Sovereign-scale, purpose-built AI GPU data center infrastructure across Central & Eastern Europe.
560 MW
Phase 1 + Phase 2
147,456
Vera Rubin GPUs
2,048
NVL144 Racks
$4.3B
Revenue at Scale
5 GW
CEE Pipeline
ALT Infrastructure SA, Zurich · AGC Poland
EXECUTIVE SUMMARY — STARGARD, POLAND
01 — Investment Thesis
The Opportunity
While the world's hyperscalers race to spend $600B+ in 2026 chasing AI compute capacity they cannot build fast enough, ALTi Global is constructing what they need: sovereign-scale, purpose-built GPU infrastructure that 90% of the world's existing data centers physically cannot deliver. This is not a bet on AI. This is the infrastructure beneath the bet — the power, the cooling, the racks, the chips — at the exact moment when demand has outstripped supply by the widest margin in the history of computing.
The buildout is staged across two phases. Phase 1 delivers 140 MW and 36,864 NVIDIA Vera Rubin NVL144 GPUs in Stargard, Poland — operational Q1 2027, generating $712M in Year 1 revenue at 70% utilization. Phase 2 triples the footprint: three additional 140 MW facilities (420 MW) begin construction Q2 2027 on confirmed expansion land, adding 110,592 GPUs and driving combined platform revenue to $4.3B annually at steady state. The 5 GW CEE pipeline (2 GW Poland, 2 GW Romania, 1 GW Bulgaria) extends the platform far beyond these initial phases.
Three structural advantages make this possible: purpose-built Vera Rubin NVL144 architecture that competitors cannot retrofit for — each rack draws ~264 kW, requires direct-to-chip liquid cooling, and operates as a unified AI supercomputer delivering 3.6 EFLOPS; 40–60% cost advantage through Poland's Special Economic Zone, labor arbitrage, and competitive energy at ~130 USD/MWh; and full vertical integration via four strategic acquisitions controlling MEP engineering, architecture, energy generation, and cooling — compressing the build timeline to under 12 months versus the 18–30 month industry standard.
Sources: CreditSights Hyperscaler Capex (Nov 2025); NVIDIA CES 2026 (Jan 2026); ALTi Global financial model
Phase 1 — 140 MW, Stargard
140 MW
Power
36,864
GPUs
512
NVL144 Racks
$712M
Year 1 Revenue
$1.01B
Yr 3 Revenue
$4.27B
Investment
Phase 2 — 3 × 140 MW Expansion (Q2 2027 Start)
420 MW
Additional Power
110,592
Additional GPUs
1,536
Additional Racks
~$3.18B
Yr 4 Revenue (P2)
~$12.8B
Investment (P2)
<12 mo
Each Build
Combined Platform — Phase 1 + Phase 2
560 MW
Total Power
147,456
Total GPUs
2,048
Total Racks
$4.29B
Steady-State Rev
~$3.05B
Steady EBITDA
~$17.1B
Total Investment
28.3%
IRR (P1 Model)
12.2×
Equity MOIC (P1)
5 GW
CEE Pipeline
How Phase 2 Works
Phase 2 construction begins Q2 2027 — immediately after Phase 1 goes live — leveraging the same vertically integrated construction platform, the same NVIDIA NVL144 supply relationship, and the same Stargard Industrial Park expansion land (70+ hectares identified). Each of the three 140 MW builds replicates the Phase 1 template: identical rack configuration (512 NVL144 racks), identical power density (~264 kW/rack), identical CAPEX profile (~$4.27B per facility). The replicability is the point — vertical integration means ALTi Global doesn't re-procure contractors, re-negotiate power, or re-engineer cooling for each build. The template is proven, the land is available, the power consortium agreement covers up to 500 MW, and the construction team is in-house. Phase 2 facilities ramp to 90% utilization within 24 months of go-live, mirroring the Phase 1 70% → 80% → 90% trajectory validated in the financial model.
Source: ALTi Global three-statement financial model (Vera Rubin variant, Q1 2026). Phase 1: WACC 8%, 6% cost of debt, 10× exit multiple, 70%→80%→90% utilization. Phase 2 projections scaled from validated Phase 1 model with 1-year construction lag.
