We control our own energy supply, compute capacity and land pipeline. That combination is rare in Australia — and from 2027 it is increasingly the only one regulators will approve.
Each builds on the one before it, so customers can start on capacity that exists now and move into owned, renewable-powered facilities as they come online.
Provisioned accelerator capacity in established Australian facilities, available by the hour or the month. It is the fastest route to production AI compute here without waiting on a build — and the same team carries you into our owned capacity as it commissions.
Facilities designed for AI density from the ground up rather than retrofitted from cloud-era assumptions, on industrial land we have already secured. Sites target 50–100 MW of IT load each. Two Sydney projects — Illawong and Olympic Park — are in development now.
Solar and wind generation, grid-scale storage and the data centre on one site — lithium-ion for short-duration response, vanadium flow for the long-duration cycling that firms renewables overnight. Your compute runs behind the meter on power we own.
Compute and energy services across the three phases — the first available today, the rest as our owned capacity commissions.
| Offering | Phase | Availability | What it is |
|---|---|---|---|
| GPU compute | 1 | Available now | Hourly and monthly rental of provisioned, GPU-equipped racks |
| Colocation & compute lease | 2–3 | From commissioning | Long-term power and space leases for hyperscalers, AI labs and enterprise |
| Firmed renewable supply | 3 | In planning | Behind-the-meter generation and storage, firming your load around the clock |
| Grid services | 3 | In planning | Frequency control and ancillary services from our storage assets |
| Renewable certificates | 3 | In planning | Large-scale Generation Certificates, for customers reporting on scope 2 |
Facilities designed around the thermal and electrical profile of modern accelerators, not adapted to it after the fact.
40–100 kW per rack, against roughly 10 kW in a traditional facility — the envelope modern GPU training clusters actually draw.
Direct-to-chip and immersion plant to manage heat at density while holding PUE below 1.25 — a prerequisite for new-build AI capacity, not an optimisation.
Power and cooling modules deployed in tranches aligned to customer demand — avoiding over-build and preserving capital efficiency.
Co-locating AI data centres with owned renewable generation eliminates long-distance transmission losses and removes the single hardest constraint on delivery in the Australian market: grid connection risk.
Where a conventional operator negotiates for firm capacity on a congested network, a Phase 3 Tesseract DC campus arrives with its own generation and storage behind the meter — and contributes to network stability rather than competing for it.
Vanadium flow batteries offer near-infinite cycle life with no capacity fade, a non-flammable electrolyte, and full discharge capability — well suited to the deep daily cycling required to firm intermittent renewables for 24/7 AI compute. Upfront cost per kWh is currently higher than lithium, but the levelised cost of storage across a 20-year asset life is materially lower, which matches the horizon of the underlying infrastructure.
| Technology | Best for | Role in the portfolio |
|---|---|---|
| Lithium-ion BESS | Short-duration (1–4 hr) fast response | FCAS market participation, intra-day arbitrage, UPS-grade reliability for compute uptime |
| Vanadium flow battery | Long-duration (4–12 hr) cycling | Shifts renewable generation to overnight AI training loads, firms solar and wind output, deep-cycle longevity of 20,000+ cycles with minimal degradation |
From 2027, Australian planning authorities and grid operators are expected to approve AI data centre developments only where they are backed by renewable energy and storage. It is the single biggest factor in whose capacity actually gets built.
It is driven by National Electricity Market emissions targets, grid connection scarcity, and political pressure to ensure the AI build-out does not derail national decarbonisation commitments. Projects without credible renewable supply face protracted delays, connection refusals, or outright rejection — which is a supply problem for anyone buying compute.
Our projects are timed to reach approval and construction precisely as the threshold takes effect. Where others are re-planning around it, our pipeline was designed for it — so the capacity we commit to is capacity that can actually be permitted and built.
| Approval type | Description | Typical timeline |
|---|---|---|
| Development Application (DA) | Local and state council consent for industrial land use change to AIDC | 6–12 months |
| Grid connection (AEMO) | NEM connection agreement and network augmentation assessment | 12–24 months |
| Generation & storage licensing | Registration of solar, wind and BESS as scheduled or non-scheduled units | 6–18 months |
| AIDC operating permit | Environmental, water, noise and emissions compliance for continuous high-density operation | 6–12 months |
Most operators buy power and rent space. We own the generation, the storage, the land and the facility — which changes what we can commit to, and for how long.
Our campuses sit behind their own meter, on generation and storage we own. Capacity is not queued behind a congested grid connection, which is the single hardest constraint on delivery in Australia today.
We were renewable-first before the 2027 threshold made it mandatory. Our approvals move with the policy rather than against it — and your compute is carbon-aligned from the first rack, not offset afterwards.
40–100 kW per rack against roughly 10 kW in a conventional facility, with direct-to-chip and immersion cooling and PUE held below 1.25. No retrofit compromises, because nothing here was designed for cloud-era workloads.
Power is 40–60% of the operating cost of an AI facility. Owning the generation and arbitraging with storage means we can hold that cost steady through wholesale volatility instead of passing it to you.
Ten industrial sites identified and 4 GW of generation secured by option, MoU or acquisition pathway. If one approval or connection slips, the programme does not — your capacity is not tied to a single planning decision.
Onshore infrastructure under Australian ownership, for workloads with data residency, sovereignty or continuity requirements — backed by an investor group spanning Australian, Gulf and Asia-Pacific institutional capital.
Two projects are in development in Sydney, with further sites in planning across the ACT, Queensland, South Australia and Western Australia. Here is where our capacity is coming from, and when.