India AI Data Centre Cooling 2026–2035

The Thesis

India has a documented data-centre cooling opportunity, but earlier proprietary market forecasts lack an auditable foundation. This report replaces uncalibrated readiness scores with attributed external benchmarks, a dated H1 2026 capacity inventory (1,789 MW operational, 509 MW under construction, 3,351 MW planned), and specific company evidence. It establishes that cooling architecture depends on the complete system design, not just chip-power thresholds, and identifies verified domestic manufacturing activity alongside the real execution risks in power, water and supply-chain qualification.

Key Numbers

Key Findings

The Framework

What It Means

For Policymakers

Public support should address testing, skills and qualification barriers with measurable outcomes. Policy inputs must distinguish enacted requirements from draft proposals. The audit found no substantiated evidence of a dedicated cooling-components PLI or a universal Indian data-centre WUE ceiling; incentives should be appraised for additionality and fiscal cost without presuming quantified import shares.

For Industry

Operators should tender a complete thermal scope and require acceptance tests. Procurement must identify which party owns the server loop, the facility loop, the control logic and the response to leakage. Manufacturers should choose a defined product, secure design rights and prove repeatable quality before adding capacity — verifying exact assembly rather than inferring capability from general HVAC presence.

For Investors

Financing milestones should link to qualified orders, customer commitments and demonstrated cash conversion. Evaluate entry routes (licensed component production, assembly, system integration) against design rights, qualification time and warranty exposure. A proposed investment must identify intellectual-property rights and manufacturing-process liability; generic cooling-segment IRRs are not asserted.

Analytical Figures

The divergence between announced megawatts and actual cooling procurement necessitates a strict accounting of project stages. Treating the entire pipeline as immediate demand overstates the addressable market. The figure below isolates the conditional arithmetic of capacity build from a single dated inventory, showing how the addressable thermal market expands only upon verified construction milestones.

These conditional cases do not assign probabilities to construction or planned projects. They serve as boundary conditions for thermal-procurement modelling. A negative return on a cooling-manufacturing investment is not repaired by assuming the automatic conversion of "planned" megawatts into immediate equipment orders.

Market sizing for India remains fragmented across external research publishers, preventing a defensible pooled headline. The next figure shows the material divergence in 2025 estimates — the reason Techadyant withholds a proprietary 2035 total addressable market without project-level quantity and price matching.

Different scope and estimation methods prevent averaging these figures. A bottom-up market forecast requires project quantities, compatible architecture, matched prices and documented overlap eliminations — which this edition does not artificially construct.

The Numbers, Tabulated

Table 1 — Capacity completion sensitivities (H1 2026 inventory)

Source: Cushman & Wakefield H1 2026 [PUB-16]; modelled by Techadyant Labs. No retirements or outside additions assumed.

Table 2 — Selected substantiated transactions (2025–2026)

Source: Techadyant Funding & Investment Tracker [PUB-03, PUB-10, PUB-38, PUB-39].

What to Watch

• 2026–2027 Commissioning of the 509 MW under-construction capacity and verification of actual thermal-procurement packages versus announced specifications.
• 2027+ Evolution of the broad EU PFAS restriction proposal and its downstream impact on immersion-cooling fluid supply chains and end-of-life treatment compliance.
• Ongoing Verification of domestic manufacturing claims versus actual indigenous intellectual property, component sourcing and repeatable yield rates.
• Ongoing Site-specific groundwater-extraction permissions and their impact on facility-level heat-rejection design choices in high-density corridors.

Frequently Asked Questions

Sources & Methodology

This edition uses desk research checked on 27 September 2026. Primary sources include company announcements, public-authority records, original research-publisher releases, and the downloaded Cushman & Wakefield H1 2026 report. A publisher is the primary source for its own estimate; that does not make the estimate an independently verified market fact. Every load-bearing claim resolves to a numbered source record (PUB-01 onward) in the report and companion workbook.

1. Cushman & Wakefield — India Data Centre H1 2026 Update [PUB-16]
2. Dell'Oro Group — Data Center Liquid Cooling Market Forecast [PUB-17]
3. Turner & Townsend — Data-centre construction cost index 2025–2026 [PUB-22, PUB-23]
4. NVIDIA — DGX SuperPOD Data Center Best Practices with DGX B200 [PUB-05]
5. Company records — Schneider Electric, Vertiv, Voltas, Blue Star, Ecolab, Trane Technologies [PUB-24 to PUB-27, PUB-03, PUB-10]

Get the Full Report

The free edition is a 12-page briefing covering the thesis, the audited capacity inventory and the evidence discipline. The full report runs to 59 pages with 22 tables and 43 sourced records — the audited company register, the technology-qualification framework, the capacity-completion model and the investment-diligence checklist.