India's Data Centre Capex Cycle: Mapping the Value Chain, Sizing the Beneficiaries

Thematic equity research. For informational purposes only — this report does not constitute investment advice. See disclosures at the end.

Executive Summary

India's data centre (DC) buildout is emerging as a structural, multi-year capex cycle spanning real estate, power infrastructure, cooling, racks/white-space and project management — and, increasingly, as its own thematic pillar within Indian real estate and industrial capital goods. This report maps the full DC value chain to listed Indian equities, sizes each segment's share of total capex, and asks a simple question of each company: is the market pricing DC-linked names in line with disclosed exposure, or has sentiment already run ahead of fundamentals?

Unlike a traditional commercial real estate cycle, the competitive edge in this theme is shifting away from land ownership toward access to uninterrupted power, fibre connectivity, faster regulatory approvals, hyperscaler partnerships and balance-sheet strength to fund capital-intensive projects. That shift widens the investable universe well beyond real estate developers into power equipment, cooling, EPC/construction, connectivity and digital-infrastructure operators — the full bill-of-materials of a data centre.

What this report does:

• Explains India's current data centre ecosystem and why AI is driving a structural capex cycle, distinct from the prior enterprise/cloud DC cycle
• Positions India within the global data centre and semiconductor landscape — capacity share, cost advantage, and the compute-vs-chips distinction
• Breaks down how a data centre is built and where every rupee of capex is deployed, layer by layer
• Maps the complete DC value chain to listed Indian equities and classifies each company by its role in the ecosystem
• Uses confirmed, sourced valuation data (Screener.in, as of July 2026) — no estimated EV/EBITDA — to compute segment medians and compare companies strictly within their own segment
• Sets out a 5-10 year outlook for the theme and the risks that could delay or de-rate it

Headline View

• Electrical Systems — 42% of total capex, the single largest line, remains the most crowded, most liquid and most bifurcated segment. A premium cluster of large-cap grid/HVDC names (Hitachi Energy, CG Power, ABB, GE Vernova T&D) trades 70-170% above the segment median of 35.0x EV/EBITDA, while a value cluster of transformer, genset and mid-cap cable names (Powerica, Voltamp Transformers, Havells, Kirloskar Oil) trades at a discount to the segment median despite disclosed DC order-book traction.
• Base Building/Shell (15% of capex) is the widest-dispersion segment — from NCC at 5.6x to PSP Projects at 20.1x EV/EBITDA. On confirmed EV/EBITDA (unlike PE, which is scale-compressed for L&T), the segment's largest constituent does not carry the lowest EV/EBITDA multiple in the segment; L&T trades above the segment median once leverage is captured in the multiple.
• Mechanical & Cooling (16% of capex) is tightly banded around Voltas's 59.7x median, with KRN Heat Exchanger — the purest listed cooling pure-play — now trading near the segment median on confirmed data rather than at a premium, a meaningfully different read from PE-based or estimated comparisons.
• Earnings contribution remains gradual, not immediate. Roughly 60% of India's announced future DC capacity sits at the planning or early-execution stage; construction typically runs three to four years before stabilised rental or services revenue begins flowing to the equity.

Key Sector Metrics

AI: India Real Estate's Next Structural Growth Pillar

Rapid expansion of artificial intelligence and hyperscale data centres is emerging as a structural opportunity for India's real estate and industrial sectors, but market watchers are consistent on one point: it is likely to remain a gradual contributor to earnings over the long term, not a near-term catalyst. Vijay Agrawal, managing director for infrastructure at Equirus Capital, frames the shift as the sector moving from an optional diversification play to a strategic, long-term asset class — while cautioning that the industry is still in the early innings of earnings contribution. Industry estimates cited by Agrawal suggest India could see ₹1.6-2 trillion of data centre investment over the next five to seven years — a capex programme large enough to move the earnings trajectory of an entire sub-set of listed industrials, power-equipment and construction companies, even though it will take years to show up in reported numbers.

