Princeton Digital Group

1.1 Identity, business model, and scale

Princeton Digital Group (PDG) develops and operates internet infrastructure and data centers for hyperscalers and enterprises in Asian digital economies. Stonepeak's July 2025 announcement identifies Singapore as headquarters and lists operations in Singapore, Japan, India, Indonesia, China, and Malaysia; its description says the portfolio supports hyperscalers and enterprises. The same announcement reports a portfolio of more than 1.1 gigawatts across six countries. DC Atlas separately tracked 23 facilities across seven countries and more than 450 megawatts of live capacity, while flagging 14% of that live capacity as estimated. The two measures use different portfolio and operating-status boundaries, so they should not be treated as directly comparable. Open Compute's SG1 profile adds a concrete example of the model: carrier- and cloud-neutral wholesale colocation, build-to-suit, on-demand data-center, connectivity, and technical services. PDG's identity and regional footprint are therefore well established, but current customer count, headcount, revenue/run-rate, occupancy, and utilization are not publicly supported in this pool.[CO001, CO002, CO003, CO004, CO011]

1.2 Leadership, sponsors, and financing

Rangu Salgame is identified by Stonepeak and Mingtiandi as PDG's co-founder, chairman, and chief executive officer. W Media's syndicated company announcement identifies Varoon Raghavan as co-founder and chief operating officer in connection with the JC3 campus. The evidence confirms senior operating leadership, but it does not provide a complete executive roster, board composition, voting rights, or succession plan. PDG's capital position is clearer: Stonepeak announced a USD 1.3 billion preferred-equity investment in July 2025, following USD 1.2 billion of debt financing, which the company described as USD 2.5 billion raised during 2025 across equity and debt. Mingtiandi reports a Mubadala-led USD 505 million equity raise in 2022, while Warburg Pincus remained the largest shareholder after the Stonepeak transaction. These figures describe financing raised, not cash on hand, revenue, or enterprise value; the reported possible USD 4 billion valuation was a transaction discussion rather than a completed valuation disclosure.[CO005, CO006, CO007, CO008, CO009, CO012, CO016]

1.3 Milestones, execution, and diligence gaps

The milestone record shows a shift from regional platform formation to AI-oriented capacity expansion. Mingtiandi dates the company to 2017 and reports institutional financing milestones, including the Mubadala-led 2022 equity raise. W Media's syndicated company announcement says PDG broke ground in November 2025 on a USD 1 billion, 120-megawatt AI-ready JC3 campus in Greater Jakarta, with first-phase service expected in the fourth quarter of 2026 and total Indonesian capacity expected to reach 230 megawatts. Asian Morning and Yahoo's reproduced GuruFocus article describe a March 2026 plan to raise as much as USD 5 billion of debt, but those accounts are not equivalent to completed financing. The evidence also contains a concrete adverse diligence issue: the financing plan implies significant leverage and asset-level execution exposure, while public sources do not disclose net leverage, covenants, contract tenor, customer concentration, realized revenue, or incident history. The company has meaningful institutional backing and a visible delivery pipeline, but the economic and governance terms needed for underwriting remain private.[CO013, CO014, CO015, CO016, CO017]

1.4 Exhibits

2.1 Market boundary and customer job

The relevant market is physical digital infrastructure: data-center development, wholesale colocation, build-to-suit hyperscale capacity, connectivity, and operating services in APAC digital economies. PDG's OCP facility listing describes both hyperscale and wholesale colocation, alongside build-to-suit, on-demand colocation, technical services, connectivity, and workplace services. This boundary excludes cloud software, semiconductors, and end-customer applications, while retaining the power, land, cooling, fiber, and operations required to deliver contracted capacity. The buyer is typically a hyperscaler or enterprise; the user is its cloud or compute platform; and the economic payer is the infrastructure budget owner under a long-term capacity agreement. Status quo substitutes include self-built campuses, alternative colocation operators, carrier facilities, and delaying deployment. PDG's advantage is local execution across multiple jurisdictions, but that advantage is exposed to power availability, permitting, construction, and talent.[CM001, CM002, CM003, CM004]

