The 500 MW Double-Count Claim: What a 172 MW Parent–Phase Audit Can—and Cannot—Prove

Executive summary

The number 500 MW appears in several India data-centre conversations, but it does not always describe the same thing. Princeton Digital Group (PDG) used it in September 2024 for a four-market, pan-Asia powered-land and expansion programme spanning Mumbai, Chennai, Johor and Jakarta. Its release named 100 MW of Mumbai expansion and a 72 MW Chennai campus as Indian components. Those two entries add to 172 MW; they sit inside the wider programme, not alongside it as a second 172 MW. The release’s other named site figures total 472 MW, not 500 MW, and it does not publish a line-by-line bridge for the remaining 28 MW.

There is a second arithmetic question inside the same release: PDG described a 150 MW MU1 campus endpoint and a 72 MW Chennai campus, while also saying its India footprint would reach 230 MW. The two named figures sum to 222 MW. The public announcement does not reconcile the eight-megawatt difference or specify a common IT-load or gross-power basis. That is an unbridged source figure, not evidence of an eight-megawatt duplicate or deliberate overstatement.

PDG’s March 2026 update is a different layer. The company said a 210 MW multi-site acquisition in Mumbai and Hyderabad would take its total operating and planned India capacity to 1 GW. The 210 MW is part of that stated total, not another 210 MW to add on top—and the 1 GW wording is not a claim that one gigawatt is already operational.

Two other “500 MW” references are different again. A November 2025 Moneycontrol interview reported CapitaLand Investment’s ambition to approach 500 MW in India by the end of the decade; that is a forward target, not a current, audited inventory. CBRE’s 2026 outlook was reported as forecasting roughly 500 MW of new supply across major Indian markets during 2026; that is a market forecast, not a project row. Neither number proves a national double count.

What “double-counted” would mean

A parent total and one of its phases may both be accurate statements. The error occurs when both are entered as independent capacity in the same roll-up, for the same date, geography and capacity basis. If a 150 MW campus total already includes a 100 MW expansion phase, adding 150 and 100 produces a 250 MW entry even though the source describes one campus with a 150 MW endpoint. If a 500 MW regional programme contains 172 MW of Indian components, adding 500 and 172 as separate India projects misstates both geography and hierarchy.

That is different from two genuinely separate sites, from an expansion that is explicitly additional to a completed baseline, or from a company announcing an India target while a market consultant forecasts national supply. The same numerals do not create overlap. A defensible double-count finding needs evidence that the same physical asset or capacity phase appears twice in a particular dataset, on the same basis and at the same cut-off date.

For this review, “500 MW” is a search and audit prompt—not a verdict. We look for the reporting entity, geography, stage, capacity metric, date and parent/child relationship behind each number. Where a source gives no project bridge, the correct status is unresolved, not “duplicate”.

PDG’s 2024 programme: the 500 MW headline was regional

On 19 September 2024, PDG said it had acquired powered land and an expansion footprint for 500 MW across Mumbai, Chennai, Johor and Jakarta. The company described the programme as pan-Asia, with activity in India, Malaysia and Indonesia. It was not a 500 MW India-only campus, and “powered land” or planned expansion is not the same status as commissioned IT load.

The release attached several figures to that programme. Its named pieces are useful because they allow a reader to test the hierarchy instead of repeating the headline without context:

Adding the four stated incremental/site figures—100 + 72 + 200 + 100—gives 472 MW. PDG’s release labels the programme 500 MW but does not publish a complete line-item bridge to the headline. The difference is 28 MW on the face of the published arithmetic. It may reflect rounding, omitted detail or a different perimeter; the release does not tell us which. It is therefore a reconciliation question, not proof that 28 MW is missing, duplicated or misstated.

The same caution applies to the India line. The release says the company’s India footprint would rise to 230 MW, while the two named future pieces—MU1 at 150 MW and CH1 at 72 MW—sum to 222 MW. The public statement does not explain the remaining eight megawatts or state whether the measures use the same capacity basis. We preserve both figures as reported. We do not create an unlisted eight-megawatt site, force the figures to match, or convert one into the other.

PDG’s wording also matters for project stage. The 2024 announcement described powered land, an expansion footprint and a planned Chennai campus; it said the first phase of the 100 MW MU1 expansion was scheduled for delivery in 2026. A scheduled year is not confirmation of completion. The source set cited here does not establish that the full 150 MW MU1 campus, or the 72 MW CH1 plan, had reached commissioning, ready-for-service or commercial operation by 1 November 2026. We therefore keep those numbers as dated company-reported totals and plans rather than upgrading them on the strength of a target date alone.

