India’s Data Centre Power Bottleneck: Grid Readiness, State Policy and the Pre-Tender Window

Executive summary

Power is one of the most important constraints on India’s data-centre expansion, but “power is the bottleneck” is too simple unless the claim is tied to a location, a phase and a project schedule. A facility needs more than generation in the state: it needs a confirmed site, a viable grid connection, transmission and substation capacity, an acceptable tariff, operational redundancy and a procurement plan. It also needs enough customer demand to justify the capital.

State policy can affect the economics and speed of a project. Gujarat’s 2026–29 policy sets a 7.5 GW capacity target and a stated investment ambition of ₹6 lakh crore, with eligibility conditions and a renewable-energy requirement. Uttar Pradesh’s 2026 policy targets mobilisation of more than ₹2 lakh crore and lists incentives and infrastructure measures. Those figures are policy targets, not operating capacity or money already spent.

For suppliers, the “early window” is real as a practical principle: qualification, design decisions and long-lead packages can be addressed before a public tender appears. But no credible general rule says every supplier has exactly two quarters. Use verifiable milestones, package-specific lead times, current sources and direct buyer confirmation. Mark an inferred date as an estimate, not a fact.

Contents

1. Executive summary
2. What the power bottleneck actually means
3. Why grid readiness is different from state generation
4. A data centre’s power stack
5. How a project can procure and manage power
6. Renewables, storage and sustainability claims
7. State policy: useful lever, not substitute for infrastructure
8. The two-quarter buying-window claim: what holds up and what does not
9. A better procurement-signal framework for vendors
10. Project examples and what they do—and do not—prove
11. A practical power-readiness checklist
12. Power readiness by project phase
13. Make power readiness visible without inventing a score
14. What the CEA review could clarify—and what remains unknown
15. Four playbooks for different decision-makers
16. Build a power-aware project timeline
17. Turn policy and power information into a vendor plan
18. Recommended labels for power and project claims
19. Sources and publication discipline
20. Frequently asked questions
21. Sources and further reading

What the power bottleneck actually means

A data centre is a large, continuous electricity user. Its power requirement supports IT equipment and the systems that keep that equipment available: cooling, electrical conversion, lighting, security, controls and other facility loads. A development can be announced, permitted, financed or under construction and still lack the power arrangement needed to operate its planned phase.

“Power bottleneck” can refer to several different issues:

• Generation adequacy: Is enough electricity expected to be available in the wider system?
• Transmission: Can electricity be moved to the relevant region and substation?
• Distribution and connection: Can the local network deliver the required load to the site?
• Firmness and reliability: Can the buyer receive a dependable supply, including through outages and maintenance?
• Commercial terms: Are tariffs, demand charges, network charges and contracts workable?
• Renewable procurement: Can the project obtain the intended clean-energy supply under applicable rules?
• Timing: Will equipment, approvals and the connection be ready before the intended commissioning date?
• Site dependencies: Are land, water, permits, fibre and construction ready on a compatible schedule?

A state may have substantial power generation capacity but still require grid upgrades or transmission investment to serve a particular campus. Conversely, a project can secure a promising connection plan but face cost, tariff or renewable-energy questions. A state-wide generation number is not proof that a project parcel has firm capacity.

Power capacity, energy use and IT load are different

A megawatt (MW) describes power at a point in time; a megawatt-hour (MWh) describes energy used over an hour. A gigawatt (GW) is 1,000 MW. A data-centre project can use capacity figures that refer to IT load, site load, grid connection or a future campus objective. Those terms should not be mixed.

IT load is the power supplied to computing, storage and networking equipment, according to the source’s definition. Facility load includes IT plus supporting systems. Grid demand can reflect a utility connection or maximum draw. Annual energy measures cumulative consumption over time. These measures answer different questions.

If a project is described as a 100 MW IT facility, its grid requirement may be higher, depending on cooling and electrical design. A project’s load factor, redundancy configuration, AI density and operating pattern also matter. Do not convert one metric into another without a clear technical basis or the source’s own definition.

Why power planning is now getting more attention

The number and scale of announced data-centre and AI projects has prompted more attention from grid planners and policymakers. In late September 2026, The Economic Times reported that the Central Electricity Authority had formed a committee to assess the sector’s impact on power demand and grid stability. The report cited a CEA estimate that data-centre electricity demand could reach about 17 GW by 2031–32 and described concerns raised by regional power-planning bodies about large, continuous loads.

That is a planning signal, not proof of a nationwide shortage or a forecast for any one state. The estimate and committee work should be attributed to the reported source and refreshed as official findings are published. A project team still needs site-level grid studies and utility discussions.

Why grid readiness is different from state generation

A data-centre campus needs usable power at a particular place and time. The path from generation to a facility can include generation assets, high-voltage transmission, regional substations, state transmission networks, distribution feeders and a site substation. Each link has technical, commercial, regulatory and scheduling dependencies.

