Insuring Data Centres in India: Electrical, Mechanical, and Cyber Risks Under One Programme

A data centre is, on the surface, a building. It has walls, a roof, electrical connections, and plumbing. But from an insurance perspective, treating a data centre like any other commercial building is a fundamental underwriting error that leads to coverage gaps large enough to bankrupt an operator after a single significant loss event.

The reason is straightforward: in a conventional commercial property, the building shell represents 60-80% of the total asset value, with contents (furniture, fixtures, inventory) making up the remainder. In a data centre, the relationship is inverted. The building shell may represent only 15-25% of the insurable value. The remaining 75-85% sits in the servers, storage arrays, networking equipment, cooling infrastructure, power distribution units, UPS systems, diesel generators, and the kilometres of structured cabling that connect it all. A 10,000 sq ft Tier III data centre in Hyderabad or Navi Mumbai can easily hold INR 100-200 crore worth of electronic equipment inside a building shell worth INR 15-20 crore.

This concentration of value per square foot exceeds almost every other commercial occupancy class. A pharmaceutical manufacturing facility might hold INR 50 crore of equipment across 100,000 sq ft. A data centre holds multiples of that value in a fraction of the space. The insurance implications are immediate: the sum insured per square foot is extraordinarily high, which means even a localised incident affecting one section of the facility can produce a claim disproportionate to the physical area damaged.

Beyond value concentration, data centres present three characteristics that set them apart from every other property risk an Indian underwriter encounters.

First, they operate 24/7/365 with zero tolerance for downtime. A manufacturing plant can shut down for maintenance windows, absorb a two-day power outage with limited financial impact, and resume production once the issue is resolved. A data centre cannot. The contractual obligations that data centre operators carry through Service Level Agreements (SLAs) with their customers impose financial penalties that begin accruing within minutes of an outage. A Tier III facility typically commits to 99.982% uptime, which translates to a maximum permitted downtime of approximately 1.6 hours per year. Any incident that breaches this threshold triggers SLA credits, customer churn, and reputational damage that compounds over subsequent contract renewal cycles.

Second, the perils that threaten data centres are different from those that threaten conventional properties. Fire is a risk, certainly, but the more frequent and often more damaging perils are electrical: power surges, UPS failures, transformer faults, and cascading electrical events that propagate through interconnected power distribution systems. Water damage, typically from cooling system leaks rather than external flooding, is another high-frequency peril. And cyber events, which have no physical manifestation at all, can render an entire facility's contents worthless without causing a single rupee of physical damage.

Third, data centres house third-party assets. In a colocation model, which accounts for a significant and growing share of India's data centre capacity, the operator owns the building, the power infrastructure, and the cooling systems, but the servers and storage equipment inside the racks belong to the customers. This split ownership creates complex insurance obligations. The operator must insure its own assets and demonstrate adequate coverage to its customers. The customers must insure their own equipment housed in the facility. And both parties face liability exposures to each other and to the end-users whose data resides on those servers.

India's data centre market is expanding at roughly 25-30% annually. Hyderabad, Chennai, Mumbai, and Pune are the primary hubs, with new capacity being built in Noida, Kolkata, and Bengaluru. The Ministry of Electronics and IT's Data Centre Policy, combined with state-level incentives offering subsidised land and power tariffs, has attracted both domestic operators (Yotta, NTT, STT GDC) and international entrants (Equinix, Digital Realty). As the installed capacity grows, the insurance market is being asked to absorb increasingly large and complex risks, often without adequate pricing models or loss experience to draw on.

The Indian insurance market's standard approach to commercial property, anchored on the SFSP policy with engineering add-ons, is structurally inadequate for data centres. It was designed for buildings that contain tangible goods, not for facilities where the primary asset is electronic equipment operating under extreme environmental sensitivity. Insuring a data centre properly requires a programme approach: multiple interlocking policies, each addressing a specific domain of exposure, structured to eliminate gaps and avoid overlapping triggers that could lead to coverage disputes.