A Different Kind of Real Estate Cycle

The traditional commercial real estate playbook — acquire land, build, lease — still applies at the margin, but the competitive moat for a data centre developer is no longer land ownership alone. It is access to uninterrupted power supply, fibre connectivity, faster regulatory approvals, partnerships with global hyperscalers or specialist operators, and the balance-sheet strength to fund capital-intensive projects. This is why developers such as Anant Raj, Hiranandani Group, Macrotech Developers (Lodha), DLF and Mindspace Business Parks REIT are positioning data centres as a strategic, long-term asset class distinct from their legacy residential or office portfolios — while power-equipment, cooling and EPC companies with no historical real-estate exposure at all are becoming equally central to the theme.

Pankaj Kumar, vice-president for fundamental research at Kotak Securities, singles out Lodha, DLF, Mindspace REIT and Anant Raj among listed developers as best positioned on land availability, execution capability and existing hyperscaler partnerships. Shariq Merchant, associate director for investments at WhiteOak Capital Mutual Fund, argues that power — not land — is now the binding constraint: meeting India's AI-scale data centre ambitions requires building out transmission infrastructure and securing dedicated power tie-ups at scale. He adds that financing, transmission build-out, suitable land, cooling infrastructure and policy clarity will together determine the pace, not the direction, of execution over the next decade.

Why the Earnings Contribution Is Gradual, Not Immediate

Pankaj Kumar of Kotak Securities puts India's data centre IT load at roughly 1.6 GW today, rising to 5 GW by FY30 and 14-15 GW by FY35, with built-up space requirements expanding from about 25 million sq ft currently to nearly 160 million sq ft by FY35. Developing a data centre requires capital expenditure of roughly ₹80-100 crore per megawatt — materially higher than a conventional office building — and construction typically takes three to four years before stabilised revenues begin.

Akshay R Shetty, an equity research analyst at Mirae Asset Sharekhan, describes the data centre cycle as creating value selectively rather than across the sector: residential sales and office rentals remain far larger profit pools for most listed developers, and most announced projects are still at the MoU, land-identification or construction stage. In Shetty's framing, what is visible today is largely land appreciation and one-time monetisation; recurring earnings arrive only once powered buildings are completed, tenants take possession and rental or services income starts flowing — a three-to-five-year story, not a two-quarter one, a characterisation this report adopts throughout.

Markets appear to be differentiating between announcement and execution. Agrawal notes that valuations already distinguish between companies with operational assets and those with merely announced expansion plans; firms that have commissioned capacity or secured long-term customer tie-ups are attracting greater investor confidence, while developers still at the MoU or land-acquisition stage face what amounts to an execution discount. Merchant adds that data centres do not yet constitute a meaningful share of valuations for most listed real estate names given still sub-scale exposures — which is precisely why there remains scope for selective re-rating as capital deployment to the segment increases and companies convert pipeline into contracted, recurring cash flow.

Source: Kotak Securities estimates, cited in Prachi Pisal, "AI data centres likely to be realty's next growth pillar," Business Standard, 19-Jul-2026.

Where Listed Developers Already Stand

• Macrotech Developers (Lodha) plans ~1 GW of build-to-suit DC capacity at Palava near Mumbai (₹10,000-11,000 crore) and has committed to a ₹1 trillion, 2.5 GW DC park in Maharashtra over time — by far the largest single developer commitment in the listed universe.
• Anant Raj has announced ₹20,000 crore of investment in Haryana in addition to an earlier ₹4,500 crore commitment in Andhra Pradesh, and is the only real-estate-led name in this report's confirmed-data set with disclosed listed DC/land economics.
• Mindspace Business Parks REIT is developing more than 1 million sq ft of data centre space in Navi Mumbai with Princeton Digital Group — a template for REIT-led, JV-based DC development that several peers are likely to replicate.
• DLF is a named participant in the sector's structural narrative; its overall listed multiples appear in Table 3, but the company does not disclose DC-specific segment revenue, so it remains classified as a Watchlist Candidate rather than a Core or Emerging Beneficiary.
• Hiranandani Group is unlisted and has no listed market data to include in this report's framework.

India in the Global Data Centre and Semiconductor Landscape

India remains a small but fast-growing node in global data centre capacity — and, importantly, a downstream beneficiary of the global AI compute build-out rather than a producer of the leading-edge chips that power it. Understanding both halves of this picture — capacity growth and the semiconductor supply chain feeding it — is essential context for sizing the listed-equity opportunity correctly.