The generation question

Generation adequacy asks whether the broader electricity system can supply expected demand. This can include conventional generation, renewables, storage, imports and demand response. A state’s installed generation capacity does not necessarily equal the power available to a new facility at every hour, nor does it establish a delivery contract.

The transmission question

Transmission capacity moves bulk electricity across regions and to substations. A new data-centre cluster can create a concentrated load that requires reinforcement. New high-voltage lines, substations and transformers have their own planning and construction lead times. A campus schedule should identify who is responsible for each upgrade and what milestones must be met before energisation.

The distribution and connection question

Even if power is available in the region, the local network and connection point must be able to support the proposed load. The project may need dedicated feeders, a new substation or staged capacity. A site with “access to the grid” may still have a long path to the full planned load. Connection approval, engineering studies, right-of-way, equipment supply and testing all matter.

The reliability question

A data centre generally needs power continuity and controlled response to outages. Utility supply is only one layer. Operators design internal electrical systems with redundancy, uninterruptible power, backup generation, transfer controls and maintenance procedures. The exact architecture depends on the service-level commitment and customer requirements.

The tariff and cost question

Power cost includes more than the headline energy rate. The project may incur demand charges, transmission and wheeling charges, open-access fees, taxes, balancing costs, backup fuel and costs associated with redundancy. The structure depends on state regulation, the supply route and contract. A state-level subsidy can change some economics, but the project must verify eligibility, duration, caps and disbursement.

The timeline question

Power delivery is a critical-path schedule. A project can have land and a customer but be unable to commission until a transformer, feeder or substation upgrade is complete. Conversely, a power connection might be available before the building and IT equipment are ready. A project dashboard should track each dependency and distinguish the promised date from a confirmed milestone.

A data centre’s power stack

The power system is best viewed as layers that work together, not as a single utility bill.

IT power and facility power

The power supplied to IT equipment is not the whole facility load. Cooling and electrical losses add to it. A common efficiency measure is Power Usage Effectiveness (PUE), the ratio of total facility energy to IT equipment energy over a stated period. PUE is useful but incomplete: it varies with climate and utilisation and does not independently describe water use, carbon intensity or renewable matching.

A low PUE does not automatically mean a project is sustainable or power-secure. A project may have efficient cooling but weak grid redundancy; it may purchase renewable certificates without hourly clean power; or it may rely on water-intensive cooling in a stressed region. Each claim needs its own metric and boundary.

Redundancy: resilience inside the campus

The terms N, N+1 and 2N describe different forms of redundancy in a system, but they are not a complete description of an operator’s resilience. N represents the capacity required to support the load; N+1 adds a spare component or capacity; 2N indicates two full independent paths in a particular design context. The precise meaning depends on the system and project specification.

Redundant site equipment cannot compensate for every shared external risk. Two internal power paths may still rely on the same utility substation or transmission corridor. A generator may be limited by fuel supply or environmental conditions. Resilience reviews should examine dependencies from the grid to the rack and test how the facility responds to failures and maintenance.

How a project can procure and manage power

The exact legal and commercial route varies by state and project. Common approaches include utility/distribution company supply, open-access procurement, captive or group-captive generation, renewable PPAs, onsite generation, storage and combinations of these. This section provides a high-level overview—not a legal opinion or assurance that any route is available to a specific project.

Utility or distribution-company supply

A project may obtain supply from its local distribution company under applicable tariff and connection rules. The developer needs to confirm load sanction, connection voltage, feeder design, reliability arrangements, tariff components, security deposits, load ramp and any infrastructure contributions. A large load may require a new connection or upgrades. The project should obtain current written information from the utility and regulator rather than relying on an old tariff table or a broker’s summary.

Open access

Open access can allow eligible consumers to procure electricity from a generator over the network, subject to applicable central and state rules, approvals, charges, scheduling and contracts. The route may involve multiple agreements and settlement responsibilities. It can improve access to renewable power or competitive supply, but charges and eligibility can change. A project should have legal and energy-market specialists review current regulations and the specific transaction.

Captive and group-captive supply

A company may participate in or develop a power-generation asset for its own consumption under applicable law and rules. Ownership, consumption, qualification and compliance tests can matter. The structure should be assessed by counsel and energy experts; a marketing description of “captive power” does not establish that the arrangement qualifies under current rules.

Power purchase agreements

A power purchase agreement sets out commercial and delivery terms between the buyer and generator or supplier. Important variables include energy volume, tariff, term, start date, delivery point, shape, renewable attributes, curtailment, change in law, credit support and termination rights. A signed PPA does not by itself prove that transmission capacity, grid approvals or the facility’s backup arrangements are complete.

Onsite generation and backup

Onsite generation can support reliability or local supply, but it has permitting, fuel, emissions, maintenance and cost implications. Backup generators are typically part of resilience design rather than a substitute for a reliable long-term energy supply. A project should state whether a generator is for emergency backup, routine operation or a broader supply plan.

