All posts
Energy Economics· 7 min read

Understanding MLF and why it costs data centres millions

Marginal Loss Factors are one of the most overlooked numbers in Australian energy economics — and one of the most expensive. Here's how MLF works, why it moves, and how a single decimal can erase millions from a data centre's business case.

By Zyntax Arc

Ask most people what drives a data centre's energy cost and they will name the wholesale price. They are not wrong — but they are missing a multiplier that quietly sits on top of every megawatt-hour, recalculated every year, and capable of moving a project's economics by millions of dollars. That multiplier is the Marginal Loss Factor, or MLF.

It is one of the least understood numbers in the National Electricity Market. It is also one of the most important for anyone connecting a large load.

What an MLF actually is

Electricity loses energy as it travels across the transmission network. The further power has to move, and the more congested the path, the more is lost on the way. The NEM accounts for these losses using loss factors assigned to each connection point.

The Marginal Loss Factor is the factor that applies between a connection point and the regional reference node — effectively, how efficiently energy moves between where it is priced and where you consume or generate it. An MLF of 1.0 means no marginal loss relative to the reference node. An MLF of 0.90 means roughly ten per cent is lost on the margin.

For a generator, a lower MLF means you are paid for less than you produce. For a load like a data centre, the effect flows through your delivered cost: your effective energy bill is shaped by the loss factor applied to your connection point.

The key facts that make MLF dangerous to ignore:

  • It is location-specific. Two sites a short distance apart can have materially different MLFs.
  • It is recalculated annually. AEMO publishes updated loss-factor tables each year. A favourable MLF today is not guaranteed tomorrow.
  • It is driven by what else is on the network. New generation, new load and network changes near you all move your MLF — often without warning.

Why a single decimal becomes millions

The arithmetic is unforgiving. Consider a hyperscale-class load drawing 100 MW at high utilisation:

  • 100 MW running at an 85% load factor is roughly 745,000 MWh per year.
  • At an indicative all-in energy cost of $90/MWh, that is about $67 million in annual energy spend.
  • Now apply MLF. The difference between an MLF of 0.98 and 0.92 is six percentage points applied across that entire volume.

Six per cent of a ~$67 million energy bill is around $4 million — every year. Over a 15-year horizon, before discounting, that is on the order of $60 million swung by a number most site shortlists never look at.

The wholesale price gets all the attention. The loss factor sitting on top of it is where projects quietly win or lose tens of millions.

And because MLFs are reset annually, the risk is not static. A site chosen on a strong MLF can degrade as new neighbours connect — turning an assumption baked into your business case into an annual surprise.

The MLF table is public — and almost nobody reads it well

AEMO publishes the loss-factor tables openly. The challenge is not access; it is interpretation and currency:

  1. Which connection point applies to your candidate site? Matching a parcel of land to the right node is non-trivial.
  2. What is the trend? A single year's MLF tells you little. The trajectory — and what is driving it — tells you the risk.
  3. How does it interact with everything else? A strong MLF in a region riddled with curtailment, or with weak renewable economics, may not be the bargain it appears.

This is precisely why MLF cannot be assessed in isolation. It is one input into a connection point's true delivered cost.

How Zyntax Arc treats MLF

Inside the DC Intelligence module, MLF Analysis is a dedicated layer, built on the latest AEMO tables (currently the May 2026 release). But its real value is in context. The platform reads MLF alongside:

  • Energy economics — negative-price hours, spot exposure and a five-year PPA projection for the same node.
  • Curtailment intelligence — whether the cheap energy near the site is actually deliverable.
  • Competitive density — who else is connecting nearby and may move your loss factor.

The result is not a raw decimal pulled from a spreadsheet. It is an MLF score that tells you whether a connection point is genuinely cheap to serve — and how durable that advantage is likely to be.

The takeaway

If you are evaluating sites for a large, energy-intensive load, treat MLF as a first-class variable, not a footnote:

  • Pull the current MLF for the actual connection point, not the region in general.
  • Look at the trend and the neighbours, not just this year's value.
  • Combine it with curtailment and price data before you trust it.

A few minutes spent understanding a loss factor can be worth millions over the life of a facility. In the NEM, the decimal place is where the money is.

Keep reading

All posts

Ready to make better infrastructure decisions?

Join the teams using Zyntax Arc to score sites, model NEM economics and pressure-test compliance before they commit capital.