$173B in Commitments, 1.8GW Operational: India’s Data-Centre Two Scoreboards

CBRE’s $173B is investment commitments since 2021; Cushman & Wakefield’s 1.789GW figure is operational capacity in its H1 2026 update. Don’t treat them as equivalent measures.

Gridvik Research · Public-source check through 12 October 2026 (IST) · Long-form market guide

Quick answer: CBRE’s September 2026 report puts India’s cumulative data-centre development commitments at about US$173 billion from 2021 through H1 2026. Its definition includes signed, proposed and ongoing development commitments. Cushman & Wakefield’s H1 update reports 1,789 MW of operational capacity—a capacity stock, not a money value. One figure is a cumulative dollar-denominated commitment measure; the other is megawatts operating at a point in time. They cannot be added, converted into one another or used as proof that US$173 billion has already been spent.

A third number belongs on a separate line: Cushman & Wakefield reports 3,860 MW of upcoming supply through 2030, split into 509 MW under construction and 3,351 MW planned. That is neither the 1,789 MW already operational nor a dollar commitment.

Research note: This article uses CBRE’s H1 2026 report and Cushman & Wakefield’s H1 2026 update, alongside the supplied Economic Times/PTI coverage. They are market-research estimates, not a single government census. Where the reports differ or leave a definition unclear, the article preserves the difference rather than averaging it away.

In this guide

1. The short answer: two scoreboards, not one
2. What CBRE’s US$173 billion counts
3. What the commitment number does not tell you
4. What Cushman & Wakefield’s 1,789 MW measures
5. India’s operational capacity by city
6. The 3,860 MW pipeline is a third scoreboard
7. Why the H1 2026 reports do not line up exactly
8. What these numbers mean for AI, cloud and colocation
9. The Singapore investor-category figure: useful, but limited
10. A practical checklist for reading India data-centre headlines
11. A four-column dashboard for investors, operators and suppliers
12. Glossary: investment, capacity and project stages
13. Frequently asked questions
14. Sources and further reading
15. Related Gridvik guides

The short answer: two scoreboards, not one

The headline pairs a stock of operating infrastructure with a cumulative pool of reported project commitments. That makes the story useful—but only if readers keep the measures separate.

The most common reading error is to compress all three into a single “market size” number. US dollars, operating megawatts and planned megawatts do not share a unit, date, project stage or denominator. A market can have large announced commitments and a smaller operating base at the same time: projects take years to secure sites, obtain power, finance, build, commission and ramp.

The sound conclusion is narrower: CBRE describes substantial investment interest and development commitments in India, while Cushman & Wakefield estimates that about 1.8 GW was operational in H1 2026 across its market coverage. The two reports do not provide a project-by-project bridge from each dollar commitment to a facility, phase, commissioning date and operating MW.

What CBRE’s US$173 billion counts

CBRE’s report, Beyond Growth: India’s Rise Among APAC’s Leading Data Centre Markets, says cumulative investment commitments in India’s data-centre sector reached about US$173 billion by the end of H1 2026, including roughly US$38 billion during January–June 2026. The Economic Times report supplied for this article carries the same figure and says CBRE expects more than 200 MW of additional capacity to become operational in H2 2026, with capacity seen approaching 2 GW by year-end.

The methodology note in CBRE’s report is essential: the commitments considered for its analysis include signed, proposed and ongoing data-centre development commitments. That is broader than an accounting ledger of paid invoices or construction expenditure. A proposal can be a meaningful signal of intent; a signed commitment can be firmer; an ongoing development can already be progressing. But they are not interchangeable stages, and the report’s headline total does not disclose a dollar-by-dollar schedule of cash paid, debt drawn or capex recognized.

The cumulative commitment series

CBRE presents the following cumulative totals. They are period-end cumulative levels, not separate annual investment amounts to be added together.

The step from US$135B at end-2025 to US$173B by H1 2026 is consistent with CBRE’s report of around US$38B in first-half commitments. It should still be described as a change in the reported commitment series, not automatically as US$38B of cash spent during six months. The source’s combined signed/proposed/ongoing definition prevents that shortcut.

How CBRE classifies the commitments

The report groups the cumulative total by development type and investor category. These are useful ways to understand how CBRE has organised the market estimate; they are not physical-capacity shares.