There is also a geographic dimension to the risk that Indian underwriters must consider. The four primary data centre hubs in India each carry distinct peril profiles. Mumbai facilities face monsoon flooding risk, particularly those in low-lying areas of Navi Mumbai and Airoli. Chennai is exposed to cyclone risk, with Cyclone Vardah (2016) having demonstrated the vulnerability of IT infrastructure to high-wind events. Hyderabad, while relatively free of natural catastrophe risk, faces chronic power quality issues due to the rapid expansion of electrical load in the IT corridor outpacing grid infrastructure upgrades. Pune facilities, concentrated in the Hinjewadi and Kharadi IT zones, contend with waterlogging during heavy monsoon events and transformer failure rates that reflect the strain on the local electrical distribution network. Each location demands location-specific risk engineering and peril-specific insurance programme design, not a one-size-fits-all approach.

The Standard Fire and Special Perils (SFSP) policy is the default starting point for insuring any commercial building in India, and most data centre operators purchase one. The policy covers the building structure, plant and machinery, and contents against fire, lightning, explosion, storm, flood, earthquake, and other named perils. For a data centre, it covers the building shell, the electrical infrastructure permanently affixed to the building (transformers, main distribution panels, bus ducts), and, if declared, the server and networking equipment housed inside.

But the SFSP policy was not designed for electronic equipment, and its limitations become apparent quickly when applied to a data centre environment.

The first limitation is the basis of valuation. The SFSP policy settles claims on a reinstatement basis for buildings and on a market value or indemnity basis for contents, depending on the policy terms. Electronic equipment depreciates rapidly. A server purchased three years ago for INR 8 lakh may have a market value of INR 2-3 lakh today, but replacing it with an equivalent current-generation machine costs INR 10 lakh. Under an indemnity settlement, the policyholder receives the depreciated value. Under reinstatement, the policyholder receives the cost of replacing with equipment of the same kind and capacity, but not superior equipment. Since server technology evolves every 18-24 months, replacing a three-year-old server with one of 'the same kind and capacity' may be impossible because the model is discontinued. This creates a valuation dispute that the SFSP wording is not equipped to resolve.

The second limitation is peril coverage. The SFSP covers fire and named perils, but many of the events that cause the most damage in data centres are not named perils under the standard wording. Power surges, voltage fluctuations, and short circuits that damage electronic components without causing a fire are not automatically covered. The SFSP can be extended to cover 'electrical installation damage' through an add-on endorsement, but this extension typically applies to the electrical installation itself (wiring, switchgear, panels) and not to the electronic equipment connected to it. A voltage spike that destroys 200 servers but causes no damage to the power distribution panel falls into a coverage grey zone under the SFSP.

The third limitation relates to temperature and humidity damage. Data centres maintain precise environmental conditions: typically 18-27 degrees Celsius and 40-60% relative humidity, per ASHRAE guidelines. If the cooling system fails and temperatures rise above 35 degrees, servers begin throttling performance, and sustained exposure above 40 degrees can cause permanent damage to processors, memory modules, and storage drives. This damage is gradual, not sudden, and the SFSP typically requires the damage to result from a 'sudden and unforeseen' event. A cooling system that degrades over several hours, eventually causing server damage, may not meet this threshold.

The fourth limitation is the sub-limit structure. Even where the SFSP does cover electronic equipment, the sums insured and sub-limits must be carefully managed. Many Indian SFSP policies apply an inner limit for any one item or any one occurrence at one location. A data centre where the total electronic equipment value is INR 150 crore but the any-one-occurrence sub-limit is INR 50 crore has a significant self-insured retention, whether or not the operator realises it.

The fifth limitation concerns debris removal and data recovery. After a fire or water damage event in a data centre, the primary recovery cost is not clearing physical debris (though that is necessary) but recovering, restoring, or reconstructing the data stored on damaged systems. Data restoration costs, including forensic recovery, backup system activation, and manual re-entry of lost records, are not covered under the SFSP. They require a separate policy or endorsement.