The narrow conclusion is clear: 172 MW is the sum of the two named India additions in PDG’s 2024 regional programme. It is a parent–component relationship, not a second India-only 172 MW sitting on top of the 500 MW programme. The same release contains two arithmetic gaps that should remain visible until the company or a better project schedule supplies a bridge.

PDG’s 2026 update: 210 MW is inside the 1 GW India total

On 24 March 2026, PDG announced a multi-site acquisition adding 210 MW in Mumbai and Hyderabad. The company said the acquisition increased its total operating and planned capacity in India to 1 GW; it also said power was secured and key land approvals were in place for the acquired sites. Those are meaningful execution signals, but they do not mean that the 210 MW—or the full 1 GW—was already operational.

PDG’s announcement identifies Mumbai and Hyderabad but does not publish a complete facility-level reconciliation of the 210 MW acquisition against its earlier programmes or the 1 GW total. It also does not establish that the acquired sites are operational: the source describes power as secured and key land approvals as in place, with a path toward ready-for-service. Without named project IDs, per-site figures and dated status evidence in the same source set, the 210 MW should remain an attributed acquisition component inside the 1 GW operating-and-planned total—not a new live-capacity subtotal.

The arithmetic rule is simple: where a company says “210 MW added, taking the operating-and-planned portfolio to 1 GW,” the 210 MW is part of the endpoint. Do not publish 1,000 + 210 = 1,210 MW as PDG’s India total. Equally, do not add the 2024 500 MW regional programme or the earlier 230 MW India figure to the 2026 total without a dated, project-by-project bridge. Some older phases may already sit inside the later portfolio perimeter; the public releases reviewed here do not map every site ID across those announcements.

For an investor or vendor, “acquired capacity with power secured and land approvals” is not a synonym for construction completion, commissioning, ready-for-service status or a tenant’s contracted IT load. Keep the acquisition, permissions, works, energisation, commissioning and commercial-service milestones in separate fields.

CapitaLand’s reported 500 MW target: a market signal, not a site inventory

A separate 500 MW figure appeared in a Moneycontrol interview published 10 November 2025. The report described CapitaLand Investment’s India data-centre footprint as 245 MW across four metro areas and its ambition to reach nearly 500 MW by 2028–2030, including a further 250 MW. We treat this as an attributed growth target and a market signal. The interview does not provide a complete asset-by-asset bridge, a capacity-basis definition for every number, or a schedule proving that the target is operational today.

There is a more granular company source to compare with that interview. On 31 December 2025, CapitaLand Investment said its CapitaLand India Data Centre Fund would acquire a 20.2% interest in three data centres from CapitaLand India Trust (CLINT). The release described the transaction and listed each asset’s IT and gross-power capacity:

The last row is a transparent sum of the company’s four asset rows—not an independently verified portfolio total or a forecast. It demonstrates why a capacity figure must state its basis: the listed projects sum to 132 MW of IT power and 200 MW of gross power. Those numbers describe the same physical developments on two different measures; adding them would produce a meaningless 332 MW.

The announced 20.2% fund interest is an ownership relationship, not an extra 20.2% of physical capacity to add to the same buildings. CLI, CLINT and the private fund may each appear in sources because they manage, own or hold interests in the assets. For a site count, the building is counted once. For an ownership analysis, the ownership percentages and entity perimeter are reported separately.

On 15 October 2026, CLI said CapitaLand Data Centre Chennai 01 would deliver 53 MW gross power and 34 MW IT power upon completion, expected by the end of 2026. The company described it as jointly held by CLINT and the CapitaLand India Data Centre Fund. The same release said CLINT’s first data-centre development in Navi Mumbai had completed in 2025 and was fully leased to a global hyperscaler. The Chennai 53/34 MW figures are the same project row already present in the December 2025 table; they are an update, not another 53 MW to add to the table. A fully leased site and its tenant are also not two separate data-centre facilities.

There is a small source-precision point worth preserving: the current CapitaLand property page lists Chennai 01 at 54 MW under a generic “Power Capacity” field, while the 15 October release specifies 53 MW gross and 34 MW IT. The page does not explain whether 54 MW is rounding or another basis. We retain the values with their source labels and dates; we do not average them or call the one-megawatt difference an additional project.

Until the reported 245 MW base and near-500 MW target have a dated project bridge, they should not be added to the 200 MW gross asset table. The difference could reflect a broader project perimeter, the fourth Bengaluru site mentioned in company materials, development stages, rounding or differing metrics. The sources reviewed do not settle which explanation applies. This is why the target is useful as a market signal but unsuitable as a stand-alone operational-capacity count.

The other 500 MW: a market-wide forecast is not a project list

CBRE’s India Alternate Sectors Outlook 2026, dated 18 March 2026, was reported by ETCIO on 1 April as projecting about 500 MW of new data-centre supply across major Indian markets in calendar 2026, alongside roughly 30% year-on-year stock growth. That is a forecast for a market-wide supply period. It is not PDG’s regional land programme, not CapitaLand’s company target, and not an independently itemised list of 500 MW of projects.