Batteries and other storage

Batteries can provide UPS functionality, shift energy, support renewable integration or help manage certain grid conditions. Their usable duration, power rating, degradation, safety, recharge, fire protection and control strategy matter. A battery’s MW rating does not reveal how many hours it can supply; both power and energy capacity should be disclosed.

The right question is portfolio design

A robust power plan often combines grid supply, contracts, renewable energy, storage and internal backup. The project should model failure cases and economic trade-offs rather than assume one resource will solve every problem. The design must match the customer’s uptime requirements, power profile, site conditions, regulations and phase schedule.

Renewables, storage and sustainability claims

Many data-centre companies seek renewable power because it can support corporate climate goals and long-term energy procurement. But “renewable powered” can mean different things. It may refer to an annual matching claim, a power purchase agreement, a physical supply arrangement, onsite generation, renewable certificates or a company-wide portfolio.

Annual matching versus hourly supply

A company may match its annual electricity consumption with renewable generation over a year. That does not necessarily mean renewable power is physically delivered to the facility in every hour. Hourly matching requires more detailed alignment of generation, consumption, storage and grid conditions. When communicating a claim, specify the method and reporting boundary.

Renewable capacity is not a facility allocation

A renewable project’s capacity may serve multiple customers, a state grid or a broader company portfolio. The project’s nameplate generation is not necessarily dedicated to one data-centre campus. A data-centre article should state whether the source confirms a dedicated contract, a shared portfolio or an aspiration.

Storage is not a substitute for all firm power

Storage can shift energy and support reliability, but its duration and cycling constraints matter. A four-hour battery, for example, does not provide an unlimited supply through a multi-day generation shortfall. Projects may need a mix of generation, grid supply, storage and demand management.

Water and power decisions interact

Cooling choices create trade-offs. A system with lower water consumption may use more electricity; a water-based design may improve energy efficiency but require water sourcing, treatment and discharge planning. Some projects use reclaimed water, desalination or closed-loop systems, but the impact depends on the local context and system design. A coastal site with desalination still needs to address energy use, brine management, permitting and marine impacts.

A practical sustainability evidence table

State policy: useful lever, not substitute for infrastructure

States are competing to attract data-centre investment with policy frameworks, incentives, approvals and infrastructure facilitation. The policy can improve a project’s economics or help coordinate agencies, but it cannot replace utility capacity, land, financing, customer demand or execution. Incentives also come with eligibility tests, caps, application requirements and time limits.

Gujarat’s 2026–29 policy

Gujarat’s Viksit Gujarat Data Centre Policy 2026–29 sets a stated target of 7.5 GW of data-centre capacity and an investment ambition of ₹6 lakh crore. NASSCOM’s analysis of the policy describes eligibility conditions focused on large projects, including a minimum approved IT load of 150 MW for the relevant policy benefits, and a requirement that at least 51% of electricity consumed for core operations come from green or renewable sources. It also describes fiscal and facilitation measures, including power-related support, duty reimbursement and assistance for eligible projects.

A developer should consult the official Gujarat policy document and current implementation guidelines before modelling benefits. The 7.5 GW target is not installed capacity, and the ₹6 lakh crore figure is a policy ambition, not cash committed to each project. Incentive value depends on eligibility, application, caps, timing, reimbursable costs and compliance.

Uttar Pradesh’s 2026 policy

Invest UP’s Uttar Pradesh Data Centre Policy 2026 states a target to mobilise more than ₹2 lakh crore. The official policy page describes support for eligible data-centre units and parks, including capital and interest-related assistance, land and duty provisions, power arrangements and renewable-energy objectives. Project teams should rely on the policy text and official notifications for exact eligibility, limits, application dates and payment conditions. The new framework should not be mixed indiscriminately with terms from the earlier UP policy.

Policy comparison should be project-specific

A policy comparison should ask:

A company should run a project model with the benefit excluded, then add only support for which it has a credible eligibility path. It should not base a site decision on a press summary or assume a future reimbursement is equivalent to cash at construction close.

Policy targets and private projects belong in separate columns

A state policy can target multiple gigawatts, while only a portion of proposals may advance. The public record may contain expressions of interest, MoUs, planning approvals, land allocations, under-construction projects and live facilities. A transparent tracker reports these stages separately.

For a business selling to project developers, state incentives can create an account-planning signal. They do not prove the customer has budget for the supplier’s package. Vendors should still find the project entity, procurement route, design consultant or EPC, qualification path and package status.

The two-quarter buying-window claim: what holds up and what does not

Some project teams may use “go-live minus two quarters” as a planning heuristic, and some procurement decisions may precede a public tender. But public evidence does not establish the rule as a universal pattern across Indian data centres, vendors and packages. A fixed two-quarter claim should not be presented as a fact.

Why a single lead time cannot fit every package

A generator, UPS, switchgear, transformer, chiller, cooling tower, rack, cable, security system and software platform have different design, approval, manufacturing and installation cycles. A package may be specified in concept design, detailed design, tender, EPC procurement or customer fit-out. Different project delivery models assign buying authority to the owner, operator, EPC, consultant, integrator or tenant.

