We finance, develop, acquire and operate liquid-cooled, high-density AI campuses and the power and cooling infrastructure that supports them — blending contracted income with selective development-stage upside.
Since generative AI emerged, model parameter counts have moved from billions to trillions, and inference now accounts for the majority of global AI compute. NVIDIA's Blackwell architecture (B300 / GB300) has become the mainstream, with per-GPU power draw up to ~1,100W and rack densities beyond 140kW. Conventional air-cooled facilities cannot dissipate that heat — so the binding constraints are no longer chips alone, but land, firm power, grid connection and liquid cooling. All financeable, hard-asset exposures.
GPU clusters & AI servers (Blackwell B300 / GB300 class) financed on an asset-backed basis with conservative residuals.
Behind-the-meter and grid-firmed supply, including renewable generation and BESS firming.
Direct-to-chip and immersion plant, MEP fit-out and containerised modular compute units.
Land, shell, substation and grid connection — the core, long-duration Class A collateral.
Core allocation of 75–100% across Australasia and Southeast Asia, with an opportunistic sleeve of up to 25% to the United Arab Emirates alongside the Fund's UAE sponsor group.
At the centre of the Asia-Pacific with submarine-cable connectivity to North & Southeast Asia and North America. Data-localisation requirements across government, finance, healthcare and defence generate sticky, long-duration domestic compute demand, and there is unrestricted access to US accelerator supply chains.
Singapore anchors regional demand but is power-constrained, pushing capacity into adjacent Johor, Indonesia, Vietnam and Thailand. Sapien has operated offices in Singapore, Jakarta, Hanoi and Bangkok since 2020 — local origination, regulatory and partner relationships that are hard to replicate remotely.
Access, not diversification. The sponsor group's established position provides accelerated site, power and hardware allocation on terms not otherwise available — alongside the world's first national AI strategy and very large sovereign-fund commitments to AI infrastructure.
Recent capital commitments in Australia alone illustrate both the depth of demand and the availability of exit and co-investment counterparties.
| Company / Investor | Announced investment | Principal projects | Scale |
|---|---|---|---|
| Microsoft | A$25bn (≈US$18bn) | Azure AI infrastructure, Australia | +140% capacity to 2029 |
| Amazon Web Services | A$20bn | Sydney & Melbourne campuses | 40MW & 35.2MW new |
| Blackstone / AirTrunk | A$24bn | SYD3, MEL1, MEL2 | 400+ MW |
| A$1bn | Digital Future Initiative | — | |
| Equinix | A$240m | SY5 (Sydney) expansion | — |
Source: sponsor programme materials, compiled from public announcements. Figures are as announced by the relevant companies and have not been independently verified.