Cushman & Wakefield’s 23 September 2026 H1 update uses a different lens. It reported 178 MW of new capacity across its top seven markets during H1 2026, taking operational capacity in those markets to 1,789 MW. It also identified 3,860 MW of upcoming supply through 2030, split into 509 MW under construction and 3,351 MW in planning. The report’s planning and construction categories are a multi-year pipeline, not 3,860 MW already live.

Market forecasts can include operator projects; an analyst’s forecast may be built from the same public announcements an individual operator publishes. But an overlap cannot be assumed without a methodology or project-level bridge. Do not add CBRE’s 500 MW expected 2026 supply to C&W’s 3,860 MW through-2030 pipeline, or either market estimate to a company’s 500 MW target, as if each were a separate set of buildings.

Six relationships that should not be collapsed into one total

These relationships are not edge cases. A modern campus may have an owner, an operating company, a private fund, a listed trust, a tenant, an announced phase and a utility connection described in separate documents. A public spreadsheet that stores just “company” and “MW” can make one asset look like several—or make a planned phase look operational. The remedy is not to discard every large figure. It is to preserve the relationships that explain what each figure includes.

A practical workflow for auditing a data-centre capacity claim

Whether the source says 500 MW, 1 GW or 3,860 MW, use the same sequence before adding it to a national or company total:

1. Define the perimeter. Is this a facility, campus, city, country, regional programme, company portfolio, listed trust or market forecast? Record which locations and entities are included.
2. Fix the time and stage. Attach a publication date, cut-off date and explicit stage: announced, land secured, approvals, under construction, commissioned, ready for service or commercially live. Do not move a project forward merely because its scheduled date has passed.
3. Label the metric. Distinguish IT load from gross facility power, utility sanction, MVA, power-availability potential and unqualified “capacity”. Keep the source’s exact term if the basis is not defined.
4. Give every physical asset and phase a stable ID. Store a parent ID for a campus or programme and a separate ID for each building/phase. Put the parent total in a separate roll-up field rather than duplicating it as a child site.
5. Map roles separately. Record developer, operator, asset owner, fund/trust, tenant and utility counterparty in different fields. A tenant’s name is not a new facility; a fund’s ownership interest is not a second campus.
6. State the arithmetic. Show which component rows add to a parent total, which are included in it, and which remain unresolved. Preserve provider/source differences rather than averaging unlike numbers.
7. Leave gaps visible. If a source says 500 MW but lists 472 MW of named pieces, record the 28 MW gap with its source, date and status. Do not invent a project, assume a duplicate, or silently force the rows to match.

A robust ledger would also store the exact source wording, link, capture date, confidence category and a correction history. For programme totals, the ledger should say whether the source includes powered land, phases in planning, contracted capacity, commissioned IT load or gross power. If the document does not say, the field should read “not stated”—not a guessed value.

What this audit does—and does not—establish

This audit establishes that the 2024 PDG 500 MW figure is regional, that two named India components sum to 172 MW, and that several published totals contain parent/phase relationships that should not be added twice. It also records two unbridged arithmetic differences in the 2024 release and a separate unqualified 54 MW property-page figure beside the 53 MW gross / 34 MW IT figures in the October 2026 company release.

It does not establish that PDG, CapitaLand, CBRE, Cushman & Wakefield or a market-data provider intentionally or actually duplicated 500 MW in an India-wide database. It does not provide a complete national project-by-project inventory or identify a specific duplicated row in another organisation’s dataset. It does not verify that every announced phase has reached construction, commissioning or commercial service. It does not assess whether any named project has a firm site-specific utility allocation.

Those limits matter. “There is a nesting risk” is supported by the source relationships; “the national market is overstated by 500 MW” would require a different dataset and a row-level reconciliation. One cannot be substituted for the other.

What readers can take away

For investors, the distinction between a market target and a live asset affects revenue timing, capital needs and valuation. For operators and site selectors, a power-secured or land-approved project is an execution milestone—not proof of commissioned IT load. For suppliers, a multi-gigawatt pipeline may signal future opportunity, but it is not a tender calendar or a purchase order. All three audiences benefit from a dated, phase-level record rather than a headline-only total.

The better question is not “Which 500 MW headline is right?” It is: What does this specific number include, what is its measurement basis, and which parent or child rows are already inside it? Until those questions are answered, keep the figures separate and describe the uncertainty plainly.

Frequently asked questions

Did PDG double-count 500 MW of Indian data-centre capacity?

