In March 2026 the Australian Energy Market Commission published a draft rule that will change how large data centres connect to the grid. It is technical, it is easy to overlook, and it will directly affect the cost, the timeline and the design of almost every serious data centre project in the country. If you are developing or operating at scale in Australia, this is a rule you need to understand now, not when your connection agreement lands.

What the AEMC has actually proposed

The draft rule does two central things. First, it raises the threshold for large inverter based loads from five megawatts to thirty megawatts. Second, and more importantly, it requires large facilities to ride through disturbances on the grid rather than tripping offline. In plain terms, a data centre must stay connected during defined voltage and frequency events and recover its power draw within set timeframes.

30 MWThe proposed threshold, written into the National Electricity Rules, above which inverter based loads such as large data centres must meet firm disturbance ride through standards.

The AEMC has also signalled that the Australian standards would be aligned with mature international regimes in Texas, Ireland and Finland. That alignment matters commercially. It means the engineering your design team already knows from other markets will map more cleanly onto Australian requirements, which reduces cost and complexity rather than adding a bespoke local burden.

Why the regulator is doing this

The concern is grid stability, and there is a concrete cautionary tale behind it. When the grid experiences a credible disturbance, such as a sudden voltage dip, facilities that are not designed to ride through can disconnect all at once. If enough large loads drop simultaneously, the result can be a cascading instability event.

The regulator points to a real incident. In July 2024, sixty data centres in the United States state of Virginia pulled about 1,500 megawatts off the grid simultaneously during a single fault. As data centres grow from a rounding error to a material share of national load, that behaviour becomes a system risk, and regulators are moving to prevent it before it happens here.

Ride through capability is moving from a nice to have to a condition of connection.

The dates that matter

The consultation on the draft rule closed on 7 May 2026, with a final determination expected around the middle of 2026. Connection agreements are expected to carry terms that keep even non registered data centres compliant with performance standards, and operators will need to provide performance data to their network. In other words, compliance will be baked into the connection contract, not left to good intentions.

What operators and developers should do about it

This is not a reason to slow down. It is a reason to design correctly the first time. Three practical steps follow.

  • Design for ride through from the outset. Retrofitting disturbance ride through capability into a facility that was specified without it is expensive and slow. Building it in at design stage is neither.
  • Confirm your threshold position. Whether your facility sits above or below the thirty megawatt line changes your obligations. For phased builds, model the load at each stage, not just at day one.
  • Read the connection agreement as a compliance document. The performance standards and data obligations will live in that contract. Treat it with the same rigour you would give a planning consent.

Key takeaways

  • The AEMC's March 2026 draft rule raises the large inverter based load threshold from 5 MW to 30 MW and mandates disturbance ride through.
  • The driver is grid stability, informed by a July 2024 Virginia event where 60 data centres shed about 1,500 MW at once.
  • A final determination is expected around mid 2026, with compliance embedded in connection agreements.
  • Australian standards are being aligned with Texas, Ireland and Finland, which lowers cost for operators who design to them early.

How DivineLab Worx helps

Regulatory change is only a threat when it arrives as a surprise. Our regulatory and compliance work maps the approvals, standards and connection obligations that apply to a project, and coordinates the qualified Australian specialists who deliver against them, before commitments are made. For data centre and AI infrastructure clients that means the grid connection standards, the planning pathway and the security obligations are scoped as one picture, so nothing lands late. If a new rule is about to reshape your project economics, the time to model it is before you sign.

Dainu Devis

Chief Executive Officer, Sharktech Global

Dainu Devis is the Chief Executive Officer of Sharktech Global, the Australian technology group building products for a world being reshaped and displaced by artificial intelligence. Through its advisory arm, DivineLab Worx, and ventures across critical infrastructure, hospitality and industrial safety, Sharktech backs the operators, builders and businesses that intend to still be standing on the other side of the AI transition. Dainu advises operators, developers, boards and governments on where to build, what to secure, and how to turn strategy into revenue. More about DivineLab Worx and Sharktech Global.

Sources and further reading

This article is general commentary for infrastructure operators, developers and investors. It is not legal, planning or engineering advice. DivineLab Worx coordinates qualified Australian specialists within each engagement.