India's Data Centre Capacity: Small Base, High Growth Rate

India hosts roughly 3-4% of global data centre capacity today, against installed capacity of under 2 GW versus a global base exceeding 40-50 GW. That low base is the point: India's construction and operating costs run 30-40% below China and the United States, aided by relatively low land prices, competitive power tariffs and government incentives — a cost advantage that is now attracting hyperscaler capital even as the absolute capacity gap remains wide. Government support has scaled up materially through 2026, including a 20-year tax exemption for foreign cloud-service providers building India-based capacity (Union Budget 2026-27) and a proposed National Data Centre Policy that would grant the sector infrastructure status, unlocking longer-tenor financing.

Table 1 — India's data centre capacity and cost position. Source: Kotak Securities/Business Standard; industry reports (Vestian, CBRE, KPMG, CARE Ratings), July 2026.

The Semiconductor Angle: India Builds the Shell, Not (Yet) the Chip

The AI data centre cycle is fundamentally a demand story for the GPUs and AI accelerators manufactured by TSMC, Samsung and a handful of other leading-edge foundries — none of which currently exist in India. India's semiconductor build-out is real but is targeted at mature process nodes (28-110nm), serving automotive, industrial and display-driver demand rather than leading-edge AI silicon. Under India Semiconductor Mission (ISM) 2.0, roughly ₹1.6 lakh crore of investment has been approved across a dozen-plus projects in six-plus states, anchored by the Tata Electronics-PSMC fab at Dholera (targeting first silicon in 28nm on a 300mm line). Leading-edge logic capability (sub-7nm, the class of node used in frontier AI chips) is not expected in India before 2030 at the earliest.

For this report's purposes, the practical implication is that India's listed-equity opportunity in the AI data centre cycle sits almost entirely downstream of the chip — in real estate, power equipment, cooling and construction that houses and services imported GPUs — rather than in domestic chip manufacturing itself. Investors seeking direct semiconductor-fab exposure to the AI cycle currently have very limited listed options in India; the data centre value chain mapped below is the more directly investable proxy for the theme on the NSE/BSE today.

Anatomy of a Data Centre: How Capital Is Deployed

A hyperscale data centre is built in layers, each with a distinct capital-intensity profile, construction timeline and supplier base. Understanding the bill of materials — not just the headline capex number — is what allows specific rupee flows to be mapped to specific listed companies, rather than treating "data centres" as a single undifferentiated theme.

The Cost Structure

A typical Indian data centre's capex bill splits into three broad buckets. Electrical Systems and Mechanical/Cooling together — the "Core Infrastructure" — account for 59% of total build cost, making the power-equipment and thermal-management supply chains the largest addressable pools for listed-equity investors. Real estate (land + shell) is 26% of the bill, and white-space/support infrastructure the remaining 15%.

Source: Kotak Securities (industry estimates); ThirdView Capital analysis.

Because Electrical Systems Dominates the Bill, It Dominates the Theme

Electrical Systems, at 42% of total spend, is nearly three times the weight of the next-largest infrastructure line. This matters for portfolio construction: a re-rating or de-rating in transformers, switchgear, gensets, UPS systems and cabling moves the aggregate "DC theme" far more than an equivalent move in racks, land or design/PM — which is why the valuation work below devotes proportionately more attention to that segment.

Build Timeline and Unit Economics

• Capex intensity: ₹80-100 crore per MW of IT load — higher than a conventional office building of equivalent built-up area, reflecting the density of power, cooling and redundancy infrastructure required.
• Construction timeline: typically 3-4 years from land/shell to stabilised revenue — the primary reason earnings contribution lags capex announcements by several years.
• Redundancy and uptime requirements (N+1 or 2N power and cooling architectures for Tier III/IV facilities) are what push Electrical Systems and Mechanical & Cooling to a combined 59% of the cost stack, well above their share in conventional commercial construction.

The Data Centre Value Chain: Segment Valuation and Company Mapping

This section maps each layer of the cost structure to its listed Indian equity universe, and values each segment on its own terms.

