Nikhil Kamath and Sreeram Vanga Put Rs 250 Crore Into CtrlS as India's AI Compute Build-Out Goes Into Overdrive

Zerodha co-founder Nikhil Kamath is putting Rs 200 crore into CtrlS Datacenters. Entrepreneur Sreeram Reddy Vanga is adding another Rs 50 crore. The combined Rs 250 crore, announced on 19 August 2026, will fund infrastructure expansion at India's largest home-grown hyperscale operator at the exact moment the country's AI compute build-out is going into overdrive.

The deal

The two cheques come as personal investments, not fund allocations. Kamath, best known for building Zerodha into India's largest retail broker, has been steadily building an infrastructure portfolio across data centres, EV charging and grid storage over the last two years. Vanga is a Hyderabad-based entrepreneur and investor with prior bets in enterprise technology. Neither took a board seat as part of the transaction.

The Rs 250 crore lands in a company that already sits well up the capital stack. In June 2026, the Canada Pension Plan Investment Board committed up to Rs 7,000 crore to CtrlS — Rs 4,000 crore for an 8.2 percent equity stake, plus up to Rs 3,000 crore for a joint venture to build hyperscale campuses across India. The Kamath-Vanga round is a smaller, faster injection on top of that base to fund the immediate build-out.

Why now

CtrlS operates 19 data centres across nine Indian markets, with 370 MW of live capacity and roughly 4.4 GW at various stages of development. Founded by Sridhar Pinnapureddy in 2007, the Hyderabad-headquartered company was one of the earliest Indian operators to bet on Tier IV design standards. That bet is paying off now that Indian enterprise workloads, government data-localisation rules and — most of all — foreign hyperscalers looking for India capacity are all pulling on the same limited pool of high-density power and cooling.

The hyperscalers are the fastest-growing customer. Meta signed a 168 MW lease with Reliance in June for AI-enabled compute in India. Google's Vishakhapatnam campus commitment runs to USD 15 billion over five years. Microsoft is investing USD 17.5 billion in Indian data centres and AI infrastructure by 2029. Every one of those workloads needs somewhere physical to run.

Kamath, in a public note explaining his investment, framed it in terms of capacity — India, he wrote, either keeps pace with the compute demand its own AI market is generating or watches the workloads and the jobs they support move offshore. The Rs 200 crore is his way of taking that argument literally.

What CtrlS actually builds

The 370 MW that CtrlS operates today is a mix of colocation and dedicated hyperscale space in Mumbai, Hyderabad, Chennai, Bengaluru, Delhi NCR, Kolkata, Patna, Lucknow and Bhubaneswar. The company's newer campuses are being built as high-density facilities that can handle the 30-plus kW-per-rack density that modern GPU clusters need — the kind of infrastructure that Nvidia's Blackwell and Blackwell Ultra chips, and the AI factories built around them, demand.

That is not a trivial design challenge. Air cooling breaks down at those densities. Liquid cooling — direct-to-chip cold plates or full immersion baths — has to be engineered into the building from the ground up, along with the power distribution, water-treatment and heat-rejection systems that go with it. The 4.4 GW pipeline is largely being designed to those specifications.

The AI-datacentre supercycle, in one company

CtrlS is one of a handful of Indian operators — alongside Yotta, NTT, ST Telemedia GDC, Sify and CapitaLand's Nxtra — competing to sell India's growing appetite for AI-serving compute. The Kamath-Vanga cheque, coming on top of CPP Investments' anchor commitment, is a marker of how quickly private and long-term institutional capital is now converging on the same thesis: that data-centre capacity is the physical bottleneck for AI in India, and that the operators who can build fastest and to the highest specifications will capture disproportionate share.

That thesis has three ingredients: policy tailwinds (data-localisation, sovereign-AI compute commitments), demand pull (hyperscalers, Indian enterprise, government), and a supply-side scarcity of shovel-ready sites with power and water. CtrlS has been quietly building its land bank against all three for close to two decades.

What it doesn't do

The Rs 250 crore does not, on its own, change the industry's shape. It is a working-capital injection, sized to keep specific campuses moving through construction and commissioning. But taken together with the CPP commitment and the wave of hyperscaler leases now landing in the market, it is a data point on how seriously the smartest Indian capital allocators are taking the AI-datacentre build-out. Kamath, who typically writes cheques into consumer, financial services or listed-equity plays, deploying Rs 200 crore of personal capital into a single infrastructure company is not a routine trade.

For CtrlS, the money buys time and speed at a moment when neither is cheap. For the wider ecosystem, it is one more line item in a build-out that has now added, in six weeks, an AI cluster commitment from Meta, a Rs 10,000–15,000 crore NVIDIA B300 factory order from Together AI to L&T's Chennai campus, and a fresh capex round for the country's largest home-grown hyperscaler. India's AI compute plumbing is, quietly, being paid for.

Sources

More from Miscellaneous

T-Hub and Japan's OKI Electric Launch Program to Take Indian Deep-Tech Startups to Japan

Hyderabad-based incubator T-Hub has partnered with Japanese manufacturer OKI Electric to help eight shortlisted Indian deep-tech startups pilot water-infrastructure and logistics solutions in Japan.

4 min read

Geospatial Startup NeoGeo Raises $20 Million to Take India's Mapping Stack Global

Gurugram-based geospatial technology firm NeoGeo has raised $20 million in a Series A round led by Neev II Fund and Aavishkaar Capital, aiming to expand its GIS platforms and enter the Middle East and the Americas.