CBRE separately says hyperscale and colocation represented close to 62% of cumulative commitments and that more than 15% were tied to AI. The report’s category table assigns 16% to AI. Do not add “AI-linked” on top of every other category as if CBRE were describing a separate physical bucket; AI describes a workload/investment theme that may overlap with facility types in the real world.

These categories describe who CBRE classifies as making commitments—not how much capacity each group has commissioned, how much each participant has funded, or what returns it earns. A hyperscaler’s self-build, a developer’s colocation campus and a conglomerate-backed proposal can all sit in the commitment universe while being at different stages.

Only part of the US$173 billion is geographically mapped

CBRE says approximately 51% of the cumulative US$173B could be mapped to specific states and locations. Its state graphic names Andhra Pradesh, Maharashtra, Telangana, Tamil Nadu and Karnataka among the leading recipient states, with reported shares of 17%, 12%, 9%, 4% and 3%, respectively. Because the report says the mapped portion is incomplete, this should be read as a partial geography view, not a complete state-by-state account of all commitments. It does not make the unmapped balance disappear; it means the report has not assigned that full balance to a specific state/location in the published analysis.

That distinction matters for anyone searching for “data-centre investment by state.” The most visible proposed campus may be in one state, while a multi-city portfolio or commitment may be classified across several locations or remain unmapped. Do not treat the top-five graphic as a comprehensive project register, and do not multiply every state share by the full total unless the report’s denominator is explicit for the use being made.

What the commitment number does not tell you

US$173B is a striking headline, but it cannot answer every question that a developer, lender, supplier or investor might ask.

It is not the same as capex already deployed

A commitment can precede a final investment decision, land closing, financial close, equipment order, construction drawdown or commissioning. The figure may include proposed and ongoing development commitments as well as signed ones. Without a line-item status and cash schedule, readers cannot infer that the full US$173B has been spent or is irrevocably available.

It is not a single financing pool

A cumulative market estimate can combine commitments from multiple developers, operators, cloud platforms and other investors. It is not a fund raising one cheque, a government appropriation, a debt facility or an official FDI inflow table. The source report does not provide a project-by-project financing schedule that would let readers trace each commitment into equity paid, loans drawn or construction expenditure.

It is not an asset valuation or profit forecast

The report’s commitment total is not the value of India’s installed data-centre assets, annual sector revenue, EBITDA, cash flow, investor returns or land value. It also does not tell us how quickly a project will lease up, how much power it will consume or whether a particular investor will earn a target return.

It is not a capacity number

The figure cannot be divided by an operational MW total to produce a universal “cost per MW.” The numerator spans a multi-year period and mixed commitment stages; the denominator is an operating-stock snapshot. Costs also vary with land, location, power systems, redundancy, cooling, water, construction, financing, server ownership and whether a project includes associated infrastructure. A quotient would mix different scopes and dates, not reveal a reliable benchmark.

It is not an official national census

CBRE and Cushman & Wakefield are market-research providers. Their estimates are valuable, but their published figures should be attributed to the provider and read with the stated coverage and definitions. They are not the same as a public government registry of every data centre, each phase, its IT load and commissioning date.

It is not a ledger of individual announcements

A national cumulative estimate may already include named corporate commitments or developments. Adding a handful of large project announcements on top of US$173B risks double counting unless the researcher knows which ones CBRE included, how the announcement was scoped and whether related power, cloud or renewable spending sits inside or outside the project value.

What Cushman & Wakefield’s 1,789 MW measures

Cushman & Wakefield’s India Data Centre H1 2026 Update says the country’s leading data-centre markets added 178 MW in the first half of 2026, taking operational capacity to 1,789 MW as of H1. The report’s market overview says capacity and facility counts include both colocation and hyperscaler cloud service providers.

The update places India as APAC’s third-largest data-centre market by operational capacity, after China and Japan. It says operational capacity has grown from roughly 400 MW in 2019 to nearly 1.8 GW in H1 2026. That is a capacity-stock comparison across the report’s market definition—not a cash-flow series.

Cushman & Wakefield also reports 147 data centres, 33 operators and 14.9% colocation vacancy in its H1 overview. These details describe different dimensions:

• 1,789 MW is its operational-capacity estimate.
• 147 facilities / 33 providers describe the reported market footprint and operator count.
• 14.9% vacancy is a colocation-market indicator; it is not a vacancy rate applied to all 1,789 MW, including self-built hyperscaler capacity.