02 — Market Context
Demand Drivers & Structural Advantage
⚡ Compute Supply-Demand Gap
Hyperscaler capex is projected to exceed $600B in 2026, a 36% increase over 2025, with ~75% ($450B) tied directly to AI infrastructure. Goldman Sachs projects $1.15 trillion in cumulative hyperscaler capex from 2025–2027. Global data center infrastructure spending reached $290B in 2024 and is projected to surpass $1 trillion annually by 2030.
Sources: CreditSights (Nov 2025); Goldman Sachs Research (Dec 2025); IoT Analytics DC Equipment Report (Nov 2025); Futurum Group (Feb 2026)
🇪🇺 European Capacity Deficit
Europe is structurally underbuilt versus North America and APAC. The EU AI Act creates demand for sovereign compute. Poland is the cost-competitive EU entry point — EU membership, NATO membership, and significantly lower costs than Western alternatives. As of late 2025, Synergy Research counted 1,297 operational hyperscale data centers worldwide with 770 in pipeline.
Sources: Synergy Research Group (Q4 2025); Data Center Knowledge (Mar 2026)
🧱 The Retrofit Wall — ALTi Global's Core Insight
~90% of existing data centers cannot be retrofitted for Vera Rubin NVL144. Each rack is a unified AI supercomputer delivering 3.6 EFLOPS of NVFP4 inference with 72 Rubin GPUs and 72 Vera CPUs — requiring ~264 kW/rack, direct-to-chip liquid cooling, and NVLink 6 at 3.6 TB/s per GPU. Competitors built for H100/Blackwell face 12–18 month rebuilds. Microsoft has publicly abandoned sites that can't accommodate next-gen architectures. NVIDIA's development leadership has independently validated ALTi Global's strategy, architecture, and implementation plan.Sources: NVIDIA CES 2026 Keynote (Jan 6, 2026); StorageReview Vera Rubin NVL72 Deep Dive (Jan 7, 2026); Tom's Hardware (Jan 5, 2026); Microsoft Fairwater disclosure (Q4 2025)
◆ Vera Rubin NVL144 — Generational Leap
Launched CES 2026, in full production Q1 2026. 5× inference performance and 10× lower cost per token versus Blackwell. 1/4 the GPUs for MoE training. 288 GB HBM4 per GPU at 22 TB/s. 260 TB/s rack-level NVLink 6. Cable-free modular tray design = 18× faster assembly. TSMC 3nm, 336B transistors. Partner availability H2 2026.
Sources: NVIDIA Newsroom (Jan 6, 2026); VideoCardz NVL72 Specs (Jan 6, 2026); Supermicro PR (Jan 5, 2026); NVIDIA Technical Blog (Feb 2026)
03 — The Asset
Phase 1 Facility — Stargard, Poland (140 MW)
Phase 1 of a 560 MW platform. Three additional identical 140 MW facilities (Phase 2) begin construction Q2 2027 on the same campus.
Specification
Detail
Total Power Capacity
140 MW (signed commitment from consortium)
GPU Architecture
NVIDIA Vera Rubin NVL144 (72 Rubin GPUs + 72 Vera CPUs per rack)
Rack Count
512 NVL144 rack-scale AI supercomputers
Power per Rack
~264 kW (next-gen liquid-cooled spec)
Total GPU Count
36,864 Rubin GPUs
Rack Performance
3.6 EFLOPS inference / 2.5 EFLOPS training per rack
GPU Price per Rack
~$6.0M per NVL144 rack (after 20% volume discount)
NVIDIA Essentials + Mission Control cluster management
Data Halls
8 purpose-built, ultra-high-density AI halls per facility
Storage
BeeGFS high-IOPS, multi-tenant parallel file system
Resiliency
Tier III Uptime Institute design standard, N+1 redundancy
Compliance
ISO 27001 ready; EU data sovereignty compliant
Sources: ALTi Global financial model; NVIDIA Vera Rubin NVL144 specs (CES 2026); Barron's NVIDIA server pricing (Q4 2025); Wolfe Research NVIDIA price target (Q4 2025)
CAPEX Breakdown — $538.4M Validated
Work Package
