India Data Centre Energy SLAs 2026: Uptime, Curtailment, BESS and Contract Design

Mission-critical power contracts for Indian data centres now need a different playbook. A plain vanilla open-access PPA or a simple retail supply agreement does not map well to the operational reality of hyperscale and colocation facilities that are targeting 99.982% to 99.995% facility availability, low diesel runtime, auditable carbon claims and predictable energy cost. In 2026, the commercial gap is no longer only between grid tariff and renewable tariff. It is between what the contract promises and what the load actually needs every 15 minutes and, increasingly, every hour.

For Indian data centres, this makes energy service-level agreements a board-level issue. The procurement stack now spans utility connection agreements, open-access PPAs, scheduling and forecasting agreements, BESS dispatch logic, EMS controls, backup power obligations, deviation settlement, outage notification rules and credit support. If these interfaces are weak, the site may still have enough annual renewable units on paper while suffering expensive shortfalls, curtailment loss, diesel starts or poor carbon matching in practice.

This article focuses on how to structure data centre energy SLAs in India for 2026. The angle is deliberately different from generic 24/7 CFE, PUE or grid connectivity discussions. The core question here is contractual: what should be promised, measured and paid for when uptime matters more than average annual supply?

Why energy SLA design has become critical in 2026

Three market changes are driving the need for tighter energy SLAs.

First, data centre loads are becoming larger and flatter. A 25 MW to 80 MW IT load with a high load factor behaves very differently from a conventional industrial plant that can tolerate process interruption or shift consumption. The cost of one avoided diesel event or one avoided UPS stress episode can outweigh a small headline tariff saving.

Second, the supply mix is becoming more complex. Many buyers now combine:

• STU or ISTS open-access renewable power
• firming through third-party or on-site BESS
• utility grid import for balancing and contingency
• diesel gensets retained only for statutory and emergency duty
• captive rooftop or gas-based standby at selected sites

Third, policy and network conditions are less uniform across states. Open-access approvals, banking treatment, standby charges, wheeling losses, forecasting penalties and curtailment practices can materially affect delivered energy. Even where central rules support non-discriminatory access, actual execution still depends on state discom, SLDC and transmission utility processes.

The implication is straightforward: the buyer cannot evaluate a data centre energy contract only on Rs/kWh. It must evaluate the SLA stack around availability, firmness, response time, outage hierarchy, metering and settlement.

The five SLA layers data centres should define

A robust contract architecture typically needs five distinct SLA layers, even if some are bundled under a single counterparty.

1. Grid availability SLA

This covers utility-side supply performance at the delivery point. Typical provisions should include:

• sanctioned demand and contract demand rights
• feeder category, redundancy configuration and restoration priority
• planned outage notice period, often 7 to 15 days depending on utility norms
• emergency outage communication protocol
• harmonic, voltage and frequency compliance range aligned to grid code and equipment tolerance
• treatment of upstream bay outages, transformer outages and protection trips

In Indian metro data centre clusters, buyers should explicitly verify whether “dual source” truly means electrically independent sourcing or only two feeders from a common upstream bottleneck. Contract language should identify the single points of failure.

2. Renewable supply SLA

This is the PPA layer. For a data centre, the key issue is not annual CUF but hourly shape. The contract should define:

• committed capacity and expected generation profile by season
• deemed generation treatment during grid non-availability or backing down
• seller forecasting responsibility and revision rights
• compensation for persistent underperformance beyond force majeure
• treatment of machine outage versus grid curtailment
• replacement power rights and price caps if the seller fails to deliver within an agreed band

In 2026, hybrid solar-wind portfolios serving RTC-oriented loads often target annual generation ratios in the 1.25x to 1.6x range against contracted flat demand before storage optimisation, depending on state resource mix and carbon goals. But overbuild alone is not an SLA. The contract must define how hourly deficits are settled.