The report does not allow a reader to infer that every operational MW is leased, fully loaded, serving customers at full utilisation, or powered at the same time. Capacity is not the same as utilisation, demand, contracted volume or revenue.

India’s operational capacity by city

Cushman & Wakefield’s H1 2026 city table gives the following operating stock. The city shares are rounded in the source; its distribution assigns a further 2% to “Others.”

Mumbai is the clear anchor market in C&W’s figures, at 890 MW or about half of its reported stock. But “Mumbai” does not mean a single municipal boundary or one campus: the report discusses multiple clusters and micro-markets such as Powai, Thane–Belapur Road and Panvel. Chennai, Delhi NCR, Bengaluru, Hyderabad and Pune form a wider second tier, while Kolkata and other markets contribute smaller shares.

The vacancy column needs particular care. C&W reports different colocation vacancy levels by city. The overall 14.9% is not the share of all operational Indian capacity that is physically empty. Hyperscaler self-builds, operator-owned capacity and colocation suites have different commercial models and should not be collapsed into one occupancy percentage.

The 3,860 MW pipeline is a third scoreboard

A large pipeline can be useful evidence of expected development, but the stage split is crucial. C&W reports 3,860 MW of upcoming supply through 2030 across colocation and hyperscaler self-built facilities:

• 509 MW under construction; and
• 3,351 MW planned.

“Under construction” is a more advanced project stage than “planned,” but neither is the same as operational. A project still needs to finish construction, secure and energise power, complete commissioning and reach whatever commercial-service definition the operator uses. Planned capacity may shift, be phased, be revised or not proceed. The report’s 3,860 MW is therefore an upcoming-supply estimate, not another 3.86 GW added to today’s live base.

The city-level “upcoming” columns sum to less than the total because the report also includes other markets and uses rounded city distributions. Do not force the table to balance by assigning the residual to a city without source support.

CBRE’s separate H2 forecast says more than 200 MW was expected to become operational in the second half of 2026 and that capacity could approach 2 GW by year-end. That is a dated forecast, not proof—on 12 October—that the second-half supply had already gone live. It also should not be substituted for C&W’s 1,789 MW H1 estimate or added to C&W’s longer-term pipeline.

Why the H1 2026 reports do not line up exactly

CBRE’s H1 India snapshot describes roughly 1,750 MW of operational stock and about 130 MW of H1 supply additions. Cushman & Wakefield reports 1,789 MW operational and 178 MW added. The reported H1 stock estimates differ by around 40 MW, and the H1 additions differ by about 48 MW.

That is a real reporting difference, but the public documents reviewed do not provide a complete crosswalk explaining it. Possible reasons could include market coverage, project inclusion, cut-off dates, capacity basis or how a phase is counted—but those are questions to test, not explanations to assert. A careful article should not declare one provider “right” and the other “wrong” without comparable project-level inventories.

A second CBRE figure needs clarification

There is also an unbridged difference inside CBRE’s own September report. Its H1 India snapshot says operational stock was about 1,750 MW after around 130 MW of H1 additions. A separate APAC comparison table in the same PDF lists India’s live capacity as 1,100 MW as of Q1 2026. The report pages reviewed do not provide a crosswalk between that Q1 table and the H1 snapshot or explain the gap.

This does not by itself establish which number is incorrect; the two tables may reflect different coverage or data inputs. But absent a published reconciliation, the 1,100 MW entry should not be silently joined to the 1,750 MW H1 series. Gridvik’s publication check should ask CBRE to clarify the basis and retain the discrepancy if unresolved.

The right approach is to name each source, snapshot and reported definition:

Do not average the stock estimates to create a “consensus 1.77 GW.” An average would hide the underlying inventory questions and would not be an agreed market total.

What these numbers mean for AI, cloud and colocation

The two scoreboards give different context for the same broad market story. The commitment ledger suggests investor appetite for data-centre development; the operational stock tells us how much capacity C&W estimates is already operating. Neither alone proves that AI demand will absorb every planned MW.

AI is a meaningful theme, not a separate unit of capacity

CBRE says more than 15% of cumulative commitments were linked to AI, and its category table places 16% in an AI grouping. That helps show that AI is influencing investment narratives and development plans. It does not mean 16% of India’s 1,789 MW operational stock is dedicated to AI, nor does it tell us how much power is currently consumed by AI workloads.

