Indian IT big Tata Consultancy Companies (TCS) has secured $1 billion from personal fairness agency TPG as a part of a multi-year, $2 billion challenge to construct a community of gigawatt-scale knowledge facilities within the nation.

The challenge, dubbed “HyperVault,” comes as demand for AI compute is rising quicker than firms can construct the power-hungry infrastructure wanted to help it.

The demand-supply hole for AI compute in India is especially stark: the nation generates practically 20% of the world’s knowledge, however accounts for solely about 3% of worldwide knowledge heart capability. Large tech firms and cloud suppliers have been investing billions of {dollars} to broaden native capability and faucet the nation’s rising adoption of AI merchandise.

With HyperVault, TCS and TPG plan to develop liquid-cooled, high-density knowledge facilities with the facility and community capability required to help superior AI workloads throughout main cloud areas, the businesses stated.

Liquid cooling and high-density rack designs are rising widespread because the GPUs wanted to energy AI inference and coaching use considerably extra energy and generate extra warmth than standard CPU servers. However such designs additionally elevate questions on useful resource use in international locations like India, the place water shortage is already a priority.

In city hubs corresponding to Mumbai, Bengaluru and Chennai, the place a lot of India’s data-center capability is concentrated, current water stress might complicate operations. S&P International, citing Uptime Institute estimates, famous {that a} 1-MW knowledge heart load can require as much as 25.5 million liters of water a yr for cooling, including strain to already strained infrastructure.

The fast constructing of AI knowledge facilities stands to additional stress India’s energy and land use, two different bottlenecks recognized by business analysts. Excessive-density AI clusters require dependable electrical energy provide and enormous parcels of commercial land, two necessities more and more troublesome to safe in main city areas.

Nonetheless, world tech firms are treating India as their frontier for constructing AI infrastructure. Native and world expertise companies have introduced investments of greater than $32 billion during the last two years to broaden knowledge heart infrastructure within the nation, in response to S&P International.

In January, Microsoft stated it will make investments $3 billion over two years in India’s cloud and AI infrastructure, and in October, Google stated it will spend $15 billion over 5 years to construct a gigawatt-scale AI knowledge heart hub within the southern state of Andhra Pradesh. And again in 2023, Amazon dedicated $12.7 billion to construct AWS cloud infrastructure in India by way of 2030.

TCS stated it will work with hyperscalers and AI firms to design, deploy and function AI infrastructure because the platform expands. The corporate plans to construct round 1.2 gigawatts of capability in its preliminary part.

Greater than 95% of India’s new data-center capability over the following 5 years will come from leased services, and the rest will likely be pushed by hyperscalers constructing devoted AI infrastructure, S&P International estimates. Native gamers like Reliance Industries and CtrlS are additionally increasing their data-center capability to satisfy rising demand.

TCS and TPG challenge that India’s complete data-center capability might exceed 10 gigawatts by 2030, up from roughly 1.5 gigawatts at present.