The nuclear energy industry has a problem: it takes years, costs hundreds of millions of dollars, and demands mountains of paperwork just to get a plant approved. Now two of the world's biggest tech companies are throwing AI at the problem.
Microsoft and Nvidia have announced an AI-powered collaboration to accelerate the development and deployment of nuclear power plants, combining generative AI, digital twin simulation, and Nvidia's Omniverse platform to streamline the nuclear lifecycle from permitting through operations. The partnership reflects a simple economic reality: AI data centres are hungry for power, and nuclear offers the kind of reliable, carbon-free electricity that solar and wind alone cannot provide.
Here's what the technology actually does. Engineers typically spend thousands of hours drafting, cross-referencing, formatting, searching, reviewing, and reworking materials, and must identify and fix inconsistencies across tens of thousands of pages. That manual work is precisely what generative AI excels at eliminating. AI tools can help identify documentation inconsistencies, unify data across the lifecycle of plant construction, and support digital twins, or virtual replicas that allow engineers to test changes; generative AI can also help align new applications with past permits and simulate projects before shovels hit the dirt.
In design and engineering, digital twins and high-fidelity simulations allow engineers to reuse proven design patterns and model the downstream effects of changes before construction begins, while construction gets 4D and 5D simulation, adding time scheduling and cost tracking to standard 3D spatial models.
The proof is already emerging. Aalo Atomics has reduced the time-intensive permitting process by 92% using the Microsoft Generative AI for Permitting solution, saving an estimated $80 million a year. For a single startup to achieve that kind of efficiency gain suggests the technology works in the real world, not just theory.
But here's where the Australian angle matters. According to the International Energy Agency, data centres demand increased by more than three quarters between 2023 and 2024 and is expected to account for over 20 per cent of electricity-demand growth in advanced economies by 2030. In NSW, the grid operator Transgrid estimates data centre power demand will grow from 4 per cent of the state's electrical use to 11 per cent by 2030, while in Victoria, the Australian Energy Market Operator estimates by 2030 data centre power usage will rise from the current 2 per cent of Melbourne's power usage to 8 per cent.
The government has signalled it wants Australia to be competitive in data centre investment. Global investment in data centres is accelerating and Australia is well-placed to lead, but this must happen on terms that benefit the community and deliver for the national interest. Yet Australia faces a structural problem Microsoft and Nvidia cannot solve with code. Nuclear is not viable in Australia in the near term, and US-style models do not fit local settings; the solution must be domestic firming and smarter site selection.
The US is moving fast. In the US alone, big tech companies have signed new contracts for more than 10 GW of possible new nuclear capacity in the last year, and Goldman Sachs Research sees potential for three plants to be brought online by 2030. Constellation Energy is expected to restart the Three Mile Island nuclear plant to power Microsoft's data centers in 2027. Meanwhile, Australia currently has no baseload replacement strategy, which is a significant reason the country continues to extend its coal generation, as Australia has no ability to import baseload power from neighbours and is dependent solely on its own energy systems.
This creates an awkward dynamic. Australian policymakers want the jobs and investment that come with hosting global AI infrastructure. Tech companies want to build here. But without baseload power sources beyond coal and renewables, the grid constraint becomes real. If local grids cannot keep up, investment will move; hyperscalers optimise for speed to operation, and if Australia cannot move faster, investment will flow to markets that can; if Australia fails to resolve the power equation, the country risks slowing AI adoption, delaying cloud migration and pushing investment offshore.
Microsoft and Nvidia's collaboration solves a genuine problem for nuclear development. It makes the permitting, design and construction of nuclear plants faster and cheaper, which lowers risk and makes projects more bankable. But tools that accelerate nuclear deployment in the US, UK or France don't automatically unlock nuclear for Australia. That requires legislative change. Policy decisions about nuclear power lie entirely with governments, not with tech companies armed with generative AI. For now, Microsoft and Nvidia are playing their hand well. Australia is still trying to figure out what cards it's holding.