Machinepower Index / Dossiers / Australia
Rank 16–19 across the scoring rules tested. How this is tested
Australia
Specialist Power — For nations among the world's best at something the frontier needs but could find elsewhere: keep earning the lead before others catch up, and trade expertise for access.
How the Score, Potential and bottleneck cost are worked out: methodology.
The three layers
Energy, grids, data centres and silicon — the physical capacity to run models at scale.
Compute tracked by Epoch AI: Australia controls 47,000 H100-equivalents, wherever the sites are. Hosted compute is shown, not scored.
Models, alignment capability, standards influence and the talent that produces them.
The capacity to wire AI into the state and the economy while holding public consent.
Where the score comes from
| Layer | Sub-dimension | Basis | Score |
|---|---|---|---|
| Watts | Planning & Permitting | Assessed | 6.4 |
| Firm Power Capacity | Dataset | 5.4 | |
| Sovereign Compute | Dataset | 5.2 | |
| Hardware & Silicon Access | Dataset | 4.9 | |
| Weights | Frontier Model Output | Dataset | 3.0 |
| Alignment Capability | Assessed | 7.6 | |
| Regulatory Standing | Dataset | 7.9 | |
| Talent Density | Dataset | 6.1 | |
| Will | Policy Capacity | Dataset | 10.0 |
| Public Sector Adoption | Dataset | 8.3 | |
| Workforce Transition | Dataset | 8.5 | |
| Public Trust | Dataset | 5.5 |
Assessed marks a measurement we judged rather than took from a dataset: 2 of the 12 here.
Sources
Epoch AI records about 47,000 H100-equivalents owned by Australian firms, chiefly Firmus. Because that exceeds what TOP500 shows, the Index raises Sovereign Compute to the level Epoch supports.
View source →Share of adults agreeing that products and services using AI have more benefits than drawbacks; the Index scores public trust as 1 + share ÷ 10. Survey of 23,532 adults across 32 countries, March to April 2026.
View source →Australia's National AI Plan (December 2025) sets three goals — capture the opportunity, spread the benefits, keep Australians safe — with a $29.9M AI Safety Institute launched in early 2026.
View source →Australia fields two systems on the June 2026 TOP500 — Setonix-GPU at Pawsey (#81) and Virga at CSIRO (#140, 14.94 PFlop/s). Foundational cluster capacity relies on commercial hyperscalers.
View source →The first AUKUS Pillar II signature project — UUV sensors, payloads and weapons — was announced mid-2026 with delivery from 2027, deepening allied technology access.
View source →Used for the workforce measurement. The IMF combines schooling, pay, labour mobility and active labour-market policies; the Index rescales it so the weakest of 173 economies scores 0 and the strongest 10.
View source →A cross-check for alignment, which stays assessed. The Research Almanac counts English-language research only, so it understates Japanese, Korean and Chinese work.
View source →Fossil, nuclear, hydro and other dispatchable plant; averaged with generation for firm power.
View source →Second source for talent, averaged with Stanford's count.
View source →Second source for public trust, averaged with Ipsos.
View source →Capital is shown beside the Score, not scored.
View source →No survey covers every nation, so consent to build data centres is shown, not scored.
View source →Bottlenecks
These hold Australia below its Potential. Its two lowest scores cost it the most: frontier model output at 3.0 (part of Weights) and hardware & silicon access at 4.9 (part of Watts).
Assessment
Australia is the West's plan B rather than a chokepoint of its own. Lynas is the largest producer of separated rare earths outside China, but that is around a tenth of world supply against China's two thirds, the separation happens in Malaysia, and Canberra has chosen to be a supplier of choice rather than to restrict anything. Around it sits AUKUS-grade trust, vast solar-and-land potential for future compute, and the familiar gaps: no fabrication, a thin frontier, and a wary public (47%).
The art of the possible
Australia's play is depth in a narrow layer, not a gate. Scale rare-earth separation and magnet production onshore before substitutes emerge — Eneabba from 2028 is the test — and stack trusted hosting on top: solar-powered, AUKUS-secured compute for allies who cannot put it anywhere else. The trade is minerals and trusted geography exchanged for frontier access and a security umbrella, negotiated while demand is high. If separation moves onto Australian soil at scale and Canberra takes offtake rights with it, the position hardens into something closer to a chokepoint. It is not one yet.
Strategic play — Specialist Power
Go deep in the one layer where it is the West's alternative to China: scale rare-earth separation and magnet production onshore before substitutes arrive, with Eneabba from 2028 as the test, and offer solar-powered, AUKUS-secured hosting alongside it, traded together for frontier access and a security umbrella.
Leverage — the nearest position. The West's materials floor: the largest separated rare earths outside China, plus lithium at scale. Cut from the chokepoint register because dominance still belongs to Beijing — but the position deepens every time Beijing squeezes.
Open Australia in the live Index → — every source behind every number, and the comparison tool.