"India's Yotta Wants $1.5B From Public Markets to Build Out the AI Cloud"

The race to build AI infrastructure has a new front, and it isn't in the United States or Europe. Yotta Data Services, the Hiranandani Group-backed data center operator, plans to list on the public markets in the January–March window of 2027, aiming to raise up to $1.5 billion — a number that puts India's AI buildout squarely on the global map. Reuters reported the plan, with CEO Sunil Gupta confirming draft papers are due to be filed this month. The stated uses are telling: repay debt, buy graphics processing units, and expand sovereign cloud capacity — infrastructure that keeps data inside national borders.

What makes this more than a routine IPO headline is the sequencing. Yotta is deliberately shrinking the public portion of its raise because it already filled much of the gap with private capital. Gupta disclosed last month that the firm took in $150 million in primary growth capital at a valuation of roughly ₹370 billion, or about $3.9 billion. In other words, private investors are doing the risky early work of proving the business, and the public listing is being kept as a smaller, cleaner top-up. That's a deliberate two-stage de-risking strategy, and it says something about how mature the AI-infrastructure asset class has become.

The first thing worth noticing is that this is a balance-sheet story as much as a growth story. Data centers are ferociously capital-intensive: GPUs are expensive, power contracts are long, and buildouts run years ahead of revenue. A firm that funds that expansion with debt eventually hits a wall where every new megawatt strains the books. Moving to equity — first private, now public — converts an interest-bearing liability into permanent capital that can absorb the long lag between laying concrete and billing customers. Yotta's decision to repay debt with IPO proceeds is the clearest signal that it's rebalancing, not just expanding.

The second insight is buried in an unusual financing structure Gupta described: partners would buy GPUs through special-purpose vehicles, share the revenue those chips generate, and then transfer ownership to Yotta after four to five years. If that sounds familiar, it's because it's essentially aircraft leasing applied to silicon. Instead of shouldering the full upfront cost of Nvidia hardware that depreciates fast, Yotta lets a partner carry the asset, shares the yield while it's productive, and takes title only after the chips have paid for themselves. It's a clever way to defuse the single most punishing cost in the AI data center business.

Positioning matters too. Gupta frames India as the "green market for everybody" precisely because the usual expansion paths are congested. Power constraints and GPU shortages are slowing buildouts in the United States and Europe, while geopolitical tensions make the Middle East an uncomfortable bet for some customers. India, by contrast, offers room to scale — and the Indian government sweetened the deal in February with a 20-year tax holiday for foreign firms that use local data centers. That's a demand-side subsidy that turns a regulatory friction point into a competitive moat, and it's already working: 75–80% of Yotta's customer base is now global, not domestic.

That last figure is the quiet revolution in the whole story. Traditionally, data centers were built where demand already existed — local businesses needed local capacity. India's AI infrastructure is increasingly being built for export, serving customers who are themselves global hyperscalers. When the majority of your clients come from abroad, the data center stops being regional plumbing and becomes a national export industry, which is exactly how the government's tax policy is treating it. Google and Amazon are both expanding their Indian footprints, and that inbound demand is what a listing like Yotta's is ultimately priced against.

The scale is real, not aspirational. Yotta already operates what it calls India's largest Nvidia-powered AI computing footprint, and its flagship campus near Mumbai was billed at inauguration as Asia's largest data center, with 210 MW of overall capacity and Tier IV certification. Tier IV isn't a marketing label; it's an uptime standard for fault-tolerant, concurrently maintainable infrastructure, and it's the kind of credential that matters to banks, government agencies, and any customer whose data can't go down. A listing gives the public a way to buy into that asset base directly.

The balanced read is that this is still a bet, not a certainty. Yotta hasn't disclosed revenue, and its sovereign-cloud ambitions put it in competition with far larger, better-capitalized hyperscalers who increasingly build their own facilities rather than rent someone else's. An IPO in early 2027 also lands against a market that will have to absorb a wave of data-center listings globally, and investor appetite for AI infrastructure won't stay bottomless forever. But the direction of travel is hard to argue with: India has the power headroom, the policy support, and a growing roster of global customers that the US and European markets are struggling to serve.

For the broader industry, Yotta's move is a leading indicator. When a data center operator goes public, it's usually a sign the market believes demand is durable enough to finance at scale rather than project by project. The fact that the firm can raise pre-IPO capital at a $3.9 billion valuation, file draft papers, and still hold the IPO in reserve as a smaller event suggests the private markets have already voted. India's AI cloud is being built — the only question left is how much of it the public will get to own.

Further reading: - Reuters via ET EnterpriseAI: India's Yotta targets Jan–March 2027 IPO, seeks up to $1.5 billion amid AI boom - Yotta Data Services — the company's sovereign cloud and Nvidia-powered AI infrastructure offerings - News18: Asia's largest data centre with 210MW capacity inaugurated near Mumbai