NextFin News - India's Yotta Data Services is preparing to tap public markets for an initial public offering of up to $400 million in the fourth quarter of this fiscal year, part of a broader $1.5 billion fundraising push that will decide whether the Hiranandani Group-backed data centre operator can convert its early lead in AI compute into a durable national infrastructure franchise. The chairman's latest comments on the IPO pipeline and growth underscore a company racing to scale ahead of rivals who are now chasing the same AI infrastructure dollar.
The Deal: A $1.5 Billion Bet Split Between Private and Public Markets
Yotta Data Services is targeting a total raise of $1.5 billion by the end of fiscal 2027, structured as a pre-IPO round followed by a domestic listing, co-founder and chief executive Sunil Gupta said in an interview. The public offering itself is expected to raise as much as $400 million, with the bulk of the capital coming from institutional investors before the listing.
"Our $1.5 billion fund-raise is a combination of pre-IPO fundraising, as well as an IPO. We are targeting Q4 of this fiscal year, subject to Sebi approval - we're working towards ensuring compliance," Gupta said. "The idea is that a big part of this $1.5 billion may get closed in the pre-IPO round itself, for which we're in touch with large funds based in the US, and also those with India offices."
The final size of the IPO will be dictated by how much money the company raises beforehand, Gupta said - a sequencing choice that signals management wants to de-risk the public debut by locking in anchor institutional capital first. That approach reflects a sober reading of the market: a $400 million IPO for a pre-profit data centre company with heavy capital needs is not a sale that can be rushed.
The company has already raised approximately $150 million from high-net-worth individuals and family offices at a valuation of around $3.8 billion, with no promoter stake sold and all proceeds being deployed into the business. Management has indicated it is now in discussions with larger institutional and sovereign wealth investors, and has set a target valuation of $5.5 billion to $6 billion by the time of the listing.
Yotta had earlier planned to list in the United States but has shifted its focus to India. The rationale is rooted in the nature of its fastest-growing business line.
"(My) business proposition is sovereign. My complete build-out is in India, and the government is a huge customer for my entire sovereign cloud - which India's private sector will also adopt as they understand geopolitical risks," Gupta said. "The nudge from the government for this was clear: why are you raising (funds) outside? Why not within India?"
The pivot from a US listing to a Mumbai IPO is more than a change of venue. It is a statement that Yotta's growth thesis is tied to India's data sovereignty regime - Reserve Bank of India rules requiring payment data to be stored domestically, and Securities and Exchange Board of India requirements for market participants - rather than to the global AI trade that values pure-play GPU capacity.
The Growth Engine: GPUs, Sovereign Cloud, and a Capacity Sprint
Yotta's expansion plan is capital-intensive by design. The company reported revenue of ₹890.7 crore and a profit of ₹11.1 crore for the year ended March 2025, according to disclosures made by the privately held firm with India's corporate affairs ministry. Against that base, it is pursuing a far larger capital programme: raising $8.5 billion by 2029 while more than quadrupling installed data centre capacity from 180MW to 800MW.
Much of that capital will finance graphics processing units and the high-density power, cooling, networking, and storage infrastructure needed to rent that compute to customers. Yotta expects to increase the number of Nvidia chips installed across its data centres from 37,000 to 85,000 by the end of this fiscal year. The company already owns between 60% and 70% of India's GPU capacity, Gupta has said.
The centrepiece of that build-out is a 60MW hyperscale facility in Greater Noida, scalable to 250MW, which will host 20,736 liquid-cooled Nvidia Blackwell Ultra GPUs - one of Asia's largest AI superclusters - representing an investment exceeding $2 billion and scheduled to go live by August 2026. The project includes a $1 billion Nvidia DGX Cloud cluster under a four-year agreement.
"This NVIDIA Blackwell Ultra supercluster reinforces India's position in the global AI value chain," Darshan Hiranandani, co-founder and chairman of Yotta, said when the deployment was announced.
The business model is also shifting to reduce the risk of owning expensive hardware. Rather than buying GPUs based on expected demand, Yotta is increasingly deploying capacity against customer contracts, with some clients signing five-year take-or-pay agreements that give the company a clearer path to recover its investment. Around 80% of Yotta's current GPU capacity is used by customers in the United States and Europe, and management expects both its colocation and sovereign cloud businesses to grow at roughly 25% a year, with sovereign cloud potentially reaching an annual revenue run rate of about $300 million from government customers alone.
The Second-Order Question: Is This a Data Centre Listing or a Sovereign-Cloud Listing?
The first-order read of Yotta's IPO is straightforward: India's data centre market is growing, AI is driving demand for compute, and Yotta is raising money to build more of it. That is true as far as it goes, but it misses the valuation question that will decide whether the listing succeeds.
Conventional data centre operators are valued on contracted, annuity-like rental income - stable, predictable, and capital-intensive. AI infrastructure platforms are valued on growth optionality - the chance that scarce GPU capacity commands premium pricing for years. Yotta is attempting to be priced as the latter while its financials still look like the former: revenue of ₹890.7 crore and a profit of ₹11.1 crore on a business whose core is colocation.
The transmission mechanism runs through the customer mix. If Yotta's growth comes predominantly from colocation and sovereign cloud contracts with Indian government and enterprise clients, the market will price it as an infrastructure utility - a respectable multiple on contracted cash flows, but not an AI re-rating. If, instead, the company can demonstrate that its GPU capacity is earning premium, export-oriented returns with long-dated take-or-pay contracts from overseas clients, the AI infrastructure premium becomes defensible.
That distinction matters because the $1.5 billion raise is not just funding growth - it is funding a bet on which business Yotta actually is. The 85,000-GPU target, the $2 billion Blackwell deployment, and the $8.5 billion capital plan through 2029 only make sense if management believes AI compute will remain scarce and premium-priced through the end of the decade. If GPU supply catches up with demand and pricing normalises, Yotta would be left carrying a balance sheet built for a scarcity environment into a commodity one.
There is also a second-order currency and capital-market dimension. By listing in India rather than the United States, Yotta gains alignment with its sovereign-cloud customers and avoids the scrutiny a US-listed AI infrastructure story would attract. But it also limits its investor base to domestic institutions and a smaller pool of India-focused global funds - precisely the investors who are most likely to price it as an Indian infrastructure name rather than a global AI play. The sovereign pivot solves a geopolitical problem and creates a valuation one.
The Counter-Thesis: India's Data Centre Boom Is Already Priced, and Yotta Is Late to the Race
The strongest case against Yotta's thesis is that the opportunity is real but no longer underpriced - and that the company is racing into a market that is about to become crowded, capital-saturated, and margin-compressed.
Yotta's expansion comes as India's data centre market enters a major investment cycle. The country had 1.5GW of data centre capacity as of mid-March 2026, according to a central government press note, and industry research projects capacity nearly doubling to around 4GW by 2028. Against that backdrop, competitors are raising equally large sums: Airtel-backed Nxtra raised $1 billion from private-equity investors Alphawave, Carlyle, and Anchorage Capital in March to build 1GW of capacity; Hyderabad-based CtrlS Datacenters raised $425 million from Canada's public pension fund to expand from 370MW to 2GW over five years; and Adani and Reliance Industries have each outlined plans to spend more than $100 billion over the coming decade on data centres.
"At $1.5 billion, Yotta's proposed capital raise would be significant even by the standards of India's rapidly expanding data-centre market," said Ashish Banerjee, senior principal analyst at Gartner. "But, it is not an apples-to-apples comparison with every other data centre fundraise. Yotta is seeking to build a GPU-intensive AI infrastructure platform, in which the investment extends beyond