AI-ready campuses may be planned for higher rack densities, larger power blocks and specialised cooling. But the report’s commitment category is a financial-development classification, not a meter reading of live AI compute. Some AI facilities may be self-built hyperscaler projects; some may be leased; some may still be planned. The capacity and money ledgers need phase-level evidence before they can be matched.

Hyperscale and colocation have different commercial routes

CBRE’s type table allocates 30% of cumulative commitments to hyperscale and 32% to colocation. These models are related but not identical. A hyperscaler may build captive capacity for its cloud workloads; a colocation provider may develop a facility and lease space or power to multiple customers. Their revenue, financing, utilisation and customer-contract patterns differ.

C&W explicitly says its operational-capacity and facility counts include both colocation and hyperscaler cloud service providers. That makes its 1,789 MW broader than a colocation-only inventory. Its vacancy statistic, however, is a colocation measure. This is why a headline like “14.9% vacancy” cannot be applied to every MW in the total.

Demand drivers are not all the same as contracted demand

The research providers cite cloud adoption, AI workloads, enterprise digitisation and digital consumption as sector drivers. Those are market-demand explanations, not a list of signed tenant contracts for every facility in the pipeline. Investors and suppliers should ask whether a project has a named customer, an executed lease, a power agreement, land control and a construction schedule—not infer those milestones from sector-wide demand trends.

The Singapore investor-category figure: useful, but limited

CBRE’s report includes an investor-origin/category table for cumulative commitments from 2021 through H1 2026. It lists India at 38%, the United States at 27%, Australia at 20%, Singapore at 8%, Japan at 3%, the United Kingdom at 1% and other categories at 4%.

That is a potentially relevant signal for Singapore-based investors and operators tracking India’s data-centre market. It is not a project-by-project source-of-funds trace. The table does not, by itself, establish the ultimate beneficial owner behind every commitment, show when cash crossed a border, distinguish all financing vehicles, or prove that capital was withdrawn from a Singapore data-centre project and redirected to India.

Use the figure as CBRE’s classification of investor origin/category in its commitment dataset. Do not re-label it as Singapore FDI, “Singapore money deployed,” or a measured Singapore-to-India capital shift. For the stronger capital-flow claim, the evidence would need to identify the same investor, a reduced or cancelled Singapore project, the released amount and a financing trail into a named Indian project.

This boundary is important because a Singapore-headquartered firm can invest in India without displacing a Singapore build; it may be expanding in both markets, using a multi-country fund, borrowing locally or deploying capital from a different entity. Corporate location and funding origin are not automatically the same thing.

A practical checklist for reading India data-centre headlines

Use this checklist before quoting a market number in a pitch deck, article, investment memo or supplier pipeline.

1. Identify the metric

Is the source reporting a commitment, proposal, capex budget, FDI flow, asset valuation, operational MW, contracted capacity, under-construction supply or planned pipeline? If the article uses the word “investment,” find the source’s actual definition.

2. Read the date and time window

“Since 2021” is cumulative. “H1 2026” is a six-month snapshot. “By 2030” is a forecast horizon. “As of H1” is not the same as “currently live on 12 October.” Put the date beside the number, not in a footnote readers will miss.

3. Check the project stage

A signed, proposed or ongoing commitment is not equivalent to cash paid. A planned facility is not under construction. Under construction is not commissioned. Commissioned does not always mean commercially serving tenants. Label each stage.

4. Check geographic coverage

Does the estimate cover all India, seven leading markets, a list of metro cities, a set of named projects or only the portion mappable to states? CBRE says about 51% of its commitments could be geographically mapped; that makes the state view incomplete.

5. Check capacity basis

Is MW described as operational capacity, IT load, utility capacity or gross power? If the source does not specify, quote its term and avoid converting it to another basis. Do not compare figures from separate providers as exact equivalents without a methodology bridge.

6. Separate hyperscaler self-build from colocation

Both can count in national operational capacity. But self-built capacity may not be offered to third-party tenants, while colocation vacancy refers to leasable capacity. Ask what asset types are included in the total and in any occupancy statistic.

7. Check overlap and phase nesting

A campus target may include several phases. A parent company target may include its Indian projects. A state project pipeline may include an operator’s national commitment. Do not add parent totals and phases, or individual announcements and a market-wide estimate, without a source crosswalk.