Methodology: segments are populated with listed companies carrying disclosed, material DC-linked revenue exposure per company filings, brokerage coverage or the source materials for this report; broader conglomerates with immaterial DC exposure are still shown, but flagged as Diversified Players rather than Core Beneficiaries. All EV/EBITDA and PE figures are sourced directly from Screener.in (confirmed, July 2026) — no multiple in this report is estimated or derived. Segment medians are the relative-valuation yardstick throughout: trades at a premium to the segment median means trading materially above the segment median, trades at a discount to the segment median means materially below, and trades near the segment median means within ~15% of median. These labels describe where a company's multiple sits relative to its listed peers only; they are not a judgement of intrinsic value or mispricing, and a premium or discount to the segment median can equally reflect scale, balance-sheet, or growth-rate differences unrelated to data centre exposure. Segments with fewer than three usable listed comparables are discussed qualitatively but not charted, since a scatter needs enough points to say anything meaningful about clustering or dispersion.

Electrical Systems — 42% of Capex, the Largest and Most Bifurcated Segment

This segment covers transformers, HV switchgear, gensets, UPS systems and cabling — the power backbone that gets electricity from the grid to the server rack. It is the most crowded, most liquid part of the DC value chain, spanning names from ~₹7,000 crore genset makers to ~₹1.6 lakh crore diversified electricals majors.

Reading the data: a clear bimodal split. A cluster of large-cap grid/power-equipment names — Hitachi Energy (93.9x), CG Power (77.1x) and ABB (69.2x) — trades 70-170% above the segment median, a level that implies elevated market expectations for a multi-year power-transmission-and-DC supercycle rather than DC exposure alone. A cluster of transformer, genset and mid-cap cable names — Powerica (15.8x), Voltamp Transformers (23.2x), Havells (27.5x) and Kirloskar Oil Engines (24.8x) — trades meaningfully below the segment median; Kirloskar Oil's discount is set against a named, disclosed contract (its 192 MW HyperNext order), while Powerica and Voltamp Transformers currently lack a similarly specific disclosed DC contract despite their low multiples. Siemens (35.0x) sits almost exactly at the median and RR Kabel (31.3x) and KEI Industries (33.0x) are close behind, trading near the segment median given their disclosed exposure.

Fig. — Electrical Systems: 15 listed comps, segment median 35.0x. Source: Screener.in (confirmed, Jul-2026).

Valuation view: trading at a premium to the segment median — the large-cap grid/HVDC names Hitachi Energy, CG Power, ABB and GE Vernova, plus the diversified industrial name Cummins India. Trading near the segment median — Siemens, KEI, RR Kabel. Trading at a discount to the segment median — Polycab, Havells, Kirloskar Oil, Voltamp Transformers and Powerica. Schneider Electric Infrastructure is the one small/mid-cap name trading at a premium to the segment median despite its size; no specific disclosed DC order or revenue line was identified behind that premium, so it is presented here as a relative-valuation observation rather than confirmed DC-specific demand. Cummins India's premium likewise reflects its broader diversified industrial/engine business rather than a disclosed DC-specific order book.

Base Building / Shell — 15% of Capex, the Widest Dispersion

Civil construction and EPC contractors that build the physical shell. India has no clean listed pure-play for DC shell construction, so this segment is a mix of large diversified EPC majors and smaller, story-driven contractors — producing the widest valuation dispersion of any segment in this study.

Reading the data: a greater than 3.5x spread, from NCC (5.6x) to PSP Projects (20.1x). Unlike a PE-based read — where L&T's scale visually compresses its multiple — L&T does not carry the lowest EV/EBITDA multiple in the peer group on confirmed data: at 15.5x it sits 36% above the segment median, because EV/EBITDA captures the company's balance-sheet leverage in a way PE does not. NCC has the lowest EV/EBITDA multiple in the peer group, but neither L&T nor NCC derives a material share of earnings from DC work specifically, so neither a premium nor a discount here reflects a clean DC-specific signal.

Fig. — Base Building/Shell: 5 listed comps, segment median 11.4x. Source: Screener.in (confirmed, Jul-2026).

Valuation view: PSP Projects trades at the widest premium to the segment median, on ownership/order-narrative rather than disclosed DC-specific intensity. Ahluwalia Contracts is the most defensible mid-point in the segment, being both listed and disclosed as having actual DC project wins. L&T's premium to the segment median reflects scale and balance-sheet characteristics rather than disclosed DC-specific demand.

Mechanical & Cooling — 16% of Capex, a Tight Thematic Band

Chillers, CDUs (coolant distribution units) and heat-exchangers that manage the thermal load of high-density AI/GPU racks. This is the smallest listed universe of the core segments (three names) but arguably the most direct read on the "liquid cooling" sub-theme within DC.

