India’s data centre boom is creating significant pockets of concentrated risks. These risks need to be identified and managed early and insurance covers to be designed around the exposures. Howden India’s Mr Amit Agarwal speaks with Asia Insurance Review about how insurance can help operators protect their assets and avoid gaps in cover as the sector expands.
India is experiencing one of the fastest expansions of data centre infrastructure globally due to AI, cloud computing, data localisation requirements, digital payments, 5G adoption and the rapid growth of digital services, billions of dollars are being invested in hyperscale facilities across Indian metros.
India’s operational data centre capacity has already surpassed 1.5 gigawatts (GW), with a significant future pipeline under development, projected to grow to nearly 6.5GW by 2030. While industry discussions often focus on computing power, connectivity and land availability, a more fundamental issue is resilience.
Speaking to Asia Insurance Review, Howden India CEO and MD Amit Agarwal said, “The increasing concentration of computing power within large hyperscale campuses has amplified this dependency. As facilities become larger and more interconnected, a single incident can create widespread business interruption across multiple industries.
“Recently, we encountered a Graphics Processing Unit (GPU) fit-out proposal where GPUs worth $300m were being deployed within a three-storey structure. It highlighted the sheer concentration of asset values now housed within modern AI infrastructure.”
Mr Agarwal said, “This growing criticality is changing expectations around resilience. Like airports, power stations and transmission networks, data centres are increasingly expected to withstand operational disruptions, extreme weather events, cyber incidents and infrastructure failures without compromising service continuity.”
Power: the defining risk
Speaking about the variety of risks that data centres face, Mr Agarwal said, “For most operators, power availability has become the single most important factor influencing both operational resilience and insurability.
“AI-driven workloads are significantly increasing energy demand, placing unprecedented pressure on electricity infrastructure. Industry research highlights power availability as a key determinant of where future data centres can be built and how quickly projects can become operational. Rising energy requirements are also driving developers to explore alternative solutions, including on-site generation, battery storage systems and hybrid energy infrastructure.”
“These solutions, however, introduce new risks. Battery storage systems create fire and thermal runaway exposures. On-site generation adds operational and maintenance risks. Increased electrical density can intensify the severity of equipment failures and business interruption losses.”
Insurers are increasingly looking beyond power availability to assess grid and substation reliability, redundancy and a facility’s ability to operate through prolonged disruptions.
A new insurance challenge
Traditionally, the insured value of a facility was concentrated in buildings, electrical infrastructure, cooling systems and conventional server hardware. Increasingly, AI-focused data centres are investing millions of dollars in high-performance GPUs and specialised AI accelerators. Unlike traditional infrastructure assets, GPUs present unique insurance challenges that include high value, rapidly evolving technology and retaining significant residual value even after newer generations are introduced.
Mr Agarwal said, “Older GPUs often continue to serve machine learning, cloud computing and lower-intensity AI workloads, creating an active secondary market. This raises important questions around valuation methodology, depreciation assumptions, salvage recovery and claims settlement.
“Determining whether a GPU should be valued on a replacement cost, current market value or residual value basis can materially impact insurance claim outcomes.”
Outside India, discussions have already begun around insuring the residual value of GPUs. The concept is straightforward:
- GPU purchased for $30,000
- Expected residual value after five years = $10,000
- Actual market value after five years = $5,000
While this product is not yet widely available for GPUs in India, discussions are emerging globally around AI infrastructure financing and equipment leasing.
Accumulation risk
Mr Agarwal said, “Accumulation risk is another major concern. The concentration of large GPU clusters within a single facility increases exposure significantly. A fire, flood, power event or cooling system failure can potentially damage assets worth tens or even hundreds of millions of dollars within a relatively small footprint.
“As the AI ecosystems mature, insurers will increasingly need to understand not only the physical infrastructure that supports data centres but also the economic lifecycle and residual value characteristics of the computing assets housed within them.”
Very often the larger data centre markets are concentrated in locations facing varying degrees of climate-related exposure.
He said, “Flooding, extreme rainfall, cyclones, heatwaves and water stress are increasingly influencing site selection decisions and long-term operational planning. Facilities located in coastal regions face greater scrutiny due to potential exposure to rising sea levels and extreme weather events.”
Higher temperatures directly increase cooling requirements, driving energy consumption and operational costs. Water scarcity is becoming another critical concern, particularly as advanced cooling systems remain heavily dependent on reliable water supplies.
He said, “Historical loss data alone may no longer be sufficient as climate patterns become less predictable. Resilience planning, catastrophe modelling and risk engineering are becoming increasingly important in infrastructure development and insurance placement. Parametric products may also offer a potential solution for certain climate-related exposures.”
No physical damage losses
One of the most significant challenges facing data centre operators is that many outages occur without any physical damage. A software malfunction, control system failure, utility disruption or cooling system issue can cause a facility to go offline without triggering traditional property insurance policies. Yet the resulting financial impact may be substantial.
Mr Agarwal said, “Situations such as above mentioned, create a significant protection gap between the exposures operators face and the coverage provided by conventional insurance systems. Industry studies increasingly identify non-damage business interruption (NDBI) as a major concern for modern data centre operators.”
He said such issues are particularly relevant because many facilities, especially in India, already operate with extensive redundancy through UPS systems, backup generators and cooling infrastructure. While redundancy reduces the likelihood of failure, it does not eliminate it.
“Even highly resilient facilities remain vulnerable to operational events that do not involve physical damage. As a result, insurers are exploring alternatives such as parametric covers and broader solutions designed specifically for digital infrastructure risks.”
New risk environments
Mr Agarwal said, “Many data centre operators are constructing new extensions and structures adjacent to existing live facilities. This creates a unique risk environment where construction activity takes place alongside highly sensitive operational infrastructure.
“Potential consequences include physical damage, equipment breakdown, testing and commissioning failures, and business interruption affecting existing tenants.”
Between 2026 and 2030, planned expansion at existing data centres is equivalent to 78% of their current footprint. Construction and operational risks therefore need to be assessed within a single programme.
Hence, global best practices increasingly assess risks across the entire lifecycle, from design and construction to commissioning and operations, providing greater visibility of interconnected exposures.
Insurance for resilience
Mr Agarwal said, “For many years, insurance was viewed primarily as a compliance requirement or a procurement exercise completed shortly before operations commenced.
That perspective is now evolving.
“As data centres grow in size, become larger, become more interconnected and more critical to economic activity, insurance is increasingly becoming an integral strategic component of risk management.”
“Operators are engaging insurers, brokers and risk engineers right from the earliest stages in the project lifecycle, bringing resilience considerations into decisions around site selection, flood protection, power infrastructure, redundancy architecture and business continuity planning.”
He said, “This improves insurability, strengthens investor confidence and supports project financing and long-term operations. It also provides greater confidence to lenders, investors and operators that all major risks have been identified and addressed.
“The next phase of growth of data centres will be defined not only by how much infrastructure is built, but by how resilient that infrastructure proves to be. In that context, insurance is evolving beyond its traditional role as a risk-transfer mechanism and becoming an increasingly important part of the broader resilience framework supporting our digital infrastructure,” he said. A