3. BESS performance SLA

If storage is part of the uptime or clean-energy claim, then BESS obligations must be contractually explicit. Core parameters include:

• usable energy capacity in MWh and guaranteed AC deliverable energy
• contracted discharge duration, commonly 1 hour, 2 hour or 4 hour blocks
• round-trip efficiency guarantee at tested conditions
• response time in milliseconds or seconds for predefined events
• annual availability guarantee, often above 97% for utility-scale service windows
• degradation curve and end-of-term capacity floor
• augmentation responsibility and cost allocation
• black-start or islanding support if applicable

A common mistake is to quote a nominal 20 MWh battery as if all 20 MWh are always available. For data centres, the SLA should refer to tested usable capacity at the AC point after auxiliary load and within temperature-controlled operating conditions.

4. EMS and dispatch SLA

As more sites optimise between grid, OA renewable supply and storage, the control layer becomes critical. The SLA for Energy management systems should define:

• telemetry granularity, usually 15-minute and increasingly 5-minute visibility
• command hierarchy between site operator, aggregator and asset owner
• dispatch objectives, such as demand cap management, renewable maximisation or outage ride-through
• cybersecurity and data retention requirements
• fail-safe control logic during communication loss
• alarm escalation and human override protocol

This matters because an excellent asset portfolio can still fail operationally if dispatch signals are delayed or conflicting.

5. Commercial settlement SLA

This is often under-specified. It should cover:

• meter hierarchy and source-of-truth meter
• time block for settlement, generally 15-minute in India, with hourly overlays for carbon accounting if needed
• treatment of losses from generator bus to drawal point
• deviation charges and who bears them
• curtailment classification and evidence standard
• outage minutes accounting methodology
• monthly reconciliation timeline and dispute process

Without this layer, parties end up arguing not about performance, but about whose meter and whose event log is valid.

Which reliability metrics matter for data centre energy contracts

Many PPAs still use generic generator metrics. Data centres need load-centric metrics.

A practical framework includes the following.

• Supply availability at point of consumption: percentage of time the contracted power stack can serve the agreed critical load band
• Unserved energy: MWh not supplied when contractually required
• Response time to disturbance: seconds from event detection to BESS or alternate source injection
• Maximum annual interruption minutes attributable to contracted supply failure
• Clean-energy coverage ratio: percentage of hourly demand served by contracted clean supply including storage discharge under the defined methodology
• Diesel displacement hours: a useful operational metric where low-runtime backup is a corporate objective

For example, a data centre may contract a 30 MW critical energy service with a monthly supply availability target of 99.9% at the import bus, backed by utility plus renewable plus BESS stack. That still allows around 43.8 minutes of monthly non-availability. If the site’s internal design objective is tighter, the contract must include layered backup and remedies rather than assume the PPA alone can deliver near-carrier-grade uptime.

In practice, buyers should distinguish between:

• energy SLA for economic and carbon performance
• n- power continuity SLA for ride-through and reliability
• facility availability SLA, which also depends on internal electrical design

This is where Load & reliability engineering becomes commercially important. Contract promises must match the actual single-line diagram, UPS autonomy, transfer scheme and backup philosophy of the campus.

Curtailment, outages and force majeure: the clauses that change project value

For mission-critical loads, curtailment clauses often matter more than nominal tariff.

In Indian open-access structures, curtailment can arise from:

• transmission constraints
• SLDC or discom backing down
• protection events on evacuation infrastructure
• generator outage
• weather-driven renewable shortfall
• scheduling and forecasting errors

These must not be lumped together.

A useful contract approach is to create an outage and curtailment waterfall.

• Seller event: plant or dedicated evacuation asset failure under seller control
• network event: STU, CTU, bay or substation outage outside seller control
• buyer event: receiving station issue or drawal restriction under buyer control
• system event: grid emergency or regulatory instruction
• force majeure event: narrowly defined exceptional events

Each category should have separate consequences.

Seller event remedies may include:

• replacement power procurement up to a capped tariff adder
• liquidated damages linked to unserved energy or benchmark market price
• deemed availability failure for repeated events

Network event treatment may include:

• deemed generation for billing if contract permits
• pass-through of documented network curtailment subject to evidence
• optional balancing from contracted BESS if physically feasible

Force majeure should not become a catch-all safe harbour. In 2026, sophisticated buyers are narrowing force majeure wording and requiring detailed notification, mitigation action and auditable evidence.

Pricing structures that align uptime with cost control

The wrong pricing model can distort performance. If a supplier is paid only for energy injected, there may be little incentive to maintain high delivery confidence during stressed periods. Data centres should assess hybrid pricing structures.

