Australia’s energy system is retiring its centre, and rebuilding at the edge

For most of Australia’s electricity history, reliability was delivered by a small number of large, centralised generators pushing power outward through transmission networks to passive consumers.

Across the country, large coal-fired generation assets are approaching retirement, many earlier than originally planned.

In the National Electricity Market (NEM):

• Eraring (NSW) - Australia’s largest coal plant - is scheduled to close in 2027
• Yallourn (VIC) is set to retire in 2028
• Vales Point (NSW) has extended its technical life but remains within the 2030s horizon
• Loy Yang A (VIC) is currently scheduled for closure in 2035

In Western Australia’s SWIS:

• Muja C retires in 2025
• Collie follows in 2027
• Muja D retires by 2029

Collectively, these closures represent many gigawatts of always-on generation exiting the system within a decade.

Their replacement is not a new generation of baseload power stations. Instead, Australia is transitioning to a fundamentally different architecture: generation and flexibility moving to the edges of the grid, coordinated through digital orchestration, virtual power plants (VPPs), and firming assets.

The Retirement Wave: ‘The Control Gap’

Australia’s large thermal fleet was largely built between the 1960s and 1980s. Much of it is now operating beyond its original design life, with increasing reliability risks, higher maintenance costs, and growing exposure to market volatility.

In the National Electricity Market and Western Australia’s SWIS alike, multiple gigawatts of large, always-on generation are scheduled to retire over the next decade. These exits create not only an energy gap, but a control gap - the loss of predictable, centrally managed demand and supply and what replaces that ‘control gap’ matters more than what replaces the megawatts.

The Myth of “New Baseload”

The question “what will replace baseload?” misunderstands what is actually leaving the system.

Baseload was never a single thing. It was a bundle of services - energy, capacity, inertia, frequency response - delivered by assets that happened to run continuously.

Australia is now unbundling those services and sourcing them from many assets, in many locations, coordinated in real time.

The question it challenges is not only what gets built, but where intelligence must sit in the system.

Where the Replacement Capacity Is Actually Coming From

Energy is increasingly delivered by wind and solar, much of it embedded deep in the distribution network: rooftops, commercial sites, industrial estates, and community-scale projects.

Firm capacity comes from batteries, pumped hydro, and selective flexible gas. System services are delivered by inverters, synchronous condensers, and fast-response digital control.

But there is a missing piece in this picture - one that does not show up as a line item in megawatts installed.

The New Scarcity: Controllable Load

As generation decentralises, controllable demand becomes as valuable as controllable supply.

Historically, large industrial loads were rare, inflexible, and poorly integrated into market operations. Today, new forms of demand are emerging that are:

• Highly energy intensive
• Location specific
• Time flexible
• Digitally controllable

This is where the architecture of the grid quietly flips.

Instead of shaping generation to meet demand, demand can now be shaped to meet generation.

Project Ohm: Turning Compute into Grid Infrastructure

Project Ohm sits precisely at this inflection point.

Rather than treating energy-intensive compute as a passive consumer of electricity, Project Ohm reframes it as a dispatchable, grid-aware resource.

By deploying GPU-dense compute nodes at the edge of the grid - often near constrained substations or renewable generation - and orchestrating workloads dynamically in response to real-time grid conditions, Project Ohm turns compute into a form of digital demand-side infrastructure.

In effect, it behaves like a virtual power plant - but operating on the load side rather than the supply side.

When renewable generation is abundant and prices are low, compute ramps up. When the grid is constrained, compute ramps down. When system services are needed, demand responds accordingly.

The innovation is energy-native computing in a changing energy landscape.

Why This Matters in a Post-Baseload Grid

As coal retires, the grid loses more than energy - it loses inertia, predictability, and large blocks of controllable load.

Project Ohm provides:

• Absorptive capacity for excess renewable generation
• Fast demand response during grid stress
• Locational flexibility aligned with network constraints
• Digital orchestration that complements VPPs and storage

In a system dominated by edge generation, these capabilities become infrastructure-grade.

They reduce curtailment. They improve asset utilisation. They defer network upgrades. They stabilise increasingly complex grids.

From Data Centres to Distributed Digital DERs

Traditional data centres are designed for reliability through redundancy and constant power availability. They are grid-agnostic by design.

Project Ohm inverts this model.

It treats compute as a distributed energy resource (DER) - one that can be scheduled, paused, shifted, and optimised in alignment with the physical realities of the electricity system.

In doing so, it bridges two transitions that are often discussed separately:

• The decentralisation of energy
• The explosive growth of AI and compute demand

At the grid edge, these transitions are converging.

The Emerging Energy Architecture: Orchestration Is the Asset

In the new energy system, advantage no longer comes from owning the biggest generator or the largest data centre.

It comes from orchestrating flexibility across distributed assets - generation, storage, and controllable load - in real time.

Project Ohm is one example of this new class of infrastructure: assets that are simultaneously digital and physical, computational and electrical, local and system-relevant.

As Australia retires its centralised generation, it is not replacing it with a single new pillar.

It is replacing it with an ecosystem.

And in that ecosystem, intelligence at the edge is no longer optional - it is the grid.