India has secured $300 billion in AI data centre commitments, but a report highlights a sharp mismatch between supply and demand. With GPU utilization near 20%, the sector faces questions about long-term returns and actual job creation. Investors are now tracking how firms like TCS and Infosys balance heavy infrastructure spending against traditional service-focused models.
India’s push to become a global hub for artificial intelligence infrastructure is hitting a practical reality check. While the country has seen over $300 billion in committed capital for data centres, with goals to reach 5 to 10 gigawatts of capacity by 2030, current usage levels suggest a significant gap between supply and demand. A recent report by the Rhodium Group indicates that subsidized infrastructure—specifically GPUs under the IndiaAI Mission—is largely underutilized, with utilization rates hovering around 20 percent.
Strategic Divergence Among IT Majors
This supply-demand mismatch is forcing India’s top IT companies to adopt vastly different strategies. Tata Consultancy Services (TCS) has moved toward an asset-heavy approach, establishing 'HyperVault' with a target of 1 gigawatt of capacity. This strategy involves significant capital spending on physical infrastructure to control the hardware layer of AI services. In contrast, Infosys has opted for an asset-light model, emphasizing capital discipline and focusing on the software and integration side of the business. For investors, this creates a clear point of differentiation: TCS is betting on the long-term value of owning the infrastructure, while Infosys is prioritizing returns on equity by avoiding heavy physical investments.
The Operational and Resource Reality
Beyond the capital spending debate, the sector faces scrutiny regarding its economic impact. State governments have offered incentives based on promises of mass job creation, but the actual on-the-ground operational headcount for these centres remains lean. Unlike traditional manufacturing, data centres require high automation and relatively fewer permanent staff once construction is complete. Furthermore, the physical footprint of these facilities poses resource risks. Many data centres are being built in water-stressed regions, leading to potential regulatory hurdles and concerns regarding power and water sustainability.
The 'Digital Export' Challenge
Currently, a large portion—roughly 60 percent—of India’s operational data centre capacity is utilized by foreign hyperscalers and international clients. While this brings in foreign revenue, it highlights that domestic enterprise demand is not yet high enough to fill the massive infrastructure being built. This risks turning these centres into 'digital export enclaves' rather than hubs for local AI innovation. Investors should monitor whether local businesses eventually adopt these services at scale, as domestic demand will be the ultimate driver of long-term profitability.
Looking ahead, the next phase for the sector will be defined by execution. The primary monitorables for investors include the pace of capacity utilization, the ability of companies to manage high energy and water costs, and whether government incentive frameworks are adjusted if employment targets are not met. The distinction between companies investing in their own infrastructure and those focusing on the 'services layer' will likely be a key factor in financial performance in the coming quarters.