Cost (USD)
Design & Engineering
$6.6M
Building Shell & Structure
$60.9M
Technical Infrastructure
$9.1M
Critical Infrastructure (Power, Cooling, UPS)
$375.4M
Office Facilities & Operations Center
$16.5M
External Works
$51.6M
Construction Management & Commissioning
$18.2M
Total Facility CAPEX
$538.4M
Phase 1 CAPEX excludes GPU procurement ($3.072B for 512 NVL144 racks at $6.0M each, 20% volume discount). Total Phase 1: ~$4.27B. Phase 2 replicates this template 3×, for ~$12.81B additional. Total platform investment: ~$17.1B across 2026–2029. Source: ALTi Global financial model; EPC contractor validated proposal
Construction Timeline — Phase 1 (140 MW)
Stage 1 · Through Q1 2026
Pre-Construction
Site acquisition, permitting, GPU ordering, power agreements finalized
Stage 2 · Q2–Q4 2026
Facility Build
Shell construction, HVAC, power infrastructure, BESS integration
Stage 3 · Q1–Q2 2027
Internal Fit-Out
NVL144 assembly, NVIDIA Mission Control config, burn-in, go-live
Construction Timeline — Phase 2 (3 × 140 MW)
Stage 4 · Q2 2027
Phase 2 Groundbreak
Three simultaneous 140 MW builds begin on Stargard expansion land. Leverages same vertically integrated construction team, same power consortium (500 MW pathway), same NVIDIA NVL144 supply chain.
Stage 5 · Q3 2027–Q1 2028
Parallel Build
Shell, MEP, power, and cooling across all three facilities. Replicable template eliminates re-engineering. In-house MEP, architecture, and cooling teams deploy proven Phase 1 design at 3× scale.
Stage 6 · Q2 2028
Phase 2 Go-Live
1,536 NVL144 racks installed across three facilities. Burn-in, Mission Control config, customer onboarding. First Phase 2 revenue: $2.136B (Year 1, 70% utilization).
04 — Site & Expansion
CEE Strategic Ranking — 5 GW Pipeline
Multi-country platform assessed across power, EU membership, SEZ incentives, connectivity, labor, stability, land, and climate.
Energy cost modeled at ~500 PLN/MWh (average Polish commercial rate), with nominal 112 MW usage and 90 MW real load. Vertical integration extends to energy generation through acquisition of renewable assets (behind-the-meter capacity). International, world-class Battery Energy Storage Systems provide peak load handling and grid stabilization. Poland's 2023 Energy Law amendment enables behind-the-meter generation.
Sources: ALTi Global financial model; gramwzielone.pl, TGE, Axpo Poland energy pricing (Q1 2025); Polish Energy Law 2023 amendment
06 — Business Model
Revenue Projections — From Financial Model
Wholesale GPU compute via fixed enterprise rack rentals at ~$3.15/GPU-hour (Year 1, with 5% annual escalation). 884,736 GPU-hours capacity per day across 36,864 GPUs. Phased utilization ramp: 70% Year 1, 80% Year 2, 90% Year 3+.
Source: ALTi Global three-statement financial model (Vera Rubin variant). GPU-hour pricing benchmarked against current H100 wholesale rates ($2–5/hr, Q4 2025).
Phase 1 — Projected P&L (140 MW, 512 Racks)
2026
2027
2028
2029
2030
2031
GPU Utilization
—
70%
80%
90%
90%
90%
Revenue
$0
$712M
$854M
$1.009B
$1.060B
$1.113B
EBITDA
($3.6M)
$494M
$603M
$718M
$753M
$790M
EBITDA Margin
—
69.4%
70.6%
71.1%
71.1%
71.0%
Net Profit
($3.6M)
($331M)
($61M)
$158M
$281M
$392M
Source: ALTi Global financial model — Analysis sheet. Utilization: 70%→80%→90%. Energy at ~130 USD/MWh. 5% annual price escalation.
Phase 2 = 3 × Phase 1 financial model, lagged by 1 year. Same GPU pricing, utilization ramp (70%→80%→90%), cost structure, and EBITDA margins. Construction Q2 2027, first revenue Q2 2028.
