Three data-centre formats, three site profiles: hyperscale campus acreage and power, colo carrier density, and standardised edge rollouts, mapped to Chennai's geography.
Hyperscale campuses are cloud and AI factories: tens to hundreds of megawatts, built in phases on large parcels, owned or anchored by a single operator with long-dated power and fibre contracts. Colocation facilities are interconnection businesses: multi-tenant buildings in metro cores where carriers, clouds and enterprises meet, selling racks, cages and cross-connects. Edge nodes are small, distributed capacity, single-digit megawatts or less, placed near population and network aggregation for latency-bound workloads, content delivery, gaming, industrial automation, increasingly AI inference.
Chennai hosts all three layers, and their land questions barely overlap.
The hyperscale land profile: 20-plus acres with expansion headroom, industrial zoning, flood-safe elevation, water for cooling, and above all a credible path to phased high-voltage power, the criterion that disqualifies most marketed parcels. Campus delivery is a multi-year programme where land execution, substation works and hall construction interleave; a title defect or easement gap discovered mid-programme cascades through every dependent contract.
This is where verification carries the most value per rupee: the 30-point land read, power-readiness verification and computed buildable envelope before commitment, as covered across this series.
Metro colocation lives on carrier density: the building's meet-me rooms, the duct routes into it, and its position relative to landing-station backhaul decide its product more than its civil specification. Land strategy here is really asset strategy, acquiring or converting buildings in network-dense pockets, with diligence weighted toward fibre records, existing tenancies and upgrade headroom for power and cooling within an urban envelope.
Chennai's colo stock clusters where the networks already are; entrants should map duct geography before touring buildings.
Edge programmes invert the hyperscale problem: instead of one large execution, dozens of small ones, parcels or rooftops of modest size near aggregation points, each needing zoning compatibility, modest but reliable power, fibre access and rapid approvals. The risk is process: without standardised site criteria, diligence templates and approval playbooks, an edge rollout drowns in per-site friction.
We support edge programmes with exactly that standardisation: a repeatable verification and execution kit, so the twentieth node closes like the second.
Hyperscale is single-operator cloud capacity at campus scale; colocation is multi-tenant interconnection space in metro cores selling racks and cross-connects. They serve different workloads and want different real estate.
Small, distributed capacity, typically a few megawatts or less, placed near users and network aggregation for latency-critical workloads. Site needs are modest but multiplied across many locations.
Indicatively: hyperscale campuses 20+ acres with phasing headroom; colocation reuses or builds metro structures on compact plots; edge nodes fit small urban parcels or existing buildings near aggregation points.