8. Look for the evidence bridge

For a commitment-to-operational story, the useful bridge is project-level: named sponsor; specific project; phase; land; power; financing; construction; commissioning; and reported live capacity. If one of those links is absent, say so.

9. Test the source’s definitions before comparing it with another provider

Two consultants can use different project lists, city boundaries, cut-offs and capacity methods. A discrepancy is a reason to inspect the methodology—not to select the larger number because it makes a stronger headline.

A four-column dashboard for investors, operators and suppliers

A useful internal dashboard keeps financial intent, construction progress, live stock and commercial performance apart.

For a supplier, the dashboard helps prevent a large “India data centre pipeline” from being mistaken for a near-term order book. For a lender, it separates sponsor intent from assets that can support cash flow. For an operator, it distinguishes physical capacity from customer demand and power availability. For an investor, it makes it easier to compare a high-level market commitment series with actual project delivery.

Reading the market without overstating it

The available evidence supports a strong growth narrative, but the exact phrasing matters:

• Supportable: CBRE estimates roughly US$173B in cumulative signed, proposed and ongoing data-centre development commitments since 2021 through H1 2026.
• Supportable: Cushman & Wakefield estimates 1,789 MW of operational capacity in H1 2026 across the market scope described in its report.
• Supportable: Cushman & Wakefield lists 3,860 MW of upcoming supply through 2030, of which 509 MW is under construction and 3,351 MW is planned.
• Not established: US$173B has already been spent or financed.
• Not established: every planned MW will be delivered by 2030.
• Not established: all operational capacity is available to colocation customers or fully utilised.
• Not established: CBRE’s Singapore investor-category share proves that capital moved from Singapore to India.
• Not reconciled: the provider differences in H1 additions and capacity inventories, or CBRE’s separate 1,100 MW Q1 APAC-table entry versus its ~1,750 MW H1 India snapshot.

The responsible verdict is not that one scoreboard is “better.” The scoreboards are useful for different questions: CBRE’s commitment series tracks investment intent and development activity; C&W’s operational figure estimates infrastructure already in service; C&W’s pipeline describes capacity at earlier stages. A good India data-centre market analysis needs all three, clearly labeled, and a project-level bridge before tying one to another.

Glossary: investment, capacity and project stages

Frequently asked questions

How much has India’s data-centre sector attracted since 2021?

CBRE’s H1 2026 report estimates cumulative data-centre development commitments of about US$173 billion from 2021 through H1 2026, including around US$38 billion in H1 2026. Its published definition includes signed, proposed and ongoing development commitments, so it should not be described as US$173 billion of cash already spent.

How much operational data-centre capacity does India have in H1 2026?

Cushman & Wakefield reports 1,789 MW in its H1 2026 update. CBRE’s H1 India snapshot gives roughly 1,750 MW and reports about 130 MW added in H1. The figures are close but not identical, and the providers do not publish a project-level reconciliation in the materials reviewed. CBRE’s same report also has a separate Q1 APAC table showing 1,100 MW for India; that entry is not bridged to the H1 snapshot.

Is CBRE’s US$173 billion investment figure money already invested?

Not on the definition found in the report. CBRE says the analysis includes signed, proposed and ongoing data-centre development commitments. It does not present the whole amount as paid capex or provide a cash-drawdown ledger.

What is India’s data-centre pipeline through 2030?

Cushman & Wakefield reports 3,860 MW of upcoming supply through 2030: 509 MW under construction and 3,351 MW planned. The pipeline should be kept separate from the 1,789 MW operational stock.

Which city has the most operational data-centre capacity?

In Cushman & Wakefield’s H1 2026 figures, Mumbai leads with 890 MW, about 50% of the reported distribution. Chennai is next at 214 MW, followed by Delhi NCR at 183 MW, Bengaluru at 162 MW, Hyderabad at 151 MW, Pune at 133 MW and Kolkata at 22 MW. A further 2% is allocated to “Others” in the source distribution.

Can US$173 billion be converted into megawatts?

Not from these reports. The commitment total covers multiple years and stages, while operational MW is a capacity stock. The project mix, costs, power systems, scope and delivery dates are not aligned well enough to derive a reliable cost-per-MW conversion.

Does the 14.9% vacancy figure mean 85.1% of India’s capacity is occupied?