2.2 Multiple sizing lenses and capital intensity

The market is large, but the available estimates describe different populations and should not be added together. Deloitte estimates approximately $800 billion of APAC data-center investment through 2030, while its electricity lens says regional data-center demand could expand up to five-fold by the mid-2030s. Those figures describe regional investment and electricity demand rather than PDG revenue or addressable capacity. PDG-specific disclosures provide a narrower operating lens: Ontario Teachers' reports a portfolio of 1 gigawatt built in less than seven years, and the same source says the company owns 21 data centers in six countries. A market-size bridge from APAC investment to PDG SOM cannot be calculated without country-level capacity, contracted-versus-available MW, pricing, utilization, capex, and financing terms. The $1.3 billion Stonepeak preferred-equity investment is financing, not market revenue or cash available for deployment.[CM005, CM006, CM007, CM008, CM009]

2.3 Adoption drivers, buyers, and constraints

Cloud adoption, AI workloads, data localization, and digital transformation support demand for regional capacity. PDG's Tokyo, Mumbai, and Johor projects were described as together exceeding 400 MW, while W.Media's account of PDG's company release describes Jakarta JC3 as a 120 MW AI-ready campus with dual-grid feeds and contracted full power capacity. These are deployment indicators, not proof of customer take-up. ISI Markets attributes regional demand to cloud, AI, and mobile connectivity, while DC Pulse's project directory shows the breadth and mixed maturity of PDG's footprint. Adoption depends on a buyer's need for speed-to-market, reliable power, connectivity, and a trusted operator across jurisdictions. Constraints are equally material: Deloitte identifies grid congestion and price volatility risks, and PDG's CEO says delivery delays and downtime are unacceptable for mission-critical customers. Power procurement, renewable availability, land, permits, modular construction, direct-to-chip cooling, and specialized talent can therefore shift both timing and returns. Public evidence does not disclose customer concentration, contract tenor, renewal rates, price per MW, or campus-level occupancy, leaving the adoption-to-revenue bridge unresolved.[CM010, CM011, CM012, CM013, CM014, CM015, CM016]

2.4 Sizing and adoption diligence gaps

The strongest conclusion is directional rather than a precise TAM/SAM/SOM estimate: APAC digital-infrastructure demand is expanding, and PDG has assembled a multi-country power and capacity platform, but public sources do not establish its share of regional demand or the economics of turning planned MW into revenue. Investor materials and trade coverage emphasize strategic growth and financing, whereas facility directories provide partial technical snapshots. The key diligence work is to reconcile the project pipeline with energized and contracted MW, tenant commitments, utilization, pricing, power costs, capex per delivered MW, debt service, and renewable-power obligations. Until that evidence is available, valuation should be tested against scenario ranges for build timing, occupancy, pricing, and financing rather than a single market multiple.[CM005, CM008, CM009, CM014, CM016]

2.5 Exhibits

3.1 Competitive landscape and buyer alternatives

The relevant competitive set is broader than other branded operators. PDG sells hyperscale and wholesale-colocation capacity, build-to-suit delivery, colocation, technical services and connectivity solutions at SG1, while the APAC market also includes hyperscaler self-build, enterprise on-premise capacity, carrier-neutral specialists, specialist AI or neocloud providers and the status quo of delaying or reallocating workloads. PDG's disclosed footprint spans China, Singapore, India, Malaysia, Indonesia and Japan, and Baxtel carries a company description of more than 20 data centres across those six economies. DC Atlas independently tracks 23 facilities across seven countries and more than 450 MW of live capacity, while labeling 14% of that live capacity as estimated. These scale and coverage signals use different bases and do not establish that PDG wins on realized price, occupancy, contract tenor or customer concentration.[CP001, CP002, CP003, CP004, CP006]

3.2 Capability, trust and distribution comparison

The strongest documented buying criteria are power and delivery certainty, technical flexibility, connectivity, security controls, certifications and the ability to serve customers across markets. SG1 lists two diverse redundant 22kV supplies, N+1 generators and distributed-redundant UPS alongside SOC1 Type 2, SOC2 Type 2 and ISO certifications. Frost & Sullivan's paid release says PDG works with customers on capacity planning and adapts technical requirements to local market conditions, while CBRE and JLL identify power availability, construction cost and grid connection delays as structural constraints. These sources support a plausible execution and access proposition, not a verified peer benchmark; competitor capabilities and service-level outcomes remain mostly unreported here.[CP004, CP005, CP007, CP011, CP015]