Combined Platform — Phase 1 + Phase 2
2027
2028
2029
2030
2031
Phase 1 Revenue
$712M
$854M
$1.009B
$1.060B
$1.113B
Phase 2 Revenue (3× P1)
—
$2.136B
$2.562B
$3.027B
$3.180B
Combined Revenue
$712M
$2.990B
$3.571B
$4.087B
$4.293B
Phase 1 EBITDA
$494M
$603M
$718M
$753M
$790M
Phase 2 EBITDA (3× P1)
—
$1.482B
$1.809B
$2.154B
$2.259B
Combined EBITDA
$494M
$2.085B
$2.527B
$2.907B
$3.049B
Combined Net Profit
($331M)
($1.054B)
($25M)
$755M
$1.235B
Phase 2 = exactly 3× Phase 1 financial model, lagged 1 year. Same pricing, utilization ramp, cost structure. Phase 2 construction begins Q2 2027 on confirmed expansion land within 500 MW consortium pathway. First Phase 2 revenue Q2 2028.
Customer Pipeline — Four Target Segments
U.S. Hyperscalers
Wholesale GPU Capacity
Active negotiations with major hyperscale cloud providers seeking dedicated GPU capacity in the European market. EU data residency requirements and sovereign compute mandates are forcing hyperscalers to secure purpose-built European infrastructure they cannot build fast enough internally. ALTi Global's Vera Rubin NVL144 architecture positions the platform as one of the first European-based next-gen GPU deployments, directly addressing $80B+ in unfulfilled Azure backlog alone.
AI Scale-Ups
Dedicated Training Clusters
High-growth AI companies — including frontier model labs, enterprise AI platforms, and vertical AI specialists — requiring dedicated NVL144 clusters for model training at scale. These companies face hyperscaler pricing constraints ($3-5/GPU-hour for H100) and 6+ month procurement queues. ALTi Global offers 30-50% cost savings, guaranteed capacity, no platform lock-in, and purpose-built Vera Rubin architecture unavailable from incumbents.
European Multinationals
Sovereign AI & Compliance
European-centric multinational corporations across financial services, pharmaceuticals, automotive, and industrial sectors facing dual pressure: the EU AI Act mandating data sovereignty for AI workloads, and competitive urgency to deploy proprietary AI models. These enterprises require European-hosted, ISO 27001-ready infrastructure with full GDPR compliance, Tier III resiliency, and no dependency on US-headquartered cloud providers.
Middle Eastern Sovereigns
Strategic Backup & Resilience
Sovereign entities across the Gulf seeking strategic compute capacity through direct infrastructure partnerships, with emphasis on geographic diversification and backup workload resilience. Given the current geopolitical situation in the Middle East, sovereign compute programs (Saudi NEOM, UAE G42/MGX, Qatar digital mandates) are increasingly prioritizing European-based backup infrastructure for mission-critical AI workloads — creating demand for NATO/EU-member-hosted facilities with bilateral treaty protections.
07 — Corporate Structure
Governance & Vertical Integration
Swiss HoldCo
ALT Infrastructure SA, Zurich
GPU assets as collateral. Revenue via equipment leasing to Polish OpCo. Swiss governance, EU capital markets access.
Polish OpCo
AGC Poland
Operations, customer contracts, employees, facility management. SEZ (up to 40% CapEx offset), EU regulatory compliance.
Vertical Integration — 4 Strategic Acquisitions
Acquisition
Strategic Rationale
Status
Acquisition A
Mechanical, electrical, and plumbing (MEP) experts; critical execution capability for high-density data center infrastructure
Closing 30d
Acquisition B
Architecture firm specializing in commercial, industrial, and high-performance compute facility design
Closing 30d
Acquisition C
Renewable energy assets providing behind-the-meter generation and long-term energy cost control
Closing 30d
Acquisition D
Specialized cooling engineering for ultra-high-density GPU deployments; NVL144-class thermal management
Closing 30d
Board & Leadership
Michel Combes — Chairman: Former CEO, SoftBank International. Institutional credibility, SWF network, telecom/infrastructure relationships.
Jan Sędęk — Founder & CEO: 19+ years industrial electrical infrastructure. Direct NVIDIA relationships. Large-scale infrastructure across CEE.