The sixth limitation is the SFSP's treatment of fire suppression system discharge. Data centres use clean agent fire suppression systems (FM-200, Novec 1230, or inert gas systems) rather than water-based sprinklers, because water would destroy the electronic equipment the system is designed to protect. These clean agent systems are expensive to recharge after activation, typically costing INR 10-30 lakh per zone depending on the agent used and the protected volume. A false discharge, triggered by a faulty smoke detector or a maintenance error, causes no fire damage but still requires full system recharge. The SFSP does not cover the cost of recharging fire suppression systems after a false discharge, because there is no insured peril and no material damage. This cost must be addressed either through a specific endorsement or through the facility's operational budget.

The seventh limitation relates to the interdependence of building systems. In a conventional commercial property, damage to the building's HVAC system is an inconvenience. In a data centre, the loss of cooling is an emergency that can cause secondary damage to IT equipment within hours. The SFSP treats each item of damage independently, but data centre losses are almost always cascading: a roof leak damages a CRAC unit, which causes a temperature rise, which damages servers. The causal chain creates attribution challenges under the SFSP that are better handled by specialist engineering and electronic equipment policies.

For these reasons, using the SFSP as the sole property cover for a data centre is insufficient. The SFSP serves a useful function as the base layer, covering the building shell and permanent installations against fire and natural perils. But the electronic equipment, the cooling infrastructure, and the data recovery exposure all require dedicated policies that address their specific valuation, peril, and claims settlement requirements.

The two engineering insurance products that matter most for data centres are the Machinery Breakdown (MB) policy and the Electronic Equipment Insurance (EEI) policy. These are distinct covers, each designed for a specific class of equipment, and a data centre typically needs both.

The Machinery Breakdown policy covers sudden and unforeseen physical damage to machinery from internal causes: mechanical failure, electrical defects, short circuits, centrifugal force, defective material, and similar operational faults. In a data centre context, the MB policy is the appropriate cover for the power generation and distribution infrastructure. This includes diesel generators (typically 1-2 MW units, with most Tier III facilities maintaining N+1 redundancy), UPS systems (rotary or static, rated at several hundred kVA to several MVA), power distribution units, automatic transfer switches, and the cooling plant (chillers, precision air conditioning units, cooling towers, and their associated pumps and compressors).

The MB policy is written on a sudden and unforeseen basis, meaning it responds to unexpected mechanical or electrical failures but excludes gradual deterioration, wear and tear, corrosion, and damage resulting from the insured's failure to maintain the equipment in accordance with manufacturer specifications. This maintenance condition is particularly relevant for data centres, where equipment runs continuously at high loads. A chiller that has not been serviced per the OEM schedule and fails due to compressor burnout will face a contested claim. Underwriters will request maintenance logs, and a gap in scheduled maintenance can void coverage for the specific failure.

The sum insured for the MB policy should reflect the replacement cost of each covered machine, including delivery, installation, and commissioning. For imported equipment, which is common in Indian data centres (Caterpillar or Cummins gensets, Schneider or Vertiv UPS systems, Daikin or Carrier chillers), the sum insured must account for import duties, freight, and installation by authorised technicians. A Tier III data centre in Chennai with 4 x 2 MW diesel generators, 3 x 1.5 MVA UPS systems, and a 600-TR chilled water cooling plant can easily carry INR 25-40 crore in machinery value under the MB policy alone.

The Electronic Equipment Insurance policy is the more specialised product and the one most data centre operators undervalue. The EEI policy covers sudden and unforeseen physical damage to electronic equipment from any cause, including but not limited to fire, lightning, power surges, voltage fluctuations, short circuits, water damage, humidity damage, mechanical impact, theft, and operator error. Its peril coverage is broader than both the SFSP and the MB policy for electronic equipment.

In a data centre, the EEI policy covers servers, storage arrays, networking switches and routers, firewalls, load balancers, structured cabling, monitoring and management systems, and any other electronic equipment used in the facility's operations. For colocation operators, the EEI covers the operator's own electronic infrastructure (top-of-rack switches, management servers, monitoring systems, structured cabling) while the customers' equipment is insured under the customers' own policies.