Reading the data: on confirmed EV/EBITDA, KRN Heat Exchanger (67.4x) — the pure heat-exchanger micro-cap — sits within 15% of the segment median rather than screening as an outlier, a materially different (and more reassuring) picture than a PE-based comparison would suggest, since KRN's PE of 106x looks stretched in isolation. Blue Star (35.7x) is the outlier here, trading well below Voltas (59.7x) despite a broadly comparable diversified consumer/commercial-AC business mix.

Fig. — Mechanical & Cooling: 3 listed comps, segment median 59.7x. Source: Screener.in (confirmed, Jul-2026).

Valuation view: the segment is currently pricing "DC cooling optionality" almost identically across large diversified players and niche pure-plays — evidence the cooling re-rating is thematic rather than company-specific at this stage. That said, KRN's earnings base remains small and undiversified; trading near the segment median on EV/EBITDA does not remove the execution risk in converting a hyperscale pipeline from pilot orders to recurring, contracted revenue.

Single-Comparable and Adjacent Segments (Not Charted)

Four segments in this report's universe have too few usable listed comparables to chart meaningfully, consistent with the methodology above. They remain part of the investable universe and are classified below, but shown here without a scatter.

Table 2 — Single-comp and adjacent segments. Source: Screener.in (confirmed, Jul-2026).

The Full Company Universe: Roles, Not Rankings

Rather than labelling companies "selected" or "rejected", every company in the universe is classified by its role in the DC ecosystem — because a large diversified electricals major and a small pure-play transformer maker can both be legitimate ways to own the theme, at very different points on the risk/liquidity spectrum. Five roles are used consistently across this report:

• Core Beneficiary — disclosed, material and growing DC-linked revenue; order-book or project evidence of DC-specific demand converting to contracted business.
• Emerging Beneficiary — building a DC-specific pipeline or early orders, but DC is not yet a majority driver of group revenue or earnings.
• Diversified Player — credible DC exposure inside a much larger, multi-segment business; DC is one growth driver among several, not the core investment case.
• Optional Exposure — indirect or minority DC linkage; the primary investment case for the stock lies elsewhere (e.g., EV/lithium-ion, MNC-subsidiary dynamics).
• Watchlist Candidate — plausible participant in the theme per public disclosures or press coverage, but without confirmed, segment-level listed financials to include in the valuation work above today.

Company Universe by Role

Table 3 — Full DC company universe by role. Source: Screener.in (confirmed, Jul-2026).

Reading the Universe by Market-Cap Bucket

The same theme can be expressed at very different points on the risk/liquidity spectrum:

• Large-cap (above ₹75,000 Cr) — liquid, diversified, capped upside. ABB, Cummins, Siemens, Polycab, GE Vernova T&D, L&T and Voltas offer the most liquid way to own the theme, but DC is one growth driver among several. Hitachi Energy and CG Power are large-cap, and their multiples imply elevated market expectations for a power-transmission supercycle, priced as scarce, high-growth franchises.
• Mid-cap (₹15,000-75,000 Cr) — the balance of growth and valuation comfort. Kirloskar Oil Engines is the single most concrete, quantifiable DC-capex data point in the universe (the named 192 MW HyperNext order), and its PE has re-rated materially since disclosure. Blue Star, KEI Industries and Havells carry disclosed DC-linked growth drivers inside businesses still priced largely for legacy end-markets.
• Small-cap (below ₹15,000 Cr) — highest torque, highest idiosyncratic risk. Voltamp Transformers, Powerica and Schneider Electric Infrastructure are classified here as near-pure DC-linked plays; Voltamp and Powerica trade meaningfully below the Electrical Systems median despite disclosed order-book activity, though the specific revenue tied to that activity is not separately broken out in the filings reviewed for this report. KRN Heat Exchanger is the purest cooling-hardware exposure; Ahluwalia Contracts, NCC and PSP Projects span the shell/EPC range from the lowest EV/EBITDA multiple in the peer group to the richest.