3.3 Moat durability, switching and displacement risk

PDG's potential moat is cumulative rather than a single proprietary feature: land and power positions, multi-country operating experience, financing relationships, repeatable campus delivery and customer-specific technical adaptation. Stonepeak's 2025 announcement confirms a USD 1.3 billion preferred-equity investment and continued support from Warburg Pincus, Ontario Teachers' Pension Plan and Mubadala, while the announced 120 MW JC3 development illustrates continued investment in Indonesia. The counterweight is equally important. JLL reports that 78% of 4.8 GW of APAC supply expected by 2027 is already preleased, Mingtiandi reports that Digital Edge raised more than USD 1.6 billion, and CBRE identifies neocloud growth and rising build constraints; well-capitalized platforms, hyperscaler self-build and specialist providers can intensify competition for power, customers and equipment.[CP008, CP009, CP010, CP011, CP012, CP013]

3.4 Competitive diligence conclusion

PDG appears best positioned where a customer values APAC coverage, carrier and cloud neutrality, capacity planning and a delivered campus rather than a single-site commodity lease. The evidence does not support a claim that PDG has lower pricing, superior uptime, higher occupancy, better customer retention or a stronger security posture than named peers. A buyer could multi-home across operators or retain hyperscaler and internal-build options, so switching costs are likely highest after site, power, network and technical design commitments are embedded. The next underwriting step is a comparable bid set covering price per contracted megawatt, power and grid timing, PUE, uptime, service credits, contract tenor, renewal and expansion rights, tenant concentration, and financing terms.[CP012, CP013, CP014, CP015, CP016]

3.5 Exhibits

4.1 Revenue model and monetization

PDG supplies the physical space, power, and cooling needed for large-scale cloud and AI computing, with public descriptions positioning hyperscalers and enterprises as the customer base. Industry reporting describes the wholesale model as leasing capacity under long-term contracts, often 10 to 12 years, which can make cash flows financeable; that tenor is an industry observation, not a disclosed PDG contract term. The retained sources do not reveal PDG's realized price per MW, kW, rack, or campus, nor revenue mix, escalation clauses, discounts, pass-through power charges, or recognition policy. Revenue quality is therefore plausible but unverified: physical capacity and long contracts can support recurring revenue, yet the public record cannot establish contracted backlog, customer concentration, or collectability.[CI001, CI002, CI005, CI016]

4.2 Capital intensity and financing dependency

PDG's buildout is capital intensive before it is revenue producing. The 120 MW JC3 campus was announced with USD 1 billion of investment, implying roughly USD 8.3 million per announced MW before adjusting for land, phasing, financing, fit-out scope, or timing. In 2025 the company said it raised USD 2.5 billion across preferred equity and debt, while Stonepeak later reported closing its USD 1.3 billion preferred investment. In 2026, reporting said PDG planned as much as USD 5 billion of additional debt, with more than USD 4 billion potentially tied to specific assets and roughly USD 750 million at parent level. These are financing amounts or intentions, not cash-on-hand, available runway, or evidence that each facility will earn its cost of capital.[CI003, CI004, CI006, CI007, CI008, CI009, CI010, CI011]

4.3 Public traction proxies are capacity, not utilization

Public traction is expressed mainly through portfolio and facility capacity rather than financial output. MLQ tracks 10 facilities totaling about 0.5 GW across six countries, explicitly combining operational, under-construction, and planned sites. Baxtel describes more than 20 facilities and reports power growth of 164% since 2022, while other reporting cites a portfolio exceeding 1.8 GW. These measures have different coverage and development-status denominators, so they should not be combined or treated as occupied capacity. Tracxn separately reports USD 2.16 billion of total funding over six rounds and a May 2026 debt round with the amount withheld. Capacity expansion and financing access demonstrate scale, but neither proves revenue, utilization, or positive unit economics.[CI003, CI012, CI013, CI014]

