Data centre investing is not conventional property investing. The assets look like real estate and are often analysed using real estate metrics, but the drivers of value are fundamentally different. Power availability, execution capability, tenant behaviour, and regulatory positioning determine outcomes far more than location, zoning, or building specification in the traditional sense. Investors who apply conventional property frameworks without modification will systematically misprice both the opportunity and the risk. This primer is the second in the Golden Section Capital Data Centres series. Where Primer 1 established the conceptual and market foundation: what data centres are, how they operate, and how geography and regulation shape supply, this primer addresses the investment questions that follow: how to evaluate data centre assets and vehicles, how to access the sector through listed markets, and how to identify the characteristics that separate high-quality investments from those that simply benefit from sector tailwinds. The investment case begins with a clear view of what makes a good data centre investment. Secured, deliverable power in a proven demand cluster is the irreducible foundation. Assets with energised or near-term deliverable megawatts in supply-constrained markets command structural pricing power. Assets with paper capacity should not be valued on stabilised assumptions. A one-year development delay reduces project IRR by approximately 0.5 to 0.9 percentage points even where the stabilised economics are otherwise intact, making time-to-power one of the most important variables in the underwriting calculus. Valuation requires triangulation rather than a single framework. Discounted cash flow, yield-on-cost, and replacement cost each capture different aspects of data centre economics, and each must be stress-tested against realistic power delivery scenarios. Cap rates applied to headline income can obscure the true economics where power pass-throughs, service revenue, and recurring capital expenditure are inconsistently treated. The metrics that matter most are contracted MW versus energised MW, WALE disaggregated by contract type, PUE trajectory, and ESG compliance against regulatory thresholds; these are examined in detail in Section 2. The sector is accessible through two structurally distinct routes. Private market platforms have captured some of the most compelling assets over the past five years, with landmark transactions including the US$40 billion acquisition of Aligned Data Centers, the US$16.1 billion acquisition of AirTrunk by Blackstone, and KKR and Singtel's S$13.8 billion consolidation of STT GDC in early 2026, illustrating the premium that strategic and sovereign investors are prepared to pay for secured power pipelines at scale. The listed market offers daily liquidity, regulated disclosure, and current income distribution, but operates within tighter capital constraints, particularly in the Singapore S-REIT framework where MAS imposes a 50% aggregate leverage limit and a minimum interest coverage ratio of 1.5 times. Singapore's listed data centre universe, spanning three pure-play REITs and three diversified REITs with material data centre allocations, provides one of the most institutionally credible and analytically transparent entry points into the sector available globally. Keppel DC REIT, Asia's first pure-play data centre REIT and an STI constituent, trades at a price-to-book of 1.29x, reflecting its sponsor pipeline, colocation model, and long WALE. Digital Core REIT, sponsored by Digital Realty, trades at price-to-book of 0.62x, presenting a potential re-rating opportunity as its Linton Hall lease-up removes the primary near-term overhang. NTT DC REIT, which completed the largest SGX REIT IPO in a decade in July 2025, brings a third pure-play option underpinned by NTT Group's global platform and a GIC cornerstone stake. The three diversified names: CapitaLand Ascendas REIT, Mapletree Industrial Trust, and CapitaLand India Trust, each offer data centre exposure at varying degrees of concentration, risk profile, and development-phase optionality. The risk landscape is material and must be understood before any position is sized. Technology obsolescence, tenant concentration, geopolitical exposure, and the structural tension between hyperscale tenants building their own capacity are all live considerations. The sector remains structurally attractive, but outcomes will continue to diverge based on access to power, execution capability, and the ability to deploy capital into markets where returns justify the risk. This primer provides the analytical framework to help make that determination.