No. Cushman & Wakefield labels the figure as colocation vacancy. It should not be applied to the full operational total, which includes both colocation and hyperscaler cloud service providers.

Does CBRE’s 8% Singapore figure mean Singapore has invested 8% of the money in Indian data centres?

It means CBRE’s investor-category/origin table assigns 8% of the cumulative commitment series to Singapore. The report table is not, by itself, an official FDI statistic, a project-level financing trail or proof of ultimate beneficial ownership or capital diverted from Singapore.

Is India’s capacity already 2 GW?

CBRE’s H1 report and October coverage described a year-end 2026 outlook approaching 2 GW, supported by an expectation of more than 200 MW of H2 supply. That was a forecast at the time of publication. It should not be stated as an achieved figure unless a later dated source confirms it.

Why do two firms give different operational-capacity numbers?

Market research providers can differ in cut-off dates, project inventories, city boundaries, capacity basis and inclusion criteria. In this case, CBRE and C&W report different H1 additions and slightly different H1 stock estimates; the reviewed reports do not provide a full crosswalk. CBRE’s own report also has a separate 1,100 MW Q1 APAC table entry that it does not reconcile with the ~1,750 MW H1 snapshot. Preserve those distinctions and ask the providers for clarification rather than averaging the figures.

Method, scope and correction note

This is a source-led comparison of CBRE’s Beyond Growth: India’s Rise Among APAC’s Leading Data Centre Markets (September 2026) and Cushman & Wakefield’s India Data Centre H1 2026 Update, checked through 12 October 2026 (IST). The supplied Economic Times/PTI report is used as secondary coverage of CBRE’s H1 announcement. No project-by-project financing audit was conducted; no national facility census was created; and no investment-to-MW conversion is attempted.

The article retains the providers’ terminology and separates cumulative commitments, operating stock and future pipeline. CBRE’s commitment definition and geographic-mapping caveat are reported as stated in its material. The unbridged 1,100 MW/1,750 MW CBRE figures and provider-to-provider H1 differences are disclosed rather than harmonized. The article does not offer an investment recommendation or claim that Singapore capital was diverted to India.

No company, executive or research-provider representative has been contacted or tagged for this draft. Before publication, CBRE and Cushman & Wakefield should have a fair opportunity to clarify the definitions, coverage and capacity discrepancies described here. Any response or non-response should be represented accurately. A headline or disclaimer does not guarantee legal immunity; normal editorial and legal review may still be appropriate.

Sources and further reading

1. CBRE, Beyond Growth: India’s Rise Among APAC’s Leading Data Centre Markets, September 2026 — report page and full report PDF — cumulative commitments, methodology note, market snapshot, investor and development-type categories, geographic-mapping note and operator-survey material.
2. The Economic Times/PTI, “India’s data centre sector attracted USD 173 bn since 2021, capacity seen 2 GW by year-end: CBRE,” 6 October 2026 — accessible secondary summary of CBRE’s H1 commitments, H1 additions and year-end outlook.
3. Cushman & Wakefield, India Data Centre H1 2026 Update — report page and full H1 report PDF — operational stock, city distribution, capacity/vacancy indicators and 3,860 MW upcoming-supply split.
4. Cushman & Wakefield, “India’s Data Centre development Pipeline through 2030 Reaches 3,860 MW,” 23 September 2026 — public summary of the H1 update and operational/pipeline totals.
5. Cushman & Wakefield, 2026 Asia Pacific Data Centre Investment Landscape — broader regional investment context; APAC estimates are not a substitute for India’s H1 operating inventory.

Related Gridvik guides

For the next layer of context, continue with these public Gridvik guides confirmed live on 12 October 2026:

• India’s Data Centre Boom and Expansion (2026–2030): Capacity, Cities, AI, Power and What Comes Next — broad market context for capacity, forecasts and delivery constraints.
• India Data Centre Investment Map 2026: Big-Tech Commitments, Operators, Finance and State Incentives — distinguishes investment announcements from deployed capital and operating assets.
• India’s Data Centre Power Bottleneck: Grid Readiness, State Policy and the Pre-Tender Window — explains why a state power signal is not proof of project-level grid readiness.
• CoreWeave’s 240 MW India AI Data Centre Plan: What the AdaniConneX Taloja Campus Means — a project-specific example of announced capacity and future phases, not present operating supply.

Browse all Gridvik Insights · Visit Gridvik Resources