Some suppliers can influence a design early, before a public tender. Other packages are not selected until the design is mature. A major long-lead item might be ordered well ahead of go-live; a service contract can be sourced closer to operation; a software product may be evaluated as a pilot long before procurement. A universal two-quarter period obscures the real sequence.

What “spec lock” means—and why timing can vary

Specifications can become difficult to change at different points. A consultant may set technical requirements before the final bill of quantities. An operator may approve a vendor list before the tender. An EPC may offer an approved design with limited brand flexibility. A customer may require a later security or interoperability review. A project can keep some alternatives open while locking others.

“Spec lock” should therefore be tied to a package and a document, such as issued-for-construction drawings, a bid package, approved-vendor list or customer acceptance criteria. If no such evidence is public, describe the stage as unknown. Do not tell a supplier that a named brand is pre-qualified or that a design is locked without a source.

Public tender date versus actual buying milestone

Not all data-centre procurement is published as an open tender. A private operator may invite bids from an approved shortlist, use an EPC framework or buy through an existing contract. A public RFP can be one milestone, but its absence does not prove there is no procurement. The correct vendor route depends on the entity and project.

A public data point can indicate project maturity but usually cannot reveal the internal budget approval or package schedule. A vendor should confirm through buyer conversations and public notices. If a buying window is estimated from a target go-live date, the estimate should explain the assumed lead time and show uncertainty.

Replace “two quarters” with a milestone range

Use a sequence such as:

1. Public project announcement.
2. Site and power plan identified.
3. Design consultant or EPC named, if public.
4. Phase, capacity metric and target delivery period confirmed.
5. Package requirements or qualification route visible.
6. RFI, EOI, tender or shortlist stage identified.
7. Award, purchase order or construction activity verified.
8. Commissioning and operational status verified.

A supplier may act at different points depending on product. A design workshop can be useful before a tender; a formal quote may make sense only after a scope and buyer are known. A project’s “go-live” date can also change, so a window based on that date should be refreshed when new evidence appears.

A better procurement-signal framework for vendors

A vendor needs a pipeline that separates public signals from verified buyer demand. A large project can be a good reason to research an account, but not yet a sales opportunity. The following framework supports honest prioritisation.

Four evidence levels

These levels are not a universal sales methodology; they are a simple way to prevent a press release from being mistaken for a purchase order. The vendor should define what evidence moves a record between stages and keep an audit trail.

Track facts, estimates and unknowns separately

A useful account record may include project, location, owner, operator, customer, capacity definition, phase, source date, construction status, package, buyer role, qualification, tender route, expected timing, confidence and next verification date. It should distinguish:

• Fact: Directly stated in a current, attributable source.
• Buyer-confirmed: Confirmed by an authorised buyer or procurement contact.
• Estimate: Derived from an explicitly stated assumption or historical lead time.
• Unknown: Not publicly confirmed.

A press release may be authoritative for what the company announced, but not necessarily independent evidence of delivery. A market report may be authoritative for its own estimate, not the project’s internal schedule. A salesperson’s estimate should be labelled as such.

Estimate a package window only when the inputs exist

A team can create an internal planning range by starting with a verified phase milestone and applying its own package-specific lead-time history. For example: “If the published target commissioning period holds, and if our transformer procurement normally requires the company’s recorded lead time, then we should seek design qualification by this range.” This is an internal planning assumption, not a public tender date.

The estimate needs at least three inputs: a reliable phase milestone, the specific package’s lead-time data, and the buyer’s procurement route. If one is missing, show a wider range or leave the date unknown. Do not infer the same lead time for UPS, cooling, fiber and software.

Questions to ask before approaching a buyer

• Which company owns the project and which entity is the operating customer?
• Who controls this package: owner, EPC, consultant, operator, tenant or integrator?
• Is the facility at concept, design, procurement, construction or commissioning stage?
• What exact specification and interface requirements apply?
• Is the product on an approved-vendor list or does it need evaluation?
• What certifications, security reviews, safety records and local service coverage are required?
• Is the buyer issuing an RFI, EOI, tender or private invitation?
• What are the budget, payment, warranty, delivery and acceptance conditions?
• What is the source and date for the claimed timeline?
• What would disqualify the opportunity?

This preparation reduces generic outreach. It also prevents vendors from calling an operations contact a budget owner without evidence, or from describing a project as a confirmed customer when no relationship exists.

A no-bid decision is part of good pipeline management

A supplier can decline an opportunity if the scope is unclear, the schedule is unrealistic, payment risk is too high, the project status is not credible, or the company lacks the required service capability. A disciplined no-bid decision preserves engineering capacity and protects working capital. Sales teams should not be rewarded only for the number of announcements entered into a CRM.