Common structures include:

• fixed energy tariff plus availability-linked adjustment
• capacity retainer for BESS or firming service plus variable throughput charge
• tolling-style storage payment where the buyer controls dispatch
• shaped RTC block with deviation collar and balancing settlement
• two-part tariff with base renewable energy rate and contingency supply premium

Indicative 2026 market ranges vary by state, tenor and credit profile, but practitioners are generally seeing:

• plain solar or wind OA landed economics often in the broad band of Rs 4.0 to Rs 6.5/kWh after state-specific charges and losses
• hybrid or RTC-oriented portfolios materially higher depending on firmness, often adding Rs 1.0 to Rs 3.5/kWh equivalent through overbuild, storage and balancing
• front-of-meter BESS service values highly dependent on duration, cycling and warranty constraints, with the economic case strongest where ToD spreads, demand charges or diesel avoidance are meaningful

For data centres, the question is not whether firm power costs more. It is whether the premium is lower than the cost of outage exposure, diesel operation, carbon reporting gaps and unmanaged peak import.

Credit support, step-in rights and lender expectations

Because data centre energy contracts are increasingly infrastructure-like, lenders and investors are paying close attention to remedy mechanics.

Bankable features typically include:

• minimum payment security from the buyer, such as LC-backed receivables
• performance security from the seller for construction and early operations
• cure periods that are short enough for operational relevance but realistic for maintenance events
• step-in rights for lenders on default by the project SPV
• clear assignment rights on refinancing or change in control
• tested acceptance protocol before COD for metering, controls and storage performance

If BESS is integrated, lenders will also scrutinise:

• degradation assumptions
• augmentation funding mechanism
• dispatch rights that may accelerate wear
• warranty exclusions linked to operating window

This is especially important where the same battery is expected to serve multiple value streams such as outage ride-through, peak shaving and market-linked balancing. The contract must define priority of use.

A practical contracting blueprint for Indian data centres

For a new campus or major expansion, buyers should work through a structured sequence.

• Define critical load bands: what must be protected at all times, what can be curtailed and for how long
• Map infrastructure reality: feeder redundancy, transformer philosophy, UPS autonomy, genset role, BESS role
• Translate operational need into SLA metrics: interruption minutes, response time, energy shortfall tolerance, carbon matching target
• Decide control philosophy: supplier-dispatched, buyer-dispatched or EMS-optimised hybrid model
• Allocate each risk explicitly: curtailment, forecast error, network outage, storage degradation, tariff change, regulatory delay
• Link payment to measurable outcomes: availability, delivered energy quality, response time and documented event handling

Where buyers are building a portfolio across metros, standardisation helps. A master commercial template with state-specific schedules is often more efficient than negotiating each site from scratch.

Growthifye supports this by combining 24/7 clean power contracting with On-site generation & BESS planning so that the legal terms, dispatch logic and infrastructure design are aligned before tendering.

What policymakers and utilities should note

If India wants data centre investment to scale while lowering diesel dependence, policy should support reliability-oriented clean power structures, not only annual energy transactions.

Key priorities include:

• faster and more transparent open-access approval workflows
• clearer curtailment reporting and event attribution at SLDC and utility level
• consistent metering and telemetry standards for hybrid and storage-backed supply
• rational standby and banking treatment that does not penalise reliability architecture
• pathways for storage to provide both consumer-side resilience and system services without double counting or double charging

Data centres are high-value, high-load-factor consumers. Better contractual and regulatory clarity reduces cost for buyers, improves bankability for developers and eases planning for networks.

Conclusion

In 2026, Indian data centre energy procurement has moved beyond tariff shopping. The hard part is now SLA engineering: specifying exactly what level of supply continuity, clean-energy coverage, curtailment protection and storage response is required, and then allocating those risks to the parties best placed to manage them.

The winning contracts are the ones that recognise a simple fact. A data centre does not buy generic electricity. It buys performance at the point of load, every block of the day, under real grid conditions.

If your team is evaluating utility supply, OA renewable contracts, storage-backed firmness or multi-site reliability standards, contact Growthifye’s advisory desk to structure an energy SLA that is bankable, measurable and fit for mission-critical data centre operations.

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.