The EEI policy offers a critical advantage over the SFSP for electronic equipment: it covers electrical and electronic damage that does not involve fire. A voltage transient that damages motherboards across a row of servers, a UPS output that goes out of specification and fries connected equipment, or a water leak from a CRAC unit that damages servers in the rack below, these are all covered under the EEI but may be excluded or contested under the SFSP.

The EEI policy also includes, as a standard or optional extension, coverage for external data media. This covers the cost of reproducing data stored on damaged hard drives, SSDs, and other storage media. The cover is limited to the cost of blank media plus the labour cost of data restoration from backup. It does not cover the intrinsic value of the data itself (which is commercially unquantifiable in most cases) or the cost of recreating data that was never backed up.

A critical underwriting consideration for both MB and EEI policies in data centres is the deductible structure. Given the high frequency of minor equipment failures in a facility running thousands of devices 24/7, underwriters typically impose higher deductibles than they would for a conventional commercial risk. Deductibles of INR 5-10 lakh per occurrence are standard for EEI policies covering data centre equipment, and waiting periods of 24-48 hours apply to any associated loss of revenue extensions. The deductible must be calibrated to eliminate attritional claims (a single server power supply failure, a single hard drive crash) while preserving coverage for significant events.

The interplay between the MB and EEI policies requires careful boundary definition. The general principle is that the MB policy covers electromechanical equipment (generators, UPS, cooling plant) while the EEI covers electronic equipment (servers, storage, networking). However, modern UPS systems contain both power electronics and digital control systems. A failure in the UPS's electronic control board that causes the unit to deliver out-of-specification power, which then damages downstream servers, involves both policies. The MB policy responds for the UPS damage. The EEI policy responds for the server damage. But if the policies are placed with different insurers, the allocation of the loss between them can become contentious. This is one reason why programme design, discussed in the final section, matters enormously.

Business interruption (BI) insurance for data centres is fundamentally different from BI cover for manufacturing or retail operations. In a factory, BI coverage compensates for the gross profit lost during the period that production is halted due to insured physical damage. The calculation follows a well-established formula: actual gross profit for the indemnity period minus the gross profit that the business would have earned during that period had no loss occurred.

Data centres complicate this model in several ways.

First, the revenue model is different. A data centre operator earns revenue through monthly recurring charges for rack space, power consumption, cross-connects, and managed services. Revenue is contractual and predictable, which in some ways makes BI calculation simpler than for a manufacturer with fluctuating order books. But the loss of revenue following an outage is not linear. A minor outage that lasts four hours may trigger SLA penalties that cost the operator 5-10% of the affected customers' monthly charges. A major outage lasting 48 hours can trigger contract termination clauses, allowing customers to exit without penalty. The resulting revenue loss is not just the lost income during the downtime period; it is the permanent loss of the customer relationship, which may represent years of future contracted revenue.

Standard BI policies in India do not accommodate this step-function revenue loss pattern. They are designed for businesses where revenue returns to pre-loss levels once the physical damage is repaired. For a data centre, revenue may never return to pre-loss levels if key customers have exercised their termination rights during the outage. This requires either policy wording that extends the indemnity period to include a 'customer reconstitution' period (analogous to the 'loss of attraction' extension used in hospitality and retail BI policies) or a separate SLA penalty cover.

Second, the trigger mechanism needs careful definition. Standard BI cover in India attaches to the material damage policy: BI is payable only if there is physical damage to the insured property caused by an insured peril, and only for the period during which operations are affected by that physical damage. For a data centre, the most disruptive events may not involve physical damage at all. A cyber attack that encrypts data across all servers causes total operational disruption but zero physical damage. A software bug in the cooling management system that causes unnecessary shutdown of IT loads produces revenue loss without any hardware damage. A utility power failure lasting 72 hours that exhausts diesel fuel supplies forces a controlled shutdown without damaging any equipment.

To address these scenarios, data centre BI programmes typically need multiple trigger mechanisms. The primary BI policy, linked to the SFSP or EEI material damage cover, responds when physical damage causes interruption. A cyber BI policy, linked to the cyber insurance programme, responds when a cyber event causes interruption without physical damage. And a 'denial of access' or 'utility interruption' extension responds when external events (power grid failure, civil authority orders, access road blockage) prevent the facility from operating without any damage to the insured property itself.