Project examples and what they do—and do not—prove

Google’s Visakhapatnam AI hub

Google announced an approximately US$15 billion commitment over 2026–2030 for an AI hub in Visakhapatnam. Its public descriptions include data-centre operations, energy and fibre components. A construction update in April 2026 reported groundbreaking and described work with local partners on new transmission lines, clean-energy generation and energy-storage systems. This is evidence that power and connectivity are part of the hub strategy and that work has begun on the broader programme.

The public announcement does not disclose every phase’s power allocation, project-level renewable contract, facility capex, grid interconnection agreement, equipment tender or live MW. A vendor can treat the hub as a high-priority research account, but should verify the package, delivery phase and buying route before forecasting business.

Meta–Reliance in Jamnagar

Meta announced a 168 MW initial AI-enabled data-centre facility in Jamnagar to be built by Reliance and leased by Meta, with an option to scale. Meta’s announcement discusses energy and water arrangements, including renewable power and desalinated seawater cooling, and says Meta will cover the full cost of the energy and water supporting the facility. The announcement is useful evidence of a customer-led project and a stated operating approach.

It does not disclose a project investment amount or establish that the site is already operational. The cost of the campus, exact power contract, grid connection milestone and procurement schedule should be treated as unknown unless later sources provide them.

Gujarat policy as a state-level signal

Gujarat’s policy provides a visible framework for large projects, renewable-energy conditions and eligible support. It can help an investor model the state’s intended offer and can help a vendor identify the type of projects the state wants to attract. It does not prove that any individual company has obtained benefits or that a 7.5 GW target has been built.

Uttar Pradesh policy as a separate framework

UP’s 2026 policy sets its own investment target and incentive provisions. It should not be mixed with the prior policy period or with policies in neighbouring Delhi and Haryana. A project’s actual address, phase and eligibility should be confirmed before applying the state’s terms.

Project comparison table

A practical power-readiness checklist

Before committing to a site, investment or equipment order, a project team should examine the following. This is a screening list, not a substitute for engineering, legal or regulatory diligence.

Site and grid

• Is land control documented and the parcel fixed?
• What is the requested load by phase, and is the metric IT load or facility load?
• Has the utility or transmission provider acknowledged the requested capacity?
• What voltage level, connection point, feeder and substation are required?
• Are upgrades required, who pays, and what are the dependencies?
• Is capacity reserved, conditionally approved, or only under discussion?
• What is the energisation schedule and how is delay managed?
• Are there independent external power routes, or do they share a critical point?

Commercial power plan

• What is the current tariff and full cost stack?
• What procurement routes are legally available in the state?
• Are PPAs executed or only under discussion?
• How will renewable energy be matched and settled?
• Are open-access charges, banking, curtailment and change-in-law risks modelled?
• What backup generation and storage are needed?
• How does the project perform under higher tariffs or delayed power?

Water, cooling and environment

• Which cooling design matches the target rack density and climate?
• What is the water source, quantity, quality and reuse plan?
• Are local water permissions and community risks addressed?
• Does the project include desalination or reclaimed water, and what are the energy and waste implications?
• Are flood, heat, cyclone and other hazard risks incorporated?

Commercial and delivery

• Is there a customer commitment for the first phase?
• Does the lease or service agreement specify power, delivery and cost pass-through?
• Are equipment suppliers and contractors qualified and available?
• Are transformer, switchgear, generator and cooling lead times confirmed?
• Is the construction and commissioning plan integrated with grid milestones?
• Does the project have enough contingency capital for delay or cost overrun?

Policy and incentive

• Which official policy is current for the exact site?
• What minimum capacity, renewable share and project type qualify?
• Are in-principle approval and application windows open?
• Which costs are eligible and what caps apply?
• Is the benefit a tax exemption, reimbursement, subsidy, tariff support or facilitation?
• When is the support paid and what documentation is required?
• Can it be combined with other schemes?

Power readiness by project phase

Power readiness changes as a project moves from an idea to an operating facility. A useful plan assigns an owner, source and evidence level to every milestone. It also distinguishes the proposed ultimate campus from the first deliverable phase. A project might have a long-term site ambition but only one near-term load request; the power team should not treat those as the same commitment.

Feasibility and site selection

At early feasibility, the developer should define an initial load envelope, likely ramp and expansion scenarios. The numbers should state whether they refer to IT load or total facility demand and should include assumptions about cooling, redundancy and utilisation. The team should compare at least more than one site or connection path where practical, then assess grid proximity, substation capability, transmission work, tariffs, renewable access, water and fibre together.

A preliminary utility conversation can identify likely studies or constraints, but it is not a final allocation. The record should show who provided the information, whether it was verbal or written, and what conditions remain. A company should not present a preliminary estimate as a power sanction to lenders, tenants or suppliers.

Application, studies and design

As a site advances, the project may submit a connection request, provide detailed load data and work with the utility or transmission provider on technical studies. Design decisions may affect connection voltage, feeder requirements, transformers, protection, metering and the campus substation. The parties need to align responsibility for assets, land access, right-of-way, deposits, approvals and construction.

