A $740 million partnership backs a buildout that outpaces three decades of Indian data center history. The grid math is the part nobody has financed yet.
By Shashi Bellamkonda · July 30, 2026
India's projected data center capacity by 2030, up from 2.2 GW today
Projected growth in AI-dedicated capacity over the same five years
CPP Investments commitment to CtrlS, announced June 17
Additional grid capacity CtrlS founder Sridhar Pinnapureddy says India needs
CtrlS built the confidence to take a $740 million partnership by financing its first data centers on customer prepayments instead of debt. The company now describes its own buildout as nearly matching what India has built in three decades. The constraint on that pace is not capital. It is a grid that has not caught up.
Sridhar Pinnapureddy remembers persuading Indian banks to trust an unknown startup with their data over established telecom incumbents. That was 2007. Nineteen years later, CtrlS operates about 370 megawatts of live capacity and has 4.4 gigawatts under development, and Pinnapureddy describes the company's current buildout as nearly equal to what the entire country has built since data centers first arrived in India (Bloomberg, 2026).
Canada Pension Plan Investment Board is financing part of that ambition directly. CPP Investments announced a 70 billion rupee partnership with CtrlS on June 17, split into a 40 billion rupee purchase of an 8.2 percent equity stake and a 30 billion rupee commitment to a joint venture that will develop new data center campuses across India (CPP Investments, 2026). CPP Investments holds 48 percent of the venture. CtrlS holds the rest.
The deal fits a pattern this publication has tracked before. CtrlS shows up as one of the domestic operators Nvidia-backed neoclouds are courting to lease inference capacity under India's new tax holiday, a deal structure this publication examined in a post on The Economic Times' reporting on that leasing wave. CtrlS also appeared as a competitor in NTT's own capital-structure story, a four gigawatt bet this publication argued depended more on financing mechanics than physical capacity. The CPP deal answers the financing question for CtrlS specifically. It does not answer the grid question for anyone.
The country is building faster than it ever has
Wood Mackenzie projects India's operational data center capacity will rise from 2.2 gigawatts in 2025 to 12 gigawatts by 2030, a compound annual growth rate of roughly 40 percent (Wood Mackenzie, 2026). AI-dedicated capacity within that total grows faster still, expanding nearly 24-fold from 275 megawatts to 6.5 gigawatts over the same five years.
Wood Mackenzie's own framing matters here. Land and capital, the firm argues, are no longer the limiting factors for data center developers in India. Reliable, cost-competitive electricity has taken their place as the primary constraint on where and how fast anyone builds (Wood Mackenzie, 2026). Pinnapureddy is making the same argument from inside a company that is trying to build through it.
Pinnapureddy's own number is the one to watch
India needs another 100 gigawatts of grid capacity over the next decade to become the data center capital of the world, Pinnapureddy told Bloomberg (Bloomberg, 2026). That is his estimate, made as the founder of a company raising capital to build into that gap, not an independent forecast. It is also roughly eight times the total capacity Wood Mackenzie projects the entire data center sector will reach by 2030.
Global electricity demand from data centers rose more than 15 percent in 2025 and is expected to nearly double by 2030, according to the International Energy Agency (IEA, 2026). India is adding generation capacity to meet that curve. Battery storage and transmission are the pieces lagging behind, according to Bloomberg's reporting, which named grossly outdated transmission infrastructure as the binding constraint rather than power generation itself.
"India needs another 100 GW of grid capacity over the next decade if it wants to become the data center capital of the world."
That single sentence, from the founder of one of the companies asking investors to fund the buildout, is the entire viability question compressed into one line.
Water looks solved. Power does not.
Pinnapureddy says only about 10 percent of CtrlS capacity still depends on water-based cooling, with a shift to recycled greywater expected within three years (Bloomberg, 2026). Wood Mackenzie treats water as a real but manageable risk, flagging it as underappreciated in water-stressed states such as Tamil Nadu and Karnataka while noting that developers adopting closed-loop cooling and zero liquid discharge systems are already ahead of likely future regulation (Wood Mackenzie, 2026). Power sits in a different category. Nobody in this reporting, including the executive with the most at stake, describes the grid gap as already being closed.
The map is shifting toward cheaper power, not cheaper land
Maharashtra and Tamil Nadu currently hold about 65 percent of India's installed data center load. Wood Mackenzie expects Andhra Pradesh and Telangana to take share from both states, driven by cheaper power procurement and state-level fiscal incentives rather than land costs (Wood Mackenzie, 2026). CareEdge Ratings frames the underlying opportunity in per-capita terms: India has 1.2 megawatts of data center capacity per million internet users, against a global average of 5 megawatts (CareEdge Ratings, 2026). Construction costs run 30 to 40 percent below China and the United States, largely on cheaper land and electricity.
The state-level shift toward Andhra Pradesh and Telangana is a power story before it is a policy story. Operators are moving toward the grid capacity that exists, not waiting for the 100 gigawatts that does not.
Unknowns and uncertainties
Pinnapureddy's 100 gigawatt figure has not been independently validated by a grid planning body in this reporting. Whether India's transmission buildout tracks the sector's 40 percent compound annual growth rate, lags it by a year, or lags it by five, is not yet answerable from public data. The CPP joint venture's 48 percent stake gives the pension fund meaningful influence over new campus development, and how that influences site selection toward power-secure states versus legacy hubs is not disclosed in either company's public statements.
Before signing a multi-year colocation contract in India, ask the operator for a state-specific power interconnection timeline, not a national capacity forecast. Pinnapureddy's own 100 gigawatt estimate is the person building the facility telling you the grid is not there yet. Get that commitment in writing before it becomes your outage.
Sources
Doshi, Menaka. "Compute Capital." India Edition, Bloomberg, 30 July 2026, bloomberg.com.
"India Data Centre Capacity to Reach 12 GW by 2030." Wood Mackenzie, 27 July 2026, woodmac.com.
"CPP Investments Commits C$1 Billion to CtrlS Data Centre Partnership in India." CPP Investments, 17 June 2026, cppinvestments.com.
"India's Data Center Capacity Gap." CareEdge Ratings, 2026, careedge.in.
"Electricity 2026." International Energy Agency, 2026, iea.org.
Bellamkonda, Shashi. "NTT's 4 Gigawatt Bet Is a Capital Structure Story, Not a Capacity Story." shashi.co, 25 Mar. 2026.
Bellamkonda, Shashi. "CoreWeave and Five Other Nvidia-Backed Neoclouds Seek to Lease India Data Center Capacity." shashi.co, 29 July 2026.