Third, the indemnity period must reflect data centre recovery realities. A factory that suffers a fire can often resume partial production within weeks, with full production resuming over several months as damaged machinery is replaced. A data centre that suffers a significant fire or water damage event may need 6-12 months to fully recover, not because the building takes that long to repair, but because the electronic equipment procurement and installation timeline extends that far. Servers and networking equipment are manufactured to order, with lead times that range from 8 weeks for commodity hardware to 16-20 weeks for specialised configurations. During the 2021-2023 semiconductor shortage, lead times for certain server components exceeded 40 weeks.

The indemnity period for a data centre BI policy should therefore be set at 12-18 months as a minimum, and operators of large facilities should consider 24 months. The cost of extending the indemnity period is modest compared to the catastrophic exposure of a gap. An operator with an INR 200 crore annual revenue that suffers a major loss at month 8 of a 12-month indemnity period faces an uninsured revenue loss for every month beyond month 12 until full operations resume.

Fourth, the gross profit definition must be adapted. In manufacturing, variable costs (raw materials, direct labour) reduce proportionally when production stops, and BI covers only the fixed costs and net profit that continue regardless. In a data centre, almost all costs are fixed: rent or lease payments on the facility, debt service on the capital invested in infrastructure, staff salaries, software licences, and network transit charges (which are typically paid on take-or-pay contracts regardless of utilisation). Variable costs, primarily power consumption, do reduce when IT loads are shut down, but they represent only 30-40% of the total operating cost. This means the gross profit percentage for a data centre is typically 60-70%, significantly higher than most manufacturing operations, and the BI sum insured must reflect this.

A practical approach to calculating the BI sum insured for an Indian data centre: take the annual revenue, multiply by the gross profit percentage (typically 60-70%), and multiply by the indemnity period expressed as a fraction of a year. For a Tier III facility in Mumbai with INR 150 crore annual revenue, a 65% gross profit rate, and an 18-month indemnity period, the BI sum insured should be approximately INR 146 crore.

Data centre operators face a cyber liability exposure that is distinct from, and in many ways larger than, the cyber exposure faced by most other businesses. The distinction arises because a data centre operator does not just hold its own data; it holds, processes, and transmits data belonging to hundreds or thousands of customers. A breach at the data centre level can compromise the data of every tenant simultaneously.

The cyber insurance market in India has matured significantly since 2020, and most large data centre operators now carry some form of cyber cover. But the policy structure that works for a typical IT services company or e-commerce business does not translate well to a data centre operator. There are three reasons for this.

First, the data centre operator's liability to its customers for a data breach is contractual, not just regulatory. When a ransomware attack compromises customer data hosted in a colocation facility, the customers' first recourse is against the data centre operator under the terms of the hosting or colocation agreement. These agreements typically contain indemnity clauses that require the operator to hold the customer harmless against data breach losses, subject to negotiated liability caps. The liability caps in Indian colocation contracts vary widely, from 12 months of contract value for mid-market operators to uncapped liability for operators serving large enterprise or government clients.

A cyber insurance policy for a data centre must therefore provide adequate third-party liability limits to respond to these contractual indemnity obligations. Standard cyber policies in India offer limits of INR 5-25 crore, which may be adequate for a single-tenant breach affecting one customer but are grossly insufficient for a facility-wide breach affecting multiple tenants simultaneously. Large data centre operators should carry cyber liability limits of INR 50-100 crore or more, structured as a tower with a primary layer and one or two excess layers.

Second, the data centre operator sits in a complex position under India's data protection regime. The Digital Personal Data Protection Act, 2023 (DPDPA) imposes obligations on 'Data Fiduciaries' (entities that determine the purpose and means of processing personal data) and 'Data Processors' (entities that process data on behalf of a fiduciary). A data centre operator is typically a Data Processor: it provides the infrastructure on which its customers (the Data Fiduciaries) store and process personal data. However, the DPDPA holds Data Processors to their own set of obligations, including implementing reasonable security safeguards and cooperating with the Data Protection Board in the event of a breach investigation.

