The data centre boom is rewriting the rules of energy investment. Surging demand for AI, cloud computing and raw processing power is driving unprecedented growth in data centres, but our energy infrastructure is not keeping pace.
For investors, this creates a landscape rich with opportunity - but only for those who understand the two fundamental bottlenecks: grid access and generation capacity. These constraints are determining which projects proceed, where capital flows, and ultimately who captures value.
The Australian Government's newly released Expectations of data centres and AI infrastructure developers (DC Guidelines) have brought these challenges into focus. Operators are now expected to fund their share of grid infrastructure costs and secure their own power generation.
The solution is deceptively simple: increase grid access and electrons. But the challenge lies in the execution...
Gridlock – brought to you by sharing
Data centre demand is accelerating rapidly, with consumption projected to reach 6% of NEM grid-supplied electricity by 2029-30, and 12% by 2049-50. The fundamental constraint is grid capacity.
In core markets such as Sydney and Melbourne, available network capacity is close to being exhausted. Transmission infrastructure is not scaling at the pace required to support incremental load, creating a structural bottleneck. Unlocking additional capacity will likely require multi-billion-dollar investment in network augmentation.
This raises an inevitable question: Who pays? The DC Guidelines establish that operators (and not electricity consumers or taxpayers) should fund their 'share of transmission and distribution infrastructure costs'. While this principle is simple in concept, its practical application is far from straightforward. For downstream connection assets (ie assets that physically connect a data centre to the grid), the model is clear and well-established: a data centre owner (or its customers) should underwrite connection costs. This aligns with current regulatory settings and prevailing market practice.
However, for upstream transmission augmentation – particularly large-scale, ‘step-change’ upgrades to shared infrastructure - the framework becomes less certain. These augmentations deliver system-wide benefits and require capital commitments that are unlikely to be underwritten by a single operator. It also highlights potential challenges in the existing regulatory framework. While cost-sharing across multiple data centre developers is theoretically efficient, in practice it introduces coordination challenges across competitors including:
- alignment on timing and capacity requirements
- allocation of capital contributions, and
- commercial, governance and risk-sharing complexities.
Against this backdrop, there is a strong case for evolving the regulatory framework to enable coordinated funding of upstream grid augmentation costs – while ensuring operators are not over-exposed to shared system costs.
The generation gap
Even if grid access constraints were solved tomorrow, Australia still faces a structural shortfall in available generation to support the next wave of data centre development.
The federal government's policy position — now formalised in the DC Guidelines — is clear: data centres should 'procure additional clean energy generation and/or storage to offset demand'.
Helpfully, the DC Guidelines provide flexibility in how this expectation may be achieved, with three examples outlined below:
1. Onsite generationNon-mandatory co-location |
The Guidelines do not require physical co-location of generation assets with the data centre it supplies. While co-location may be viable in select regional markets, it is unlikely to be economically viable for data centres located in metropolitan areas where space is often priced at a premium, and the cost of land is higher. This reinforces the case for off-site generation paired with contractual structuring, rather than balance-sheet-heavy real estate allocation. |
2. Ownership of generation assetNot required |
There is no requirement for data centre operators to take equity ownership in generation or storage assets. While co-ownership may appeal to some infrastructure-backed operators (particularly for data centres owned by superannuation or infrastructure funds who already invest in generation and storage), mandating co-ownership would materially increase capital intensity to the already high costs and funding requirements of projects. |
3. Third-party offtakePPA-led model |
The DC Guidelines appear to allow offtake or power purchase arrangements with third-party projects. This could be viewed by data centre operators as a capital-efficient solution and aligns with hyperscalers such as Microsoft or Amazon continuing to underwrite new renewable and storage projects via long-term contracts. For investors, this reinforces a familiar model where counterparties anchor project finance structures, supporting bankability and scale deployment. |
Despite this helpful flexibility, the definition and practical application of ‘additionality’ (that data centre operators should actively contribute to the development of new renewable energy projects, rather than merely drawing upon existing generation capacity) remains unresolved. One key area is whether buying power from a renewable energy project that is already supported by government (for example, under the ESEM) meets the ‘additionality’ requirement.
A central question for the market is:
Does contracting with a government-supported renewable project satisfy the additionality requirement, or must capacity be incremental beyond existing policy-backed pipelines?
There would be clear benefit in allowing data centres operators to buy power from government-supported projects. In particular, the government support schemes are not intended to provide all of the support required to underwrite a project. Instead, they are designed to incentivise projects to sell contracts to large customers (like data centre operators and customers), meaning that data centre demand for power would help realised the objectives of the schemes.
The bottom line
The data centre boom is not a problem to be managed - it is an opportunity to be seized.
Grid constraints and generation shortfalls are real, but so is the demand - and it is not going away. For generators and energy developers willing to move quickly, this is one of the most compelling opportunities in the current market.
Those who can deliver firmed, reliable power at pace will be well positioned to capture the gains.