The developer should identify dependencies and decision dates: what study must be completed, who approves it, which equipment is long-lead, and what happens if a result changes the planned capacity. A connection request, a completed study and an approved agreement are different milestones. A tracker should preserve the precise status rather than compress all three into “power secured.”

Procurement and construction

During procurement, the project should align orders for transformers, switchgear, backup systems and power controls with the confirmed technical design and delivery schedule. A vendor needs the package scope, interface conditions, test requirements, warranty, commissioning support and local service plan. If the project changes its phase or rack-density target, the electrical package may need redesign or repricing.

Construction progress should be tracked for both the data-centre campus and the external grid assets. A building can advance while a transmission line or substation upgrade is delayed. The schedule should name the party responsible for each dependency and include contingency options. The target energisation date should be labelled as contractual, official, target or estimated, depending on the source.

Energisation, integrated testing and operations

When a site or phase is energised, the project should verify the actual available connection capacity and the boundaries of the energised system. The operations team then tests power quality, transfer sequences, UPS behaviour, battery charging, generator start and load acceptance, alarms, protection settings and integrated failover. Acceptance criteria should define who witnesses and signs off each test.

After service begins, readiness is not a one-time approval. The operator monitors load growth, power quality, fuel and battery maintenance, utility events, equipment condition and planned outages. Expansion may require additional feeders or substation work, so the next phase should have its own readiness review. A live first phase is not proof that the full campus has power allocated.

Risk ownership and change control

Every material power risk should have a named owner, mitigation, due date and evidence source. If the utility revises a date, the developer changes its schedule and informs affected customers and suppliers. If a tariff, policy rule or renewable contract changes, the commercial model should be updated. A change in IT density may affect cooling and facility load, so engineering, energy procurement and finance must use the same assumptions.

A compact power risk register might include connection approval, transmission delivery, equipment lead time, tariff exposure, renewable availability, redundancy, water, climate and customer ramp. It should distinguish inherent risk from residual risk after mitigation. A numeric risk score can be useful internally if its method is documented; an unexplained score should not be presented as a verified public fact.

Make power readiness visible without inventing a score

A district-level “power score” combining surplus or deficit, tariff, DISCOM health, open-access ease and renewable availability can look attractive. These are relevant factors, but one score can create false precision unless its underlying data, method, geography and update cycle are defined. Public data may be state-wide, stale or incomparable across utilities. Use a transparent evidence card rather than claiming that a district is “80% power-ready.”

A site-level power evidence card

A confidence label should explain what is known, not rank a company’s credibility in the abstract. “Utility confirmation dated 14 August” is stronger for a connection claim than an undated promotional slide. “Not disclosed” is more honest than a guessed value. If a location is only known at city level, do not publish a precise substation or parcel based on an inference.

Keep evidence age proportional to the claim

Some policy pages change annually; grid connection status can change during a project; a state-wide tariff may be updated less frequently. A 30-day stale threshold from the working draft is not a universal rule. Use a review cadence matched to volatility and decision impact. For example, recheck a tender deadline as soon as the buyer updates it; recheck policy when a new notification is published; recheck a company’s capacity statement before republishing a market total.

Each entry should record the date verified and why it needs another check. An old source can still be valuable for project history, but it should not be presented as the current status without a more recent confirmation.

What the CEA review could clarify—and what remains unknown

The reported CEA committee is a useful signal that policymakers are examining how data-centre loads interact with electricity demand and grid planning. It is not yet a published finding that the sector is destabilising the grid or that India lacks power. The distinction matters: a review explores questions; its mandate and eventual evidence determine what conclusions can be drawn.

For operators, utilities and policymakers, useful questions include:

• How should state demand forecasts represent continuous, phase-based data-centre loads?
• How do project locations cluster around existing transmission and distribution capacity?
• Which upgrades are needed at generation, transmission and local distribution levels?
• How should demand ramps, backup testing and AI workload profiles be represented in planning?
• What information do developers need to provide to secure a credible connection timeline?
• How can grid planning coordinate with renewable procurement and storage?
• What technical and cybersecurity safeguards are relevant to power-sector connections?
• Which public data can be reported without revealing commercially sensitive site details?

A future committee report could improve national planning, but it would not replace site-specific studies or project approvals. Until formal findings are available, any projection should be attributed to its source and described as an estimate, not a guaranteed demand level.

Four playbooks for different decision-makers

Data-centre developer or operator

Start with a load plan by phase: IT load, expected facility load, ramp schedule and expansion triggers. Align the land and utility process early. Ask the utility and transmission provider what studies, deposits, security and construction milestones apply. Design redundancy across both internal systems and external dependencies. Build the commercial model with multiple power scenarios, including delays, higher network charges and lower renewable availability.

Do not treat an in-principle discussion as a final allocation. Track each connection milestone, responsible party and risk owner. If a project depends on new transmission, include the delivery schedule and contingency plan in the customer plan and financing documents.