Outlook: The Next 5-10 Years

India's data centre IT load is projected to grow from 1.6 GW today to 5 GW by FY30 and 14-15 GW by FY35 — a near-9x expansion in a decade — with built-up space demand growing from roughly 25 million sq ft today to 160 million sq ft by FY35. On the ₹80-100 crore/MW capex intensity discussed above, that trajectory is broadly consistent with the ₹1.6-2 trillion five-to-seven-year investment estimate cited throughout this report, and implies a further, larger wave of capital deployment in the second half of the decade as capacity scales from gigawatts to double-digit gigawatts.

What Has to Go Right

• Power and transmission infrastructure — sector participants are unanimous that power, not land, is the binding constraint; India's ability to build dedicated transmission capacity and secure large-scale power tie-ups will set the pace of the entire cycle.
• Financing and policy clarity — proposed infrastructure status for data centres and continued tax/regulatory support (the 20-year foreign-cloud tax exemption is an early signal) would materially lower the cost of capital for developers and, by extension, the capex flowing to the listed supply chain.
• Execution over announcement — with ~60% of announced capacity still at the planning or early-execution stage, the differentiation between companies that convert MoUs into commissioned, revenue-generating assets and those that do not will likely widen over the next 3-5 years, and valuations should follow that split more closely than they do today.
• Cooling-technology differentiation — as AI/GPU rack densities rise, the divergence between air-cooled and liquid-cooled architectures should become more visible to the market, which could break the current undifferentiated pricing of the Mechanical & Cooling segment into distinct sub-clusters.

Key Risks

• Capex disappointment / demand-cycle risk — the large-cap grid/HVDC cluster (Hitachi Energy, CG Power, ABB, GE Vernova) trades at valuation levels that imply elevated market expectations for a multi-year power-transmission-and-DC supercycle; any slowdown in AI capex globally, or in India's power-transmission buildout specifically, would be high-beta for this cluster in particular.
• Land, power availability and financing cost — Tier-I city constraints (power quality, permitting, community pushback) are pushing developers to less-tested Tier-II locations, while much of the capital funding this cycle is foreign or dollar-linked, making rupee volatility and global rate paths a pass-through risk to project IRRs.
• Competitive intensity from regional peers — Malaysia (Johor), Japan and Vietnam are actively competing for the same hyperscaler capital; India's cost advantage narrows if regional peers close the policy or infrastructure gap.
• Valuation risk in the premium cluster — several Core Beneficiary and Diversified Player names in Electrical Systems already trade 70%+ above segment medians; a de-rating here would move the aggregate "DC theme" narrative disproportionately, given the segment's 42% weight in the underlying capex bill.

Conclusion

Electrical Systems is the segment that matters most to this theme — it is 42% of total capex and shows the clearest bifurcation between a large-cap grid cluster trading at a premium to the segment median and an under-covered small/mid-cap transformer-and-genset cluster (Powerica, Voltamp Transformers, Havells, Kirloskar Oil) that trades below the peer median. Base Building/Shell's apparent cheapness on a PE basis is materially revised once confirmed EV/EBITDA is used — L&T does not have the lowest EV/EBITDA multiple in the segment once balance-sheet leverage is captured — while Mechanical & Cooling still trades as one undifferentiated basket regardless of company size, a pattern that may not persist once cooling-technology differentiation becomes clearer to the market.

Position sizing should reflect that earnings contribution remains gradual — a 3-5 year story — even where the long-term capacity trajectory is not in question.

Across the full universe, the segments trading below their own peer medians despite disclosed order-book traction — the small/mid-cap Electrical Systems names (Powerica, Voltamp Transformers, Havells, Kirloskar Oil), alongside Ahluwalia Contracts and Blue Star — show the widest gap between disclosed exposure and current pricing on confirmed data. The large-cap grid/HVDC cluster and Schneider Electric Infrastructure trade at levels that imply elevated market expectations already, leaving less margin of safety if execution disappoints.

---

Important disclosures: Prepared by ThirdView Capital for informational purposes only; this does not constitute investment advice or a solicitation to buy or sell any security. Valuation multiples and market-capitalisation figures are sourced from Screener.in as of July 2026 and are subject to change. Segment classifications, medians and valuation views reflect the authors' analytical judgement applied to the confirmed data set and should not be the sole basis for any investment decision. Companies marked "Watchlist Candidate" or "not in the confirmed data set" are included for narrative purposes only; no valuation view is expressed on them. Readers should conduct independent research and consult a qualified financial adviser before investing.