quantum-technology-targets-ais-growing-energy-appetite

This article is for general information and education only. It is not legal or financial advice.