India Data Centre PPA and Energy SLA Design 2026: Availability, Curtailment, LDs

India’s data-centre market has moved beyond simple renewable procurement. In 2026, the real commercial challenge is converting renewable supply, grid access, backup systems and storage into contract structures that align with uptime requirements and lender expectations. A vanilla solar or wind PPA does not solve a hyperscale campus’s hourly load shape, outage exposure, or compliance risk. Equally, an energy services agreement that promises “RTC power” without clear metering, curtailment allocation and performance remedies can create more dispute than resilience.

For Indian data centres, contract design now sits at the intersection of open access, scheduling, deviation settlement, state banking rules, BESS dispatch rights, backup fuel economics and service-level accountability. Developers want revenue certainty. Operators want cost visibility and uptime protection. Lenders want dispatch assumptions, remedies and change-in-law provisions that are auditable. Utilities and regulators want compliant use of open access, forecasting and DSM settlement. That means the PPA and the operating SLA must be designed together, not as separate documents.

This article focuses on a distinct question: how should Indian data-centre PPAs and energy SLAs be structured in 2026 so that availability, curtailment, liquidated damages and BESS responsibilities are allocated in a bankable and operationally workable way?

Why standard RE PPAs fail for data-centre loads

A conventional utility-scale renewable PPA in India is designed around energy delivery, not critical-load service. Typical provisions cover contracted capacity, scheduling, tariff, deemed generation in narrow circumstances, payment security and termination. That works for industrial buyers with flexible processes. It is insufficient for data centres where even a short grid event can trigger expensive switching, DG starts, UPS cycling and thermal-management risk.

Three structural gaps show up repeatedly.

• First, the generation profile rarely matches the load profile. A 50 MW IT load with PUE of 1.45 translates into a facility demand of about 72.5 MW before considering charging loads, cooling peaks and redundancy margin. A solar-heavy supply portfolio may cover large daytime blocks but leave steep evening deficits.
• Second, outage causation is split across multiple interfaces: generator, pooling substation, STU/CTU network, drawal node, on-site switchgear, BESS, EMS and diesel backup. Without explicit boundary definitions, “availability” becomes impossible to enforce.
• Third, compensation mechanisms in standard PPAs are usually energy-centric. Data centres need consequence-aware remedies that reflect replacement power costs, BESS cycling, DG fuel burn and repeated event frequency.

As a result, many 2024-2025 contracts that were labelled RTC or firm power are being redrafted in 2026 to separate energy supply obligations from reliability service obligations.

Start with the correct contract stack

For most large Indian data centres, one document cannot efficiently carry all obligations. A better structure is a contract stack with distinct functions.

• Renewable supply PPA: governs source-specific generation from solar, wind or hybrid assets, scheduling, tariff, availability testing and metering.
• Firming and shaping agreement: defines balancing energy, BESS dispatch hierarchy, third-party market purchases and imbalance allocation.
• Energy SLA: sits at the delivery-node or campus level and defines service metrics such as delivered availability, interruption thresholds, restoration times and reporting.
• Interconnection and operating procedure schedules: clarify telemetry, trip logic, synchronisation, SCADA points, black-start sequence and control-room authority.

This layered approach allows the seller, aggregator or integrated energy-service provider to commit to outcomes without pretending that a single solar or wind plant can meet a data centre’s end-use reliability need.

For example, a developer may offer a 25-year solar-wind hybrid PPA at Rs 4.40-5.20/kWh under interstate open access, while a separate shaping layer procures market energy and dispatches BESS to achieve a monthly delivery target. The SLA then measures what the data centre actually receives at the interconnection point. This is far more robust than inserting generic “best efforts RTC” wording into a generation PPA.

Define availability at the right electrical boundary

The most common drafting error is using a vague availability definition. Data centres should insist on at least three separate metrics.

• Source availability: whether contracted generating units were capable of producing and scheduling power, net of approved outages.
• Delivery availability: whether energy could be delivered up to the contracted drawal node, considering transmission and open-access path availability.
• Critical-load support availability: whether the total contracted system, including BESS or firming supply where applicable, supported the agreed import profile at the data-centre interface.

These metrics should be measured at explicit boundaries:

• Generator bus / pooling station
• Interconnection with STU or CTU
• Open-access drawal meter
• Data-centre incomer or dedicated receiving substation

Without this separation, sellers may claim the plant was available while the buyer experiences curtailment or zero drawal because of path congestion, communication failure or scheduling error.

