Indian IT elephantine Tata Consultancy Services (TCS) has secured $1 cardinal from backstage equity steadfast TPG arsenic portion of a multi-year, $2 cardinal task to physique a web of gigawatt-scale information centers successful the country.
The project, dubbed “HyperVault,” comes arsenic request for AI compute is rising faster than companies tin physique the power-hungry infrastructure needed to enactment it.
The demand-supply spread for AI compute successful India is peculiarly stark: the state generates astir 20% of the world’s data, but accounts for only astir 3% of planetary information halfway capacity. Big tech companies and unreality providers person been investing billions of dollars to grow section capableness and pat the country’s increasing adoption of AI products.
With HyperVault, TCS and TPG program to make liquid-cooled, high-density information centers with the powerfulness and web capableness required to enactment precocious AI workloads crossed large unreality regions, the companies said.
Liquid cooling and high-density rack designs are increasing communal arsenic the GPUs needed to powerfulness AI inference and grooming usage importantly much powerfulness and make much vigor than accepted CPU servers. But specified designs besides rise questions astir assets usage successful countries similar India, wherever h2o scarcity is already a concern.
In municipality hubs specified arsenic Mumbai, Bengaluru and Chennai, wherever overmuch of India’s data-center capableness is concentrated, existing h2o accent could complicate operations. S&P Global, citing Uptime Institute estimates, noted that a 1-MW information halfway load tin necessitate up to 25.5 cardinal liters of h2o a twelvemonth for cooling, adding unit to already strained infrastructure.
The accelerated gathering of AI information centers stands to further accent India’s powerfulness and onshore use, 2 different bottlenecks identified by manufacture analysts. High-density AI clusters necessitate reliable energy proviso and ample parcels of concern land, 2 requirements progressively hard to unafraid successful large municipality regions.
Techcrunch event
San Francisco | October 13-15, 2026
Nonetheless, planetary tech companies are treating India arsenic their frontier for gathering AI infrastructure. Local and planetary exertion firms person announced investments of much than $32 cardinal implicit the past 2 years to grow information halfway infrastructure successful the country, according to S&P Global.
In January, Microsoft said it would invest $3 cardinal implicit 2 years successful India’s unreality and AI infrastructure, and successful October, Google said it would spend $15 cardinal implicit 5 years to physique a gigawatt-scale AI information halfway hub successful the confederate authorities of Andhra Pradesh. And backmost successful 2023, Amazon committed $12.7 cardinal to physique AWS unreality infrastructure successful India done 2030.
TCS said it would enactment with hyperscalers and AI companies to design, deploy and run AI infrastructure arsenic the level expands. The institution plans to physique astir 1.2 gigawatts of capableness successful its archetypal phase.
More than 95% of India’s caller data-center capableness implicit the adjacent 5 years volition travel from leased facilities, and the remainder volition beryllium driven by hyperscalers gathering dedicated AI infrastructure, S&P Global estimates. Local players similar Reliance Industries and CtrlS are besides expanding their data-center capableness to conscionable rising demand.
TCS and TPG task that India’s full data-center capableness could transcend 10 gigawatts by 2030, up from astir 1.5 gigawatts today.















English (US) ·