NSE deploys direct-to-chip liquid cooling for AI workloads, plans expansion to 224 GPUs

The on-premises deployment will scale beyond 80 GPUs this quarter, with NSE planning to extend the cooling technology to its co-location data centre based on market demand.

The National Stock Exchange of India (NSE) has deployed Direct-to-Chip Liquid Cooling (DCLC) technology in its on-premises data centre to support its growing artificial intelligence (AI) workloads.

The infrastructure will accommodate up to 224 GPUs as the exchange expands its AI computing capacity.

NSE has already deployed multiple GPUs and plans to scale the environment to more than 80 GPUs during the current quarter.

The DCLC infrastructure has been designed to support this expansion across high-density racks, with capacity to scale up to 224 GPUs.

NSE describes the deployment as the first of its kind in Indian capital markets.

The exchange said the solution will support high-density GPU-based computing through a scalable cooling architecture as it expands AI use cases across the organisation.

The deployment comes as AI workloads place greater demands on computing infrastructure, particularly in terms of compute density, power and cooling.

Direct-to-chip liquid cooling is part of NSE's approach to preparing its data centre for these requirements.

NSE prioritises control over AI infrastructure

The DCLC solution has been deployed within NSE's on-premises data centre, allowing the exchange to retain control over its AI data, computing resources and access.

NSE said the approach strengthens data sovereignty and supports its security governance and applicable regulatory and compliance requirements as AI adoption scales.

NSE executive director Viral Mody said the exchange continues to invest in technology to build resilient, scalable and future-ready market infrastructure.

He said the liquid cooling deployment will help NSE prepare for evolving computing needs while advancing a more energy-conscious technology ecosystem.

Sampath Manickam, chief technology officer at NSE, said AI is driving higher requirements for compute density, power and cooling. The exchange's approach is to anticipate these requirements and build infrastructure ahead of demand.

He added that the DCLC deployment provides a scalable and energy-efficient foundation for high-density GPU computing and will support NSE's evolving technology requirements.

Co-location data centre expansion planned

NSE also plans to extend DCLC technology to its co-location data centre in the near future, depending on the needs of market participants.

The proposed expansion brings the cooling technology into the exchange's co-location infrastructure, although NSE has not disclosed its capacity, investment or implementation timeline.

The move could have implications for data centre infrastructure providers and technology companies working on high-density computing environments.

However, NSE has not identified the DCLC technology provider or implementation partners involved in the current deployment, or disclosed plans for external vendor participation in the next phase.

The exchange has also not specified the GPU models, number of racks, power capacity or quantified energy savings from the deployment.

The current rollout and planned co-location expansion reflect NSE's efforts to build infrastructure that can support growing AI computing requirements while retaining operational control over its technology environment.