Investor or lender

Separate generation availability, site connection, contracted energy and operational resilience. Review who bears the risk if a substation, line or transformer is delayed. Ask whether the customer’s lease start date and payment obligations depend on energisation. Stress-test project economics for a delayed phase, tariff increase, lower utilisation and incentive non-receipt. Check whether renewable claims are supported by contracts and what happens if supply is curtailed.

For a phased campus, assess each phase independently. The ultimate campus target can make a project look large, but early capex and revenue are driven by the first deliverable phase. A lender should not assume that future phases will obtain the same tariff or policy benefits without confirmation.

Equipment supplier or services firm

Map which packages are affected by grid readiness: transformers, switchgear, UPS, batteries, generators, electrical installation, monitoring, cooling and commissioning. Determine whether the buyer is the operator, owner, EPC, utility or tenant. Public project status can help identify accounts, but the supplier still needs a package-level qualification path and buyer confirmation.

A project with a stated power dependency may be a reason to research a substation or long-lead package, not proof that the buyer is ready to purchase. Avoid claiming an exact tender date without a published notice or buyer-confirmed schedule. Record an estimate as a range with assumptions and a next verification date.

State government or utility planner

Compare announced data-centre demand with actual applications, approved load, power allocation, construction progress and operational use. Coordinate target capacity with generation, transmission, distribution, water and renewable procurement. Use clear definitions and publish progress by stage, not only by MoU value. Track the difference between projects seeking capacity and projects that have reached energisation.

Transparent planning can reduce uncertainty for investors without promising capacity that the system cannot deliver. It can also help identify whether incentives are rewarding projects that contribute to grid readiness, efficient energy use, jobs and local capability.

Build a power-aware project timeline

A simple timeline should connect public events to the infrastructure work that must follow. It should not treat every project milestone as an equal signal.

A project-tracking dashboard should store both the milestone and the source that proves it. If the source uses ambiguous language such as “secured power,” quote it accurately and flag the details that remain unresolved. Do not silently convert “power strategy announced” into “power connection approved.”

Turn policy and power information into a vendor plan

A supplier should avoid a generic pitch that says “India’s data-centre boom is growing, so you need our product.” Instead, connect its capability to a verified project need and identify the correct timing and buyer.

1. Choose a defined package: e.g., power quality monitoring, switchgear, backup systems, cooling controls or commissioning services.
2. Select a buyer type: operator, developer, EPC, utility, tenant, consultant or system integrator.
3. Find a public or buyer-confirmed project: record the source, status, phase and location.
4. Check power dependency: determine whether the package depends on grid study, design freeze, tender or commissioning.
5. Verify qualification: learn the approved-vendor, safety, security and local-service requirements.
6. Prepare evidence: references, test reports, certifications, service coverage, warranty and implementation plan.
7. Offer a low-risk next step: technical review, site assessment, pilot, design consultation or formal tender response.
8. Update the record: mark whether the buying window is confirmed, estimated or unknown.

A fictional example

A power-monitoring supplier sees a public announcement for a proposed 200 MW campus. The company records the project as a market signal, not an active tender. It verifies the developer, identifies a public design consultant and checks for a current planning document. The team discovers that the first phase is not yet publicly dated. It contacts the appropriate business-development channel with a use case for metering and fault analytics, asks whether supplier qualification is open and offers a short technical session.

If the buyer confirms the package will be specified next year, the supplier can record the range as buyer-confirmed but still subject to project timing. If there is no response, the record remains an inferred watchlist item. The company does not invent a “Q1 tender” from the campus announcement and does not buy inventory against a speculative date.

This approach is slower than repeating a fixed two-quarter formula, but it improves the quality of the pipeline and reduces the risk of misleading buyers or internal executives.

Recommended labels for power and project claims

A concise status vocabulary helps customers, investors and suppliers understand what a record means. These labels should be accompanied by a source date and the relevant project phase.

• Announced: A company or government has publicly described an intention, plan or target.
• Proposed / MoU: A site or project has been discussed or formalised at a non-final or framework stage; binding details may be unknown.
• Requested: The developer has stated that it has applied for a connection or capacity, but the application is not a final allocation.
• Under study: A utility or project team is assessing the technical requirements; the result may change.
• Approved / sanctioned: A named authority has issued a defined approval or sanction, subject to its terms and conditions.
• Under construction: Work is under way on the specified facility or network asset.
• Energised: Power has been made available to a defined asset or phase; this does not necessarily mean the data centre is open to customers.
• Commissioned / operational: The operator or a credible source confirms a phase is in service; state which phase and metric.
• Policy target: A government ambition or target, separate from a project’s funding or operational status.
• Estimate: A calculation based on explicit assumptions, not a source-confirmed fact.

A company announcement may use a different term. In that case, preserve the source wording and add a plain-language explanation rather than silently upgrading its status. If sources disagree, show the conflict and prefer the latest primary evidence where appropriate. The objective is not to make every record sound certain; it is to make clear what a reader can safely infer.