Hunter Lee Soik — Founder & Vice Chairman: Capital strategy & institutional investor pipeline. SWF/government partnerships across USA, UAE, Europe.
Marek Pawletko — Founder & Board Member: Board Member of top-five Polish insurance consortium. Governance and strategic oversight.
SHA completed by Baker McKenzie (Zurich). Big Four audit engagement in process. 5-year lock-up, bad leaver at 50% FMV, 1–2 Series A board seats.
08 — Financial Overview
Capital Structure, Returns & Valuation
Capital Layer
Amount
Status
Founder Capital
$7M
Invested
Seed Round
$64M
Closing within 30 days
Equity Investment (Phase 1)
$1.2B
From financial model
GPU Financing (Debt/Lease)
$3.072B
6% cost of debt, 5-yr term
Total Phase 1 Investment
$4.272B
Equity + GPU Finance
Phase 2: 3× Facility Replication
~$12.81B
3 × $4.27B (identical template)
Total Platform Investment (P1 + P2)
~$17.1B
Staged over 2026–2029
Source: ALTi Global financial model — DCF Analysis & Leasing sheets. GPU equipment: 512 NVL144 racks × $6.0M = $3.072B (20% volume discount from nominal).
Phase 1 Investment Returns — From Financial Model
28.3%
Unlevered IRR
53.7%
Levered IRR
2.96×
MOIC (Total)
12.17×
MOIC (Equity)
$6.53B
P1 DCF EV
$7.90B
P1 Exit (10× EBITDA)
Source: ALTi Global financial model — IRR Analysis, DCF Analysis, Investment Metrics sheets. WACC 8%, terminal growth 2%, 10× EV/EBITDA exit. 5-year hold period (2026–2031). Phase 2 returns expected to replicate Phase 1 IRR profile given identical template and cost structure.
ALTi Global — Implied Valuation Range
EV/Revenue Multiple
Phase 1 Only (Yr 3: $1.01B rev)
Phase 1 + Phase 2 (Steady: $4.29B rev)
Methodology
8.0× (Conservative)
$8.1B
$34.3B
Traditional DC REIT range
11.6× (Wtd. Avg. Comps)
$11.7B
$49.8B
Weighted avg. of 10 comps
15.0× (GPU Cloud Premium)
$15.1B
$64.4B
CoreWeave / Aligned tier
Phase 1 DCF: $6.53B (WACC 8%, 10× exit). Phase 2 multiples applied to combined steady-state revenue ($4.29B). Pre-revenue assets trade at significant discounts. Total platform investment: ~$17.1B.
SWFs deployed $66B+ into AI & digital infrastructure in 2025. The hyperscaler capex supercycle — projected to exceed $600B in 2026 — is creating unprecedented demand for co-investment capital from sovereign and institutional allocators, reshaping how AI infrastructure gets financed globally. Sources: Global SWF Annual Report (Q4 2025); CreditSights Hyperscaler Capex Report (Nov 2025)
MGX
🇦🇪
MGX Mubadala-backed
AI Investment Vehicle — Abu Dhabi
Co-led $40B Aligned DC deal with BlackRock/GIP. Anchor investor in Stargate ($500B) alongside OpenAI & SoftBank. Raising $25B for dedicated AI infrastructure fund. Key member of the AI Infrastructure Partnership (AIP). CNBC (Dec 2025)
MIC
🇦🇪
Mubadala $302B AUM
Sovereign Wealth Fund — Abu Dhabi
Record $32.7B deployed in 2025 across 40 transactions — most active SWF globally. $12.9B directed to AI & digital infrastructure. Parent entity of MGX. Strategic partner to BlackRock. Bloomberg (Q4 2025)
PIF
🇸🇦
Saudi PIF $930B AUM
Sovereign Wealth Fund — Saudi Arabia
$36.2B deployed in 2025 — largest SWF dealmaker globally. AI is a priority mandate through NEOM and advanced technology investments. Saudi Arabia introducing DC investment frameworks for foreign providers. Global SWF (Q4 2025)
QIA
🇶🇦
QIA $524B AUM
Sovereign Wealth Fund — Qatar
Anchored Blue Owl's $3B+ DC platform with $1B initial equity commitment. Investor in Anthropic Series F ($13B round). Committed $500B to US over next decade with AI & data centers as priority sectors. SEC Filings (Q4 2025)
ADIA
🇦🇪
ADIA $1T AUM
Sovereign Wealth Fund — Abu Dhabi
World's 3rd largest SWF. Active in DC sector, backed Thoma Bravo's $12B Dayforce acquisition. Co-investor in Blackstone/Hologic $18.3B deal. Increasingly deploying directly into digital infrastructure and AI-adjacent assets. Bloomberg (Q4 2025)