The sources reviewed do not establish that. PDG’s September 2024 500 MW programme covered four markets across India, Malaysia and Indonesia. The 100 MW Mumbai expansion and 72 MW Chennai campus are named Indian components within that wider programme. No row-level national inventory has been produced here showing the same 500 MW entered twice.

Is PDG’s 500 MW an India-only project?

No. The company’s 19 September 2024 release described powered land and expansion across Mumbai, Chennai, Johor and Jakarta. The 500 MW headline is regional; the two named Indian additions total 172 MW.

Why do the named India components total 172 MW?

The release names a 100 MW expansion at Mumbai’s MU1 campus and a new 72 MW Chennai campus. Their arithmetic sum is 172 MW. The announcement does not label them as a common IT-load or gross-power measure, so 172 MW here means only the sum of the source’s stated numbers.

Was PDG’s 1 GW India portfolio operational in March 2026?

No. PDG described the 1 GW as total operating and planned capacity after a 210 MW acquisition in Mumbai and Hyderabad. The statement is not a claim that 1 GW was live or ready for service.

Is CapitaLand’s near-500 MW figure current operational capacity?

No. It was reported in a November 2025 interview as a forward ambition for 2028–2030. The sources reviewed do not provide a complete project-by-project bridge from the reported 245 MW base to the near-500 MW target or label every number on one common IT/gross basis.

Should a fully leased data centre be counted again under its tenant?

No. The host facility is the physical asset; the tenant relationship describes who leases or uses capacity in it. Track the tenant and its contracted load separately, and do not infer that the lease equals the building’s gross power or IT design capacity.

Can gross power and IT load be added together?

No. They are different measures of the same facility boundary. In the December 2025 CapitaLand table, the same listed projects sum to 132 MW IT and 200 MW gross power. Adding them to 332 MW would count the same developments on two bases.

Does CBRE’s 500 MW 2026 supply forecast overlap with company projects?

It may include some of the same projects, but the forecast is not a public project-level list that can be matched here. The forecast, company announcements and C&W’s through-2030 pipeline have different scope, status and horizon; do not add them unless a dated project bridge proves they are disjoint.

What evidence would prove a national double count?

A source-dated national inventory with stable project and phase IDs, the same capacity basis and cut-off date, and two entries that map to the same physical capacity would be needed. The audit should show the parent total, its component rows and the exact duplicated MW, while ruling out ownership, tenant, status and geography differences.

Sources and research method

This article was checked on 1 November 2026. It uses public company releases, company project pages and dated market reporting. Company-reported capacity is labelled as company-reported; market forecasts are attributed to the relevant research provider as reported; arithmetic sums are shown explicitly. No direct company outreach was made for this article. A reported target is not treated as a contract, committed construction schedule or operational inventory.

Where a source did not provide a common measurement basis or a project-level bridge, we left the figure unresolved. The 28 MW regional arithmetic difference, eight MW India-total difference, and 53/54 MW Chennai source variation are recorded as source-specific observations—not conclusions of error. The dates and statuses refer to the sources listed below and should be rechecked before use in a financial model or site decision.

• Princeton Digital Group — 500 MW powered-land and expansion programme across four Asian markets, 19 September 2024.
• Princeton Digital Group — MU1 expansion, CH1 and reported 230 MW India footprint, 19 September 2024.
• Princeton Digital Group — 210 MW India acquisition and 1 GW operating-plus-planned portfolio statement, 24 March 2026.
• Moneycontrol — reported CapitaLand India footprint and near-500 MW end-decade ambition, 10 November 2025.
• CapitaLand Investment — announced CIDCF 20.2% interests and three-project IT/gross power table, 31 December 2025.
• CapitaLand Investment — Chennai 01, 53 MW gross / 34 MW IT and Navi Mumbai lease status, 15 October 2026.
• CapitaLand — Chennai 01 property page and its unqualified 54 MW power-capacity field, checked 1 November 2026.
• CBRE — India Alternate Sectors Outlook 2026, 18 March 2026.
• ETCIO — CBRE’s reported 2026 new-supply forecast, 1 April 2026.
• Cushman & Wakefield — H1 2026 operating capacity and development pipeline, 23 September 2026.

Gridvik is an independent research firm. This article is informational and is not investment, legal, tax, engineering, energy-market or environmental advice. All figures are attributed to their sources; verify material claims with the relevant company, regulator or project counterparty.

Related Gridvik research

• Reliance’s Bhogapuram project: what the land records do—and do not—show
• Karnataka’s sustainable data-centre policy and grid-readiness questions
• India data-centre investment map: commitments, projects and stages
• India’s data-centre expansion: capacity, cities and project stages
• India’s data-centre power bottleneck: policy and grid-readiness signals

Browse all Gridvik Insights · Visit Gridvik Resources