4.4 Financial verdict and diligence blockers

The financial profile is consistent with a long-duration infrastructure platform: potentially contracted revenue sits against large upfront development costs, operating expenses, and debt service. The main risk is timing. An industry analysis notes that PDG has committed USD 2.5 billion to India and frames future occupancy as the test of whether hyperscaler demand arrives on schedule. That risk becomes more consequential alongside the planned 2026 debt program. Without campus-level occupancy, contracted MW, tenant concentration, revenue, EBITDA, power pass-throughs, maintenance capex, cash, net debt, covenants, and recourse waterfalls, investors cannot calculate gross margin, debt-service coverage, payback, burn, or runway. The appropriate verdict is financially credible but not underwriteable from public metrics alone.[CI003, CI004, CI015, CI016]

4.5 Exhibits

5.1 Product surface and customer workflow

PDG's product is a facility-and-operations service for hyperscalers and enterprises that need power, cooling, physical space, network access, and on-site support. The Open Compute Project listing for SG1 enumerates hyperscale data center, wholesale colocation, build-to-suit, on-demand data center, colocation, technical, connectivity, and workplace services, and describes the facility as carrier- and cloud-neutral. That breadth gives PDG several ways to package capacity, from a tailored large deployment to a supported colocation footprint, but the public material does not disclose product-level pricing, contracted megawatts, occupancy, service-level performance, or attach rates for ancillary services. Portfolio boundaries also vary by tracker: ArusGrid counts nine active projects in four countries, while DC Atlas counts 23 facilities across 16 markets in seven countries. These are useful operating maps, not a reconciled asset register or evidence that every listed facility has the same product scope.[CE001, CE002, CE003, CE004, CE005]

5.2 Facility architecture and technical differentiation

The clearest technical evidence is site-specific. SG1 is described with two diverse 22 kV utility supplies, N+1 diesel generation for IT load, distributed redundant UPS, water-cooled chillers, CRAC units on 2N power, two fiber-entry vaults separated by 180 metres, biometric access, continuous guards, alarm monitoring, fire suppression, and on-site engineers. Its listed accreditations include SOC1 Type 2, SOC2 Type 2, ISO 9001, ISO 27001, ISO 45001, and TVRA. Structure Research adds a separate proof point: TY1 in Tokyo earned NVIDIA DGX-Ready partner certification for liquid-cooling capability. DataM Intelligence describes planned AI-ready capacity at MU1 and CH1, but planned capacity is not commissioned capacity. The defensible differentiation is therefore execution around scarce land, power, cooling, fiber, and facility operations; the available evidence does not establish proprietary hardware, software, or a fleet-wide technical specification.[CE006, CE007, CE008, CE009, CE010, CE011, CE012]

5.3 Maturity, deployment dependencies, and diligence risks

Public trackers show why pipeline and operating product must be separated. ArusGrid reports 1,146 MW of planned IT load but only 80 MW of disclosed running capacity across its nine-project scope; it classifies two projects as operational, one as operational and expanding, two under construction, one at land and permitting, and three as announced. It also reports confirmed AI workloads at three sites and explicitly says that undisclosed tenants reveal nothing about occupancy. DC Atlas reports a broader 450+ MW of live capacity but flags 14% of that live capacity as estimated, demonstrating a material methodology gap rather than a clean conflict. Mingtiandi's report that PDG opened a 96 MW first Japan campus adds evidence of delivery, while Stonepeak's description of greenfield development and M&A shows that integration and build execution remain central dependencies. Current Singapore vacancies for critical-facilities maintenance and building-management-system engineering are a practitioner-signal proxy, not proof of staffing sufficiency, uptime, or incident performance.[CE003, CE004, CE005, CE013, CE014, CE015, CE016]

5.4 Exhibits

6.1 Customer segments and buying proposition

PDG is positioned as a wholesale infrastructure supplier to global hyperscalers and large enterprises rather than as a mass-market hosting provider. Company-originated descriptions identify those two demand groups, while the SG1 facility listing makes the offer more concrete: hyperscale data-center capacity, wholesale colocation, build-to-suit, on-demand colocation, connectivity, technical support, and workplace services. This makes cloud and AI platforms the likely anchor buyers for large blocks of power and space, with enterprises representing a second colocation and connectivity segment. The public materials do not allocate contracted capacity or revenue between these groups. Geography is broad, but geographic presence is not itself evidence of a diversified customer mix or of realized demand in each country.[CU001, CU002, CU014]