The penalties under the DPDPA can reach INR 250 crore per violation, though the actual penalty regime and enforcement procedures are still being developed through rules. A cyber insurance policy for a data centre operator should cover regulatory defence costs (the cost of responding to investigations by the Data Protection Board), regulatory fines and penalties to the extent insurable under Indian law, and the costs of breach notification and credit monitoring for affected individuals.

Third, there is a boundary problem between the property and cyber policies that is unique to data centres. Consider a scenario: an attacker gains access to the building management system (BMS) through an internet-connected vulnerability. The attacker manipulates the cooling controls, causing the CRAC units to shut down. Temperatures in the server hall rise to 60 degrees over four hours, causing permanent physical damage to servers worth INR 40 crore. Is this a property claim (physical damage to electronic equipment caused by overheating) or a cyber claim (damage resulting from a malicious cyber act)?

Under most Indian EEI policies, the physical damage to the servers would be covered because the EEI covers damage from any cause, including overheating. But many EEI policies now include a 'cyber exclusion' (typically NMA 2914 or NMA 2915 wording, adapted for the Indian market) that excludes damage arising from a cyber act. If the cyber exclusion applies, the EEI policy steps back, expecting the cyber policy to respond. But the cyber policy may argue that the loss is physical damage to tangible property, which is typically excluded from cyber policies (which are designed to cover intangible losses: data, business interruption, liability).

This gap, where the property policy excludes cyber-caused damage and the cyber policy excludes physical damage, is known as the 'silent cyber' problem. It is perhaps the single most dangerous coverage gap in a data centre insurance programme. Addressing it requires either a specific cyber-physical damage buyback endorsement on the EEI policy or a property damage extension on the cyber policy. Both options are available in the Indian market, but neither is standard, and both require explicit negotiation with the underwriter.

The annual premium for a data centre cyber policy in India varies widely based on the operator's security posture, revenue, data volume, and claims history. For a mid-size colocation operator with INR 100-200 crore revenue and INR 50 crore cyber limits, premiums typically range from INR 40-80 lakh, or approximately 0.8-1.6% of the limit. Operators with ISO 27001 certification, SOC 2 Type II reports, and demonstrated incident response capabilities receive preferential pricing.

If there is one loss scenario that every data centre insurer fears, and that standard policy wordings handle poorly, it is the cascading electrical failure. Understanding this scenario in detail is essential for anyone designing or underwriting a data centre insurance programme.

A cascading electrical failure begins with a single point of failure in the power distribution chain and propagates through interconnected systems until it causes widespread damage to IT equipment. The sequence typically unfolds as follows.

Step one: an initial fault occurs in the utility power supply or the facility's own electrical infrastructure. This might be a voltage transient on the incoming 33kV or 11kV utility supply, a failure in the main transformer, or a fault in the medium-voltage switchgear. In Indian data centres, utility power quality is a persistent concern. Voltage fluctuations, frequency deviations, and momentary outages are more common than in markets with mature grid infrastructure. Even data centres with dedicated utility feeds from substations experience power quality events several times per year.

Step two: the UPS system is called upon to absorb the disturbance and deliver clean, stable power to the IT loads. In a properly functioning system, the UPS (whether static, rotary, or hybrid) absorbs the transient and the IT equipment continues operating without interruption. But UPS systems can fail to respond correctly for several reasons: a firmware bug in the UPS controller, a battery string that has degraded below its rated capacity without detection, a faulty static transfer switch that fails to engage, or a configuration error that causes the UPS to shut down on a false alarm rather than riding through the disturbance.

Step three: when the UPS fails to deliver clean power, the disturbance passes through to the power distribution units (PDUs) and from there to the IT equipment. The IT equipment receives out-of-specification power: voltage too high, too low, or with excessive harmonic distortion. Some equipment has its own internal power supplies that can tolerate a degree of variation. Other equipment is more sensitive. The result is a differential failure pattern: some servers survive, others lose power abruptly (causing data corruption on spinning disk drives), and others suffer permanent hardware damage from overvoltage.