Sources and publication discipline

Power and policy information is easy to overstate because announcements often combine a project, a state target and an infrastructure plan. Use the source that supports each claim:

• Use the official policy to describe eligibility and terms.
• Use the company announcement to describe what the company committed to or said it would build.
• Use a utility or regulator source for tariff, connection or grid-rule claims.
• Use market research for an attributed estimate or forecast.
• Use independent reporting to provide context, while identifying it as reporting.
• Use internal inference only when clearly labelled as an estimate and supported by stated assumptions.

When an amount or technical detail is not public, write “not disclosed.” If a secondary article quotes an official but the underlying document is unavailable, attribute the statement to the article and avoid presenting it as independently verified. A source link should point to the underlying evidence, not to a search result or an unrelated promotional page.

Frequently asked questions

Is power really the biggest bottleneck for every Indian data-centre project?

Power is a major constraint for many large projects, but it is not necessarily the biggest risk in every case. A project may be more constrained by land, financing, water, construction, customer demand, fibre or approvals. The relevant answer is site- and phase-specific. “Power is the bottleneck” should be supported with project or grid evidence, not used as a universal slogan.

Does a state with excess generation guarantee data-centre power?

No. The project also needs transmission, a local connection, approved capacity, suitable commercial terms and delivery timing. Generation capacity at the state level does not prove power is available at a particular parcel.

How much power does a data centre need?

It depends on the facility’s IT load, cooling design, redundancy, hardware density, utilisation and scope. A public MW number may refer to IT capacity, total facility draw or a campus target. Always verify the definition and phase before comparing.

Are AI data centres always much more power intensive?

AI clusters can have high rack density and significant power and cooling needs, but the load depends on the hardware, workload, design, utilisation and deployment. Not every AI service uses the same physical footprint. Avoid using one rack-power number as a universal industry fact.

What is the difference between a PPA and grid connection?

A power purchase agreement is a commercial contract for energy. A grid connection is the physical and regulatory route that enables electricity to be delivered to the site. A PPA does not automatically establish transmission access, connection capacity or backup resilience.

Do renewables mean a facility is powered by clean electricity every hour?

Not necessarily. A company may report annual renewable matching, which is different from hourly matching or continuous physical renewable supply. Ask for the reporting boundary, time period, contracts, certificates and residual grid consumption.

Does a two-quarter period before go-live reliably predict procurement?

No universal two-quarter rule is established. Lead times vary by package, design, procurement route and project. A supplier should use package-specific history, verified milestones, public tender data and buyer confirmation, and label any date estimate as an estimate.

When should a vendor approach a data-centre project?

It depends on the package. Early design discussions may be useful before a formal tender; a price quote may be premature without scope and qualification. Vendors should identify the actual buyer and project stage, then ask how the company is qualifying suppliers.

Are state incentive targets guaranteed?

No. Targets are not project approvals, award letters or payments. Benefits depend on the policy’s eligibility conditions, application, caps, approvals and compliance. Verify the current official policy and project-specific status.

Does Gujarat’s 7.5 GW target mean 7.5 GW is under construction?

No. It is a state policy target. Individual projects have their own stage and capacity; operational capacity must be reported separately.

Is the UP 2021 policy still the right policy to quote?

The current official Invest UP page is the 2026 Data Centre Policy. Do not use older provisions as if they were current without checking the 2026 framework and project-specific transition rules.

Can a company create its own captive renewable power?

Potentially, subject to applicable laws, rules, ownership/consumption conditions, approvals and state processes. The legal structure should be reviewed by qualified counsel and energy-market advisers. A marketing label alone does not establish compliance.

Are project power claims in press releases enough for a buyer or lender?

No. Public releases can establish what a company has announced, but diligence usually requires utility correspondence, connection studies, contracts, site plans, permits, technical design and financial models. Confidential documents should be verified by the project’s professional advisers.

How can Gridvik’s research support power and project tracking?

A public-source tracker can help organise project status, locations, sources and potential infrastructure signals. It cannot guarantee grid allocation, disclose confidential tenders, replace utility engagement, or confirm an exact buyer budget. Users should verify each record’s source, date, confidence and scope. See Gridvik’s Resources page, methodology and Services.

Sources and further reading

• Central Electricity Authority committee and grid-planning coverage, The Economic Times, 29 September 2026
• Google — India AI hub investment announcement
• Google — construction update on the Visakhapatnam AI hub
• Meta — Reliance AI-enabled data-centre announcement
• Gujarat Data Centre Policy 2026–29, official policy publication
• NASSCOM — analysis of Gujarat’s data-centre policy
• Uttar Pradesh Data Centre Policy 2026, Invest UP
• Wood Mackenzie — India capacity forecast and power/water constraints
• Gridvik: India data-centre expansion guide, investment map, profit-pool analysis, business models

This article is informational, not legal, energy-market, engineering, environmental or investment advice. Power access, state incentives and procurement depend on current laws, regulations, contracts and project-specific facts. Verify with the relevant utility, regulator, government policy and qualified advisers.