6.2 Capacity growth is visible; customer adoption is not

The clearest longitudinal evidence is physical portfolio expansion, not customer usage. A 2022 industry publication described 20 data centers with more than 600 MW of potential power and roughly half operational; July 2025 reporting described more than 20 assets and a 1 GW-plus platform; November 2025 reporting described more than 1.2 GW of IT capacity across more than 20 campuses in seven countries. An April 2026 analysis then described a 1 GW India portfolio target within 1.8 GW of operating and planned global capacity. These figures use different definitions and include pipeline, so they cannot be read as occupied MW, active accounts, or revenue. They show supply-side scale and expansion ambition while leaving the commercial conversion rate unresolved.[CU007, CU008, CU009, CU010, CU011, CU012, CU013]

6.3 Named accounts are attributed, but not independently confirmed by customers

Dgtl Infra's 2022 profile is the only retained source that names customers. It attributes Amazon, Alibaba, Pinduoduo, and Lazada in e-commerce; Microsoft Azure, Google Cloud, Alibaba Cloud, and IBM Cloud among cloud providers; and Facebook, now Meta, in content. This is more useful than a generic hyperscaler label, but the assigned evidence does not contain a customer-authored announcement, campus mapping, contracted MW, production-versus-pilot status, or measured outcome for any name. The table therefore presents the names as dated third-party attributions, not as verified current tenants. Facility certifications and resilience features may support procurement, but they do not establish that a named account is deployed or retained.[CU003, CU004, CU005, CU006, CU014]

6.4 Durability and concentration remain the underwriting gaps

No retained source provides NRR, GRR, churn, renewal rates, contract lengths, cohort behavior, occupied capacity, or customer-level revenue concentration. Those quantities are treated as gaps rather than estimated. Stonepeak identifies greenfield development and M&A as growth paths, and PDG's Jakarta and South Korea projects show how geographic and campus expansion could follow hyperscaler demand. The adverse case is equally important: an independent 2026 analysis argues that the India investment depends on hyperscaler demand absorbing 1 GW at adequate returns. Until signed capacity, occupancy, renewal schedules, termination rights, and top-customer exposure are reviewed, portfolio growth should not be interpreted as durable customer expansion.[CU011, CU012, CU013, CU015, CU016]

6.5 Exhibits

7.1 Regulatory and policy exposure

Singapore is tightening oversight of digital infrastructure at the same time that PDG operates a Singapore facility and expands across APAC. A proposed Digital Infrastructure Bill would license commercial data centers at or above 3 MW and major foundational infrastructure at or above 10 MW, with requirements spanning security, continuity, incident reporting, and efficiency. The proposed penalties are material, but the detailed technical codes and final legislation remain unresolved. Singapore's prior moratorium and competitive allocation of new capacity also show that power, land, sustainability, and permitting are strategic constraints rather than routine administrative matters. The immediate diligence question is whether each PDG site falls within current or proposed regimes and what capex is required to comply.[CR001, CR002, CR009, CR010]

7.2 Resilience, security, and service continuity

SG1's public facility description documents layered controls: diverse utility feeds, N+1 generators, distributed-redundant UPS, water-cooled chillers, fire suppression, two separated fiber entries, biometric access, 24x7 guards, and multiple certifications. These are design and control assertions for one facility, not proof of portfolio-wide uptime or incident performance. Singapore's cyber authorities report that critical-infrastructure owners will soon have to report suspected advanced persistent threat incidents, while APT activity detected in Singapore rose more than four-fold from 2021 to 2024. PDG's risk is therefore both technical and disclosure-related: an incident, prolonged outage, or delayed notification could affect tenants and licensing, while the public record does not establish PDG's incident history, testing cadence, or customer-specific recovery commitments.[CR003, CR004, CR011, CR012, CR014]