Step four: the diesel generators, which should start automatically on utility power failure, may or may not come online in time. Indian data centres typically specify a 10-15 second start-up and load transfer time for generators. If the UPS battery capacity (typically 5-15 minutes at full load) has degraded, the generators must start within the battery runtime or the IT loads will crash before generator power is available.

Step five: even when the generators start successfully, the return to utility power after the utility supply is restored creates another vulnerable transition. Retransfer from generator to utility power involves another switching event, and if the synchronisation between the generator output and the utility supply is imperfect, a second transient can damage equipment that survived the initial event.

This five-step cascade can unfold in seconds to minutes. The damage is typically spread across multiple subsystems: the UPS, the PDUs, and dozens to hundreds of servers and storage devices. The total loss can easily reach INR 20-50 crore in a mid-size facility.

Now consider how the insurance policies respond.

The SFSP policy, if it covers the electronic equipment at all, responds to fire and named perils. An electrical cascade that does not produce fire or explosion may not trigger the SFSP. Even with an 'electrical installation damage' endorsement, the SFSP may cover the damage to the electrical installation (the UPS, the switchgear) but not the downstream damage to the IT equipment.

The Machinery Breakdown policy covers the electromechanical equipment that failed: the UPS system, the static transfer switch, possibly the generator if it was damaged during the event. But the MB policy does not cover the servers and networking equipment that were damaged as a consequence of the power failure. The MB policy covers the machines that broke down, not the equipment damaged by the breakdown.

The Electronic Equipment Insurance policy covers the servers and networking equipment. But here the deductible and exclusion language matters. Some EEI policies exclude damage caused by 'failure of the external power supply' or limit coverage for damage caused by 'defects in the electricity supply.' If the EEI policy treats the UPS as the 'electricity supply' for the purpose of this exclusion (a debatable but not unreasonable interpretation), the servers damaged by UPS failure may be excluded.

The result is a three-policy pileup where each policy covers part of the loss, but the boundaries between them create grey zones. The UPS damage is covered under MB but not EEI. The server damage is covered under EEI but may face an exclusion argument. The building electrical infrastructure damage is covered under the SFSP but only with the correct endorsement. And the business interruption loss requires a valid underlying material damage claim, so if any of the material damage policies decline coverage, the BI claim is also at risk.

The solution is programme-level coordination. The policies must be placed with consistent trigger definitions, aligned deductibles, and explicit buyback endorsements that close the gaps between them. The cascading electrical failure scenario should be written into the programme as a 'agreed loss scenario' that has been modelled, with each policy's response pre-agreed. This is discussed in the final section.

A properly structured data centre insurance programme in India requires five to seven distinct policies, coordinated to operate as a single protective framework. The goal is to ensure that every foreseeable loss scenario triggers exactly one policy response, without gaps where no policy responds and without overlaps where multiple policies dispute which should pay.

Layer One: SFSP Property Policy. This covers the building shell, permanent electrical installations (transformers, HV/MV switchgear, bus ducts permanently affixed to the building), and the civil infrastructure (raised floors, fire suppression systems, cable trays). The SFSP is placed on a reinstatement value basis. The sum insured reflects the rebuilding cost of the physical structure, not the total facility value. Recommended endorsements: debris removal (enhanced), architects' and engineers' fees, additional cost of construction due to local authority requirements, and spontaneous combustion. The SFSP should explicitly exclude electronic equipment (covered under EEI) and movable machinery (covered under MB) to avoid coverage disputes.

Layer Two: Machinery Breakdown Policy. This covers all electromechanical plant: diesel generators, UPS systems (static and rotary), chillers, cooling towers, precision air conditioning units (CRAC/CRAH), pumps, compressors, automatic transfer switches, and medium-voltage switchgear that is not permanently affixed to the building. The MB policy is placed on a replacement value basis including import duty, freight, installation, and commissioning. The sum insured should include all machinery at every location covered by the programme. Key endorsement: 'surrounding property' damage extension, which covers damage caused to nearby property by the breakdown of an insured machine. This is relevant where a generator failure causes a fuel leak that damages the building floor or where a chiller refrigerant leak damages nearby equipment.