KIA
🇰🇼
KIA $969B AUM
Sovereign Wealth Fund — Kuwait
$6B deployed into AI & digitalization in 2025 — second-largest Middle East SWF allocation to the sector after Mubadala. Growing infrastructure and technology mandate. Global SWF (Q4 2025)
ADQ
🇦🇪
ADQ / XRG $200B+ AUM
Sovereign Holding & Energy — Abu Dhabi
ADQ assets doubled in 4 years, investing in critical minerals for AI. XRG (ADNOC's intl. arm) focused on energy systems powering data centers, bridging hydrocarbon capital with AI infrastructure demand. Reuters (Q4 2025)
DI
🇮🇩
Danantara $900B AUM
Sovereign Wealth Fund — Indonesia (est. 2025)
Indonesia's newest SWF, 7th largest globally. AI and digital ecosystem designated as priority investment areas. Sovereign AI Fund planned for 2027–2029. $4B co-investment partnership with Qatar announced. Asia House (Sep 2025); GovInsider (2025)
CPP
🇨🇦
CPP Investments C$675B AUM
Pension Fund — Canada
Canada's largest pension fund. Active infrastructure investor with growing digital and data center portfolio. Long-duration capital aligned with AI infrastructure build cycle. CPP Investments AR (Q4 2025)
CAN
🇨🇦
CDPQ / OTPP / BCI C$1T+ combined
Canadian Pension Alliance
CDPQ (~C$473B), Ontario Teachers' (~C$266B), and British Columbia Investment (~C$250B) collectively represent massive infrastructure capital. CDPQ invested in European digital infrastructure; OTPP active in data center JVs. CDPQ AR (2025)
🇺🇸
BlackRock / GIP $11.5T AUM
Asset Manager — New York
Led $40B Aligned DC acquisition (largest DC deal ever). AI Infrastructure Partnership (AIP) targeting $30B equity / $100B total deployment. GIP manages ~$170B, scaling to $210B by 2026. CNBC (Dec 2025)
🇺🇸
Blackstone $1.1T AUM
Private Equity — New York
Acquired QTS for $10B (2021) and AirTrunk for A$24B (2024). Aggressively expanding both portfolios. The dominant PE player in global data center ownership with billions in additional committed capital. Bloomberg (Q4 2025)
🇨🇦
Brookfield $1T AUM
Alt. Asset Manager — Toronto
Major DC developer in Europe and Americas. Power and renewables portfolio critical for energy-for-AI. Developing gigawatt-scale campuses including facilities in Sweden (Strängnäs) and North America. Brookfield IR (Q4 2025)
🇺🇸
KKR $637B AUM
Private Equity — New York
Growing digital infrastructure portfolio with active Middle East franchise. DC investments increasingly co-invested with SWF capital. Targeting infrastructure-grade AI assets across US and Europe. KKR AI Infrastructure Report (2025)
🇸🇪
EQT €130B AUM
Private Equity — Stockholm
Leading European PE firm with dedicated infrastructure strategy. Active in power generation, cooling systems, and secondary DC infrastructure across Europe. Growing AI-adjacent portfolio through EQT Infrastructure funds. EQT Group (Q4 2025)
🇫🇷
Amundi €2.2T AUM
Asset Manager — Paris
Europe's largest asset manager. Expanding real assets and infrastructure allocation with growing focus on digital infrastructure and energy transition assets that support AI data center buildout. Amundi AR (2025)
🇦🇺
Macquarie $600B+ AUM
Infrastructure Manager — Sydney
Pioneered DC infrastructure investing. Sold Aligned ($40B) and AirTrunk (A$24B) — two of the largest DC exits ever. Deep global digital infrastructure portfolio spanning towers, fiber, and colocation. Macquarie IR (2025)
🇺🇸
Microsoft / NVIDIA Strategic
Corporate AI Infrastructure Investors
Both AIP members. Microsoft investing $120B+ in FY2026 AI DC capex. NVIDIA building own AI Factory Research Center in Manassas, VA for Vera Rubin reference designs. Combined strategic influence shapes the entire AI infrastructure market. Futurum Group (Feb 2026); NVIDIA Newsroom (Jan 2026)
🇺🇸
Goldman Sachs / TPG $334B dry powder
PE & Infrastructure Funds
Active across the DC supply chain. Goldman sold Boyd Thermal to Eaton ($9.5B). TPG investing in power generation and secondary DC infrastructure. $334B in infrastructure-focused dry powder as of end-2025. Goldman Sachs Research (Dec 2025)
GIC
🇸🇬
GIC $770B+ AUM
Sovereign Wealth Fund — Singapore
Co-investor in Anthropic and Blackstone/Hologic deal. Deep portfolio in digital infrastructure, global DC assets, and semiconductor supply chain. Long-duration capital ideal for AI infrastructure build cycle. GIC AR (2025)
TEM
🇸🇬
Temasek $389B AUM
Sovereign Investment — Singapore
Owns ST Telemedia Global Data Centres with facilities across APAC. Active investor in AI companies, semiconductor firms, and digital infrastructure platforms. Strategic bridge between APAC and global AI capital flows. Temasek AR (2025)
DI
🇮🇩
Danantara $900B AUM
Sovereign Wealth Fund — Indonesia (est. 2025)
7th largest SWF globally. AI & digital ecosystem designated as priority investment areas. Sovereign AI Fund planned for 2027–2029 period. $4B co-investment partnership with Qatar. Oracle partnership confirmed for AI data sovereignty. Asia House (Sep 2025)
NPS
🇰🇷
NPS $800B AUM
National Pension Service — South Korea
World's 3rd largest pension fund. Expanding alternative investments allocation with growing digital infrastructure mandate. Active in global infrastructure co-investments alongside PE sponsors. Reuters (Q4 2025)
KNB
🇲🇾
Khazanah Nasional $30B AUM
Sovereign Wealth Fund — Malaysia
Malaysian SWF with growing digital infrastructure portfolio. Strategic investor in Southeast Asian technology and data center assets. Partner in regional sovereign compute initiatives. Khazanah AR (2025)
AT
🇦🇺
AirTrunk / Blackstone A$24B platform
APAC Data Center Platform
Blackstone's A$24B AirTrunk acquisition created the dominant APAC DC platform. Expanding hyperscale capacity across Australia, Japan, Singapore, and Malaysia. Validates APAC demand for sovereign-adjacent infrastructure. Bloomberg (Q4 2025)
The following materials are available or in preparation for qualified investor review upon execution of a non-disclosure agreement.
Corporate & Legal
Swiss HoldCo formation documents (ALT Infrastructure SA, Zurich); Polish OpCo registration (AGC Poland); Shareholders' Agreement (Baker McKenzie, Zurich — complete); Investment Agreement template; board resolutions; Big Four audit engagement documentation.
Financial
Three-statement financial model (scenario-based, U.S./international SWF standard); five-year projections with DCF, IRR, MOIC analysis; capital structure and use of funds; comparable transaction analysis; GPU lease terms and financing documentation.
Technical
EPC contractor CAPEX proposal (full 46-page scope of works across 7 work packages); facility design specifications; NVIDIA Vera Rubin NVL144 architecture documentation; cooling and power distribution schematics; Tier III design certification.
Site & Energy
Power consortium allocation documentation (140 MW confirmed, 500 MW pathway); site plans and surveys; SEZ confirmation and tax incentive calculations ($30M+ identified); BESS integration specifications; behind-the-meter renewable energy documentation.
Commercial
Customer pipeline summary with engagement status; market analysis and demand projections; competitive landscape assessment; pricing methodology and GPU-hour benchmarking; off-taker strategy documentation.
Team
Full biographies and reference contacts; organizational chart; advisory board composition; key partnership agreements (NVIDIA, Delta, EU Networks, Baker McKenzie); vertical integration acquisition documentation.