7.3 Capital intensity and expansion dependencies

PDG's expansion model combines greenfield development and M&A across six APAC markets, with public reporting describing a 1.3 billion dollar preferred-equity investment, more than 20 assets, and project financing for campuses in Mumbai, Langfang, and Tokyo. That capital access is a strength, but it creates execution, refinancing, and partner-dependence risks: construction schedules, power availability, equipment procurement, tenant ramp, and financing terms must all align. The public record does not disclose net leverage, covenant headroom, recourse by project, or cash-flow waterfalls. Investment underwriting should therefore treat the scale claims and sponsor support as evidence of capacity to pursue growth, not as evidence that portfolio-level returns or liquidity are secure.[CR006, CR007, CR008, CR013, CR015]

7.4 Monitoring indicators and thesis-break conditions

The risk posture is monitorable, but several decisive indicators are private. Regulatory monitoring should track enactment, licensing thresholds, PUE and water-efficiency codes, and any site-specific enforcement. Operational monitoring should track reportable cyber events, material outages, recovery-test results, SLA credits, and certification renewals. Expansion monitoring should track energization dates, construction delays, tenant commitments, project-finance drawdowns, refinancing spreads, and covenant headroom. A thesis break would be a material inability to obtain or retain required licenses, a repeated service disruption or reportable cyber incident without credible remediation, or a funding and construction slowdown that leaves planned capacity uncommissioned. These triggers convert the current evidence gaps into specific diligence requests.[CR009, CR011, CR014, CR015]

7.5 Exhibits

8.1 Investment case and anti-thesis

PDG's investment case is an infrastructure-scale growth thesis rather than a software-multiple thesis. The company describes a six-country operating footprint, while independent and industry sources describe more than 20 assets and a wholesale model serving hyperscalers. Stonepeak's completed USD 1.3 billion preferred-equity investment and PDG's reported USD 1.2 billion debt financing show institutional willingness to fund expansion, but they do not disclose revenue, EBITDA, occupancy, or the liquidation-preference economics that determine common-equity value. The 500 MW powered-land acquisition and the Jakarta 120 MW campus demonstrate a substantial development pipeline, yet also increase execution and financing exposure. The anti-thesis is therefore concrete: a large capital program can create value only if hyperscaler demand converts into contracted, occupied capacity at returns above the cost of capital. The available evidence does not yet verify that conversion.[CV001, CV002, CV003, CV004, CV005, CV006]

8.2 Price discovery and entry discipline

The public price signal is incomplete. Mingtiandi reports that earlier discussions contemplated a transaction potentially valuing PDG at about USD 4 billion, but the source describes that figure as a reported potential transaction value rather than a completed valuation. The later Stonepeak release confirms the USD 1.3 billion preferred investment but gives no post-money valuation, conversion terms, dividend or liquidation preference. Comparing PDG with listed infrastructure operators is also constrained because PDG does not publish a revenue base in the retained evidence. Equinix's October 2026 TTM price-to-sales page reports a P/S ratio of 9.46 in its headline, while another line on the same page reports 6.44 as current; that internal inconsistency makes it a directional market reference, not a clean multiple input. Entry discipline should therefore be staged: require the cap table and preferred terms, then test contracted MW, occupancy, revenue and project-level leverage before accepting any implied valuation.[CV007, CV008, CV009, CV010, CV011]

8.3 Scenarios, exit readiness, and thesis-break triggers

Scenario analysis should be anchored to capacity conversion and financing, not invented revenue forecasts. In the bull case, the 500 MW expansion and 120 MW Jakarta campus reach service on schedule, hyperscaler contracts absorb the powered capacity, and long-duration infrastructure financing supports attractive equity returns. The base case assumes demand remains strong but commissioning and occupancy occur in stages, requiring disciplined leverage and continued sponsor support. The bear case is a timing and capital-structure problem: PDG raises substantial debt against projects whose contracted demand or cash flow is delayed, leaving capital tied up while preferred claims and interest reduce common-equity optionality. Exit readiness is not publicly demonstrated; the company remains private and the evidence does not establish a sale, IPO, or transparent secondary market. The thesis should break if contracted MW, occupancy, project-level debt service, or commissioning milestones miss agreed thresholds. Those are diligence asks, not facts inferred from the public record.[CV012, CV013, CV014, CV015, CV016]

8.4 Exhibits