Layer Three: Electronic Equipment Insurance Policy. This covers all IT and electronic equipment: servers, storage systems, networking equipment, structured cabling, monitoring systems, security systems (CCTV, access control), and the data centre's own management and operations equipment. The EEI is placed on a replacement value basis, defined as the cost of replacing the damaged equipment with equipment of equivalent specification and performance available at the time of loss, not the original purchase price. The EEI should include: external data media restoration cover (covering the cost of restoring data from backup after a physical damage event), increased cost of working cover (covering the cost of renting temporary equipment to maintain operations during the repair period), and a cyber-physical damage buyback endorsement that confirms coverage for physical damage caused by a cyber event, closing the silent cyber gap.

Layer Four: Business Interruption Policy. This covers the gross profit lost during the period of interruption following physical damage to insured property. The BI policy should be linked to all three underlying material damage policies (SFSP, MB, and EEI), so that physical damage under any of the three triggers the BI cover. The sum insured is calculated as: annual revenue multiplied by the gross profit percentage, multiplied by the indemnity period (expressed as a fraction of a year, e.g., 18/12 for an 18-month indemnity period). Recommended extensions: utility interruption (covering BI resulting from failure of the external power supply, water supply, or telecommunications link, even when no damage has occurred at the insured premises), prevention of access (covering BI when civil authorities prevent access to the facility), and SLA penalty cover (covering the contractual penalties payable to customers following an insured outage). The indemnity period should be set at a minimum of 18 months for Tier III facilities.

Layer Five: Cyber Insurance Policy. This covers first-party costs (incident response, forensic investigation, data restoration, business interruption from a cyber event, extortion payments) and third-party liabilities (claims by affected customers and individuals, regulatory defence costs, regulatory fines to the extent insurable). The cyber policy should include explicit coverage for data centre operators' contractual liability to their colocation customers. The limit should be calibrated against the aggregate contractual liability exposure across all customer agreements, which for a mid-to-large operator typically requires INR 50-100 crore. The cyber policy should also include a 'failure to supply' extension covering the operator's liability when a cyber event prevents it from delivering services to customers.

Layer Six: General Liability and Professional Indemnity. The general liability (CGL) policy covers bodily injury and property damage claims arising from the data centre's operations. The professional indemnity (PI) policy covers claims arising from errors, omissions, or negligence in the managed services provided by the operator. For operators that provide only colocation (rack space, power, cooling), PI exposure is lower. For operators that provide managed hosting, cloud services, or security operations centre (SOC) services, PI exposure is significant.

Layer Seven (optional but recommended): Directors and Officers Liability. Covers the personal liability of the data centre company's directors and officers for wrongful acts in their capacity as directors, including regulatory investigations, shareholder claims, and creditor actions following a major loss event.

Programme coordination principles: All policies should be placed with the same renewal date to facilitate annual programme review. Deductibles should be aligned so that the BI waiting period equals or exceeds the material damage deductible across all underlying policies. Policy wordings should be cross-referenced: the BI policy should list all three material damage policies (SFSP, MB, EEI) as underlying covers, and the trigger language should confirm that damage payable under any of the three satisfies the material damage requirement. The cyber policy and the EEI policy should contain reciprocal language ensuring that no cyber-physical damage event falls between the two. Ideally, all policies should be placed through a single broker who can see the entire programme and identify gaps, rather than sourcing individual covers from different intermediaries.

The total premium for a well-structured data centre insurance programme in India varies based on facility tier, location, security measures, and claims history. As a benchmark, the total programme cost for a Tier III colocation facility with INR 200 crore total insured value (across all policies) and INR 150 crore annual revenue typically falls in the range of INR 80 lakh to INR 1.5 crore annually, representing roughly 0.5-1.0% of the total insured value. This is a modest investment against the catastrophic financial exposure that a major uninsured or underinsured event would create.