Technical study · Technical Advisory & Assurance

Power Strategy for a 50 MW Data Centre Campus

How a 50 MW data centre campus can give tenants capacity three years before full grid supply arrives, with bridging power, battery-led backup and an hourly-matched renewable supply.

Aerial view of a data centre building surrounded by fields

What the client needed

A developer was planning a 50 MW (IT load) data centre campus in South East Queensland, with tenants wanting capacity from 2028. The network could offer 10 MW straight away, 25 MW from 2029, and the full 70 MW only after a new substation in 2031.

The question they needed answered: How do we deliver capacity on the tenants' timeline when the grid can't, and how do we meet their renewable energy and backup requirements?

  • 3 yearsearlier capacity than waiting for the grid
  • ~3 : 1lease revenue against the extra cost of bridging power
  • ~85%of load matched hour by hour with renewables and storage

What we did

  • Built the campus demand ramp (IT load and PUE) against firm grid capacity, year by year.
  • Assessed bridging power options: modular gas engines, battery storage and staged grid supply.
  • Compared backup strategies: conventional diesel generators, generators on renewable diesel, and battery-led backup with less generator capacity.
  • Developed a renewable supply strategy and tested how well different PPA mixes match campus load hour by hour.

What we found

Grid capacity, not land or buildings, sets the timing. AEMO's 2026 forecasts have data centre consumption in the NEM rising from about 5 TWh to around 34 TWh by 2035-36.

Bridging power costs perhaps $100/MWh more than grid supply, around $7 to $8 million a year on 10 MW of average load. Against that, 10 MW of leased AI capacity can earn around $24 million a year.

The campus needs up to 17 MW of bridging power from 2028 to 2030. Leased modular gas engines plus a 10 MW / 20 MWh battery cover the gap, and the battery stays on for backup, power quality and grid services when the substation arrives.

Battery-led backup lets the diesel generators be sized for extended outages only, and they can run on renewable diesel.

A solar-only PPA matches about a third of campus load hour by hour. Adding wind lifts that to about two thirds, and adding storage to around 85%.

Bar chart of campus demand from 2027 to 2033 against firm grid capacity, showing the bridging power needed
Campus demand ramp against firm grid capacity. The green segments are met by bridging power until the new substation arrives.

What we recommended

  • Sign the staged connection offer now and apply for the full capacity to hold the queue position.
  • Lease bridging power as a service with a firm exit date, so it can't become a stranded asset.
  • Design the battery as a permanent asset from day one, with the connection and controls to earn revenue once bridging ends.
  • Build the renewable portfolio around hourly matching, not annual totals.

Key assumptions

PUE 1.25. IT load ramp based on tenant pre-commitments. Grid capacity dates are indicative. Hourly matching based on typical Queensland solar and wind profiles. References: AEMO 2026 ESOO and ISP.

Related work

Talk to our engineers

Tell us what you're working on, and one of the people below will get back to you.

Talk directly to

  • Andrew LelliottOperations Director
  • Alistair WardropeTechnical Director

Want to join us? We're always keen to hear from engineers who value strong fundamentals and real delivery experience.

Careers at Amarna