Nvidia and eight Australian partners say their sites could support up to 2 gigawatts of AI capacity by 2027. The announcement names the companies, but it does not give one full project schedule or energy plan.
New to this? Read it in simple words
- Nvidia and eight Australian partners shared a plan for more AI computing in Australia. Nvidia would supply the hardware, and local companies would run the sites.
- They say their sites could support up to 2 gigawatts of AI computing by 2027. This is a possible total, not capacity that is already running.
- Local access could help Australian universities, startups and companies keep sensitive work in Australia. It could also reduce delays when overseas systems are busy.
- The plan does not give a full schedule or energy plan. A gigawatt measures power, not useful AI work.
- Gigawatt
- A very large unit for measuring electric power.
- Data centre
- A building full of computers that run online services and AI.
The plan joins several kinds of company
Nvidia named cloud providers, data-centre operators, and a network company. The group includes Firmus, Sharon AI, IREN, Megaport, ResetData, CDC, NEXTDC, and AirTrunk. Each partner would provide a different part of the system.
The partners plan to use Nvidia DSX systems. Nvidia would supply computing hardware, networking, software, and technical support. The local companies would operate the sites and sell access to customers.
Nvidia says the combined buildout could reach 2 gigawatts by 2027. Reuters reported the same upper target. The words “up to” matter because they describe a possible total, not capacity that is already running.
The partnership says up to 2 gigawatts by 2027. Sites, grid connections, operating systems, customer access, and measured use still need proof.
Power is only the first limit
A gigawatt measures power, not useful AI work. Two sites with the same power limit can deliver different results. Chip choice, cooling, network speed, software, and how often the machines are busy all change the real output.
Large data centres also need grid connections, land, water or other cooling systems, and planning approval. The announcement does not give one complete schedule for every partner. It also does not explain the full energy mix for the combined target.
Local access could help Australian universities, startups, and companies keep sensitive work closer to home. It could also reduce delays when overseas systems are busy. Those benefits depend on price and fair access, not only the number of machines.
Watch the delivery, not the headline number
The first useful check is built capacity. Partners should report which sites are approved, connected, and open to customers. They should separate firm projects from early plans.
The second check is public impact. New electricity demand should not quietly raise costs or weaken reliability for nearby users. Clear energy contracts and grid studies can show who pays for new supply and network work.
The third check is customer value. A regional AI system matters when researchers and smaller companies can buy useful time at a fair price. A large target can still produce limited public value if most capacity has one private buyer.
Australia may gain an important local computing base from this partnership. For now, readers should treat 2 gigawatts as a ceiling. The stronger evidence will be working sites, clear energy plans, customer prices, and measured use.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1Primary source · September 9, 2026NVIDIA expands AI infrastructure capacity with Australia’s data center ecosystem Nvidia
- 2Research · September 9, 2026Nvidia teams up with Australian partners to build AI factory capacity Reuters via Investing.com
We used Nvidia’s announcement to identify the partners, platform, target, and planned users. We checked the target against Reuters. We call 2 gigawatts an upper goal because the announcement says “up to” and does not give one completed-project list.