India power and flexibility outlook: making capacity bankable

The investable question is delivery, not the headline

For a refinery, distribution utility or industrial group, a national capacity announcement is only the beginning of an investment case. The commercial question is whether a specific portfolio can deliver the required quantity of power, at the required connection point, during the hours when failure is expensive. A megawatt added in a region with congestion is not interchangeable with a megawatt available at an industrial busbar. Likewise, a low generation tariff is not the same as a low total procurement cost.

Enalyt’s central analytical view is that the next competitive advantage lies in matching three ledgers: the physical system, the contractual allocation of rights and the cash cost of delivering energy. Build the physical ledger in MW, MW h and hourly profiles. Build the contractual ledger around charging entitlement, scheduling, degradation guarantees and remedies. Build the cash ledger around capacity payments, charging energy, losses, network costs and financing. Keeping those ledgers separate prevents apparent savings from being counted twice.

This report uses verified public evidence as context and then constructs reproducible scenarios. It does not claim to forecast exchange prices or individual company earnings. For a buyer considering a long-duration commitment, the useful output is a range of conditions under which a project works, plus observable signposts that would invalidate the original decision. That is a stronger basis for procurement than extrapolating one recent auction or a single month of demand.

Read the data through consistent boundaries

NITI’s dashboard reports 558.90 GW total installed generation capacity for September 2026 and approximately 1,707 BU electricity requirement for FY2025-26. MNRE separately reported 297.36 GW non-fossil capacity as at 30 June 2026. These are differently dated observations; they should not be combined into a current share calculation.[S1][S2]

A robust market database should preserve the original reporting period, publication date, unit, scope and provisional status. It should also retain old versions. A later publication can repeat an older observation; its publication date does not refresh the underlying data. Similarly, apparent disagreement between sources may reflect treatment of captive generation, hydro, storage or data revisions. Reconcile the boundary before deciding which number is wrong.

Use demand scenarios rather than a point prediction

The 2026 CEA resource-adequacy plan projects a 2035-36 peak of 459 GW in its main demand case and separately considers an alternative demand path. These are planning scenarios, not a guarantee of realised demand. Earlier 2023 NEP figures should remain identifiable as an earlier forecast vintage.[S3][S7]

A procurement committee should stress-test both the level and shape of future demand. Annual energy growth can coexist with slower peak growth if utilisation broadens across the day. Conversely, cooling demand can increase the top few hours disproportionately. For a battery, the relevant variable is not simply national electricity growth but the number of economically useful discharge events and the duration of each event. A small change in these inputs can dominate a modest equipment-price reduction.

The spread between these cases is 57.8 GW. A diversified portfolio may be preferable to a single irreversible bet when the cost of under-procurement and the cost of surplus capacity are asymmetric. Translate each case into regional load, contracted resources and a residual hourly requirement. Re-run the analysis against adverse weather, delayed connections and outages. Do not allocate the full national peak sensitivity to one state or company without a defensible bottom-up load model.

Separate the storage pipeline from delivered assets

The Ministry of Power’s 30 June 2026 snapshot distinguishes BESS under construction, awarded capacity and tendering capacity. The associated energy-to-power ratios differ, so treating all procurement as a uniform two-hour product would distort the opportunity.[S4]

For a supplier, the opportunity funnel should narrow from announced potential to eligible tenders, executable orders and funded construction. For an investor, the same funnel should distinguish land, grid connection, offtake, financing, procurement and commissioning evidence. A project can remain visible in a pipeline after one of those prerequisites has failed. The work is not merely collecting announcements; it is maintaining a timestamped change history and a consistent project identifier.

For a buyer, pipeline monitoring is also a competitive tool. A large volume of projects seeking the same commissioning window can create delivery and supply-chain pressure even as bid prices fall. Ask whether a low price assumes a different duty cycle, subsidy, domestic-content treatment, tax position or end-of-life warranty. A project count without those attributes is insufficient for procurement benchmarking.

Size a service at the point of delivery

Consider an illustrative requirement to provide 100 MW for four consecutive evening hours at the AC delivery point. The net delivered energy requirement is 400 MW h. Assuming 90% retained capacity at the contractual end of life and a 95% usable discharge window gives approximately 468 MW h of initial AC-equivalent discharge energy before those two deratings. This is not a DC battery-cell nameplate rating: conversion to DC nameplate also requires a separately stated discharge-path efficiency and an explicit treatment of auxiliary loads.

At an assumed 85% net AC-to-AC round-trip efficiency, the 400 MW h delivery event requires approximately 470.6 MW h of AC charging energy:400/0.85. Round-trip efficiency is used for this charging-energy bridge; it is not a substitute for a separate discharge efficiency in a DC nameplate calculation. The design still needs a power-conversion system capable of the contracted MW, adequate connection capacity, temperature-dependent performance and consistent auxiliary-load boundaries. A larger energy allowance does not automatically solve inverter limits or a whole-site outage.

Availability must also be handled correctly. Multiplying this energy allowance by the inverse of an annual availability percentage does not guarantee delivery through an outage. Reliability may require spare power modules, redundancy, portfolio diversification or an alternative supply arrangement. The buyer should request event-level performance simulations and clear remedies for failure during critical hours. An annual average availability guarantee is weaker than a guarantee aligned to the actual hours of value.

Convert capacity charges into a comparable delivery cost

A capacity charge quoted per MW per month cannot be compared directly with a generation tariff quoted per kWh. The denominator matters. For an illustrative 1 MW/two-hour contract atRs 200,000/MW/month, annual fixed payment isRs 2.4 million. At 330 cycles and 95% usable availability, annual discharged energy is 627,000 kWh. Fixed cost is thereforeRs 3.83/kWh. Charging atRs 3/kWh and 85% round-trip efficiency addsRs 3.53/kWh, producingRs 7.36/kWh before network charges, taxes, trading, deviations and other excluded costs.

This sensitivity suggests a practical diligence order. Confirm dispatchable cycles and the charging plan before negotiating a small capacity-price difference. Then test the charging tariff at the same delivery boundary as the discharge benefit. Cheap midday energy at another node is not necessarily accessible to the battery. Curtailment cannot be treated as free charging unless the project has the relevant physical and contractual rights.

Revenue stacking needs the same discipline. CERC has an ancillary-services framework, but eligibility does not create guaranteed utilisation or revenue.[S6] A firming commitment may reserve the same MW or state of charge needed for another service. Allocate availability across products and simulate conflicts before adding two independent revenue forecasts.

There is a useful comparison hidden in the cycle sensitivity. A25% reduction in the illustrative capacity charge, fromRs 200,000 toRs 150,000/MW/month, does not fully offset utilisation falling from 330 to 200 cycles. At 200 cycles, the lower charge still produces approximatelyRs 8.27/kWh including charging, above the Rs 7.36 base case. To preserve the base result at 200 cycles, the capacity charge would need to fall to aboutRs 121,200/MW/month, assuming unchanged efficiency and charging cost. This is arithmetic rather than a price forecast. It demonstrates why an apparently dramatic headline tariff reduction can coexist with worse delivered economics when the usable dispatch opportunity deteriorates.

Treat support schemes as conditional cash flows

The 4 November 2026 scheme announcement providesRs 14,500 crore support for 92 GW h of storage and specifies a 40% minimum domestic-content requirement for BESS. The split is 30 GW h BESS,60 GW h pumped storage and 2 GW h alternative technologies.[S5]

That announcement changes the diligence agenda, but it does not justify subtracting an assumed grant from every project’s capital cost. Identify the implementing route, eligible recipient, disbursement milestones, security requirements and consequences of delay. Trace whether the commercial benefit appears in a lower capacity charge, a developer capital contribution or another contractual mechanism. A grant payable after commissioning can still leave a material construction-funding requirement.

The domestic-content condition should be converted into an auditable bill-of-materials question. Which component categories count, which evidence proves origin and who bears the risk if a supplier’s declaration is rejected? Build the compliance cost into the base case and a delay scenario into working capital. Do not assume that a low imported-cell quotation and the full policy benefit are simultaneously available without checking the applicable rules.

The historical SECI ESS-I record illustrates another gate: a tariff discovery and an award are different from regulatory adoption. CERC’sJanuary 2025 order is a concrete reminder to track the approval sequence rather than treating an auction price as an immediately executable market benchmark.[S8] Transaction-specific legal and regulatory review is required before commitment.

Model correlated execution and operating downside

Single-variable sensitivities are useful for understanding a model, but they can understate commercial risk. Consider a buyer whose storage contract begins six months after the expected date, while expensive evening purchases and lower-than-expected useful cycling occur in the first operating year. The delay raises replacement-supply cost; lower cycling spreads the fixed storage payment over fewer delivered units; expensive charging narrows the value of the eventual dispatch. These exposures should be evaluated together.

Create a monthly cash-flow bridge from contract award through stable operation. Include development spending, deposits, construction payments, connection milestones, subsidy receipts where eligible and payment delays. Then overlay the operational dispatch model. This reveals whether a project with an acceptable long-run unit cost still creates an unacceptable funding gap. A low unlevered cost of energy is not proof that the project can finance its way through the construction and ramp-up period.

The same approach applies to an industrial buyer assessing a supply contract. The relevant downside includes the cost of replacement power during delay, production consequences of non-delivery and the recoverability of contractual damages. A penalty receivable is not equivalent to physical supply. Test counterparty credit and security alongside technical performance rather than assigning full recovery automatically.

A decision system for the next twelve months

A useful corporate workflow begins with an hourly load and portfolio model, not a technology shopping list. Split the residual requirement into short evening peaks, sustained overnight energy, reserve and seasonal gaps. Screen each segment against demand response, market purchases, bilateral contracts, batteries and pumped storage. Preserve the option to contract different services instead of forcing one asset to solve every problem.

• For procurement teams:normalise every offer to the same delivery boundary, cycle entitlement, end-of-life guarantee and excluded-cost schedule. Request a machine-readable assumptions sheet.
• For development teams:track one project identifier through announcement, bid, award, signed agreement, tariff adoption, financing, construction and commissioning. Record documentary proof of every transition.
• For strategy teams:maintain forecast vintages and confidence ranges. Review demand shape and regional constraints when macro assumptions change.
• For finance teams:stress charging prices, utilisation, degradation, delays and payment timing together. Correlated downside is more informative than isolated single-variable sensitivities.

Enalyt’s view is that a high-quality energy database should make these decisions faster and more auditable. Its value is not the number of rows alone. It is the ability to compare a bid with the appropriate peer group, reconstruct a historical assumption and explain why a project moved from opportunity to risk. The useful subscription product is therefore a maintained evidence system, connected to models that users can interrogate.