In practice, a mission-critical facility may seek annual critical-load support availability above 99.5% for the contracted service layer, while recognizing that renewable source availability alone will be much lower on an hourly basis. The SLA should also define event categories such as:

• Planned outage with minimum notice, for example 7-15 days depending on equipment class
• Forced outage under 15 minutes
• Sustained interruption above 15 minutes
• Partial shortfall, such as delivered power below 90% of scheduled block

For each category, reporting windows should be tight. A 15-minute SCADA-based timestamp convention, aligned with Indian scheduling blocks where relevant, is typically easier to audit than narrative incident logs.

Curtailment allocation is now a first-order bankability issue

Curtailment language in 2026 cannot be generic. For data-centre energy structures, there are at least five different curtailment buckets, each with different commercial treatment.

• Grid security curtailment by SLDC/RLDC/NLDC
• Transmission congestion limiting schedule acceptance
• Distribution-level restrictions at the drawal side
• Economic curtailment due to negative or adverse market spreads if merchant balancing is involved
• Technical curtailment caused by inverter, evacuation or communications failure at the project side

The buyer should not pay full tariff for technical curtailment arising from seller-side equipment or forecasting failure. But neither should the seller absorb all force-majeure style grid curtailment if the transaction economics assumed some level of curtailment risk sharing.

A workable approach is to classify curtailment as compensable or non-compensable.

Compensable to buyer:

• Seller equipment outage
• Seller evacuation failure up to defined boundary
• Scheduling or forecasting non-compliance attributable to seller or seller-appointed QCA
• Failure to maintain contracted storage availability, where storage is part of the service

Shared or pass-through treatment:

• State or regional grid security curtailment beyond seller control
• Open-access denial caused by system-wide transmission constraints, subject to documented evidence
• Change in law affecting scheduling rights, banking or transmission usage

The payment formula should not stop at “deemed generation”. For data centres, the true exposure is replacement cost. If a 20 MW shortfall during an evening peak forces purchase at Rs 8-11/kWh in DAM or RTM, while contract tariff is Rs 5/kWh, the economic damage is not just lost renewable energy revenue. A rational remedy can be linked to the positive difference between documented replacement energy cost and contract tariff, subject to a cap. This is often more defensible than arbitrary fixed penalties.

Liquidated damages should reflect reliability consequences, not just lost MWh

Liquidated damages in these contracts are often under-designed. Traditional RE PPAs may impose delay LDs during construction and limited performance LDs after COD. Data centres need operational LDs that map to how outages create cost.

A practical 2026 framework uses three LD layers.

• Capacity shortfall LDs: if the contracted firming layer cannot support a minimum MW profile during specified windows.
• Energy shortfall LDs: if monthly or annual delivered energy falls below floor volumes for reasons attributable to seller.
• Service-event LDs: if repeated interruptions, restoration delays or telemetry failures breach SLA thresholds.

For example, a contract could specify that if firm delivery during designated critical hours, say 18:00-24:00, falls below 95% of agreed schedule more than three times in a month due to seller-controllable causes, the seller pays event-based LDs plus documented replacement energy differential. Another method is a service-credit regime that escalates after recurrence. The latter is useful where an integrated energy-services provider is managing supply, BESS and controls under a broader operating contract.

LD caps remain necessary for financeability, but they should not be so low that they are meaningless. In current Indian market practice, developers may push for annual liability caps near 10-20% of annual billing. Critical-load buyers often seek higher effective protection through separate indemnities for wilful misconduct, metering fraud, non-compliant dispatch instructions or repeated SLA breach. The middle ground is to keep general LD caps bankable while carving out specific uncapped or higher-cap items tied to gross negligence, data integrity and payment diversion.

BESS obligations must be explicit: who controls, who pays, who cycles

Where storage is part of the solution, the contract must answer four questions clearly.

• Who has dispatch authority?
• Who bears degradation and augmentation cost?
• What round-trip efficiency and usable capacity are guaranteed?
• Is the BESS optimised for reliability, tariff arbitrage, renewable firming or all three?

This matters because the same battery cannot be committed twice. A 40 MWh on-site system designed to provide 20 MW for 2 hours during grid disturbance cannot simultaneously be assumed available for daily arbitrage or renewable shaping without explicit priority logic.

The operating schedule should define a hierarchy such as:

• First priority: ride-through and critical-load support
• Second priority: scheduled peak support or contract firming
• Third priority: tariff optimisation

Performance schedules should include:

• Minimum guaranteed usable energy at beginning of year and end of warranty year
• Availability target, often 97-99% depending on maintenance philosophy
• Maximum response time in milliseconds or seconds depending on service class
• Auxiliary consumption treatment
• Charging-energy source rules and settlement

If the seller controls the battery under an integrated service, the buyer must receive transparent dispatch logs. If the buyer controls the battery, the seller should not be liable for shortfalls caused by buyer dispatch that leaves insufficient state of charge. These interface risks are exactly where Growthifye’s On-site generation & BESS and Energy management systems work becomes commercially important, because the technical operating envelope must be mirrored in the legal drafting.

Metering, telemetry and settlement rules are no longer boilerplate

Disputes in data-centre energy contracts increasingly arise from data mismatch rather than pure non-performance. The settlement annexure should specify:

• Primary and check meters at all commercial boundaries
• Time synchronization standard
• SCADA granularity, typically 1-minute operational data and 15-minute settlement data
• Meter hierarchy in case of conflict
• Estimation methodology during meter failure
• Deadline for preliminary and final statements

Because Indian markets still involve a mix of OA billing, SLDC charges, DSM, transmission losses and cross-subsidy related pass-throughs depending on structure, invoices should be unbundled. At minimum, separate:

• Energy charges
• Capacity or availability charges, if any
• OA and transmission charges
• SLDC and scheduling charges
• DSM or imbalance charges
• BESS service fees
• Replacement energy adjustments
• Taxes and duties

This avoids the common problem where buyers cannot trace whether higher landed cost came from market purchases, losses, curtailment or scheduling penalties.

Change in law and open-access risk need sharper drafting in 2026

Open-access economics remain exposed to state-level volatility. Banking restrictions, additional surcharges, CSS treatment, TOD structures and captive-compliance scrutiny can all change project viability. A robust data-centre energy contract should list which items are pass-through, which trigger tariff rebasing, and which create termination rights.

In 2026, key sensitivities include:

• State-specific open-access eligibility and procedural tightening
• Banking withdrawal or reduced settlement value
• Changes in ISTS waiver treatment for eligible projects and timelines
• New forecasting, telemetry or QCA compliance costs
• Landed impact of transmission-loss revisions and wheeling-loss adjustments

Not every change should reopen the whole contract. Threshold mechanisms work better. For instance, if documented change-in-law impact moves landed power cost by more than, say, 7-10% on a sustained basis, parties can reopen tariff or service structure. Below that band, costs may be shared according to a pre-agreed matrix.

What sponsors, lenders and operators should check before signing

Before financial close or execution, stakeholders should stress-test the contract against real operating scenarios.

• Evening renewable deficit with RTM price spike above Rs 9/kWh
• STU congestion during high-wind season
• Data-centre load growth 15-20% above base forecast
• BESS degradation faster than modelled
• DG dispatch due to grid outage plus low battery SOC
• QCA forecasting failure causing repeated DSM charges

Sponsors should verify that the project company is not assuming unlimited balancing ability at thin spreads. Lenders should examine whether LDs and replacement-cost exposure are insurable, capped or back-to-back with O&M and equipment suppliers. Data-centre operators should ensure the SLA measures service at the electrical point that matters to uptime, not just at a remote generation bus.

The best contracts in today’s market are not the cheapest on headline tariff. They are the ones where risk allocation is explicit, metering is auditable, BESS duty is realistic, and curtailment compensation matches actual cost exposure. Done well, this improves both bankability and operational resilience.

For data-centre owners and developers planning new campuses or retrofitting existing procurement structures, the next step is an integrated legal-technical-commercial review. Growthifye supports this through Load & reliability engineering and 24/7 clean power contracting, translating power-system realities into workable PPAs, SLAs and BESS operating frameworks.

If you are structuring a data-centre energy contract in India, contact Growthifye’s advisory desk for a transaction-focused review of your PPA, firming layer, SLA and storage integration strategy.

Explore Growthifye's related capabilities

This analysis connects directly to our advisory practice: Load & reliability engineering · 24/7 clean power contracting · Grid connectivity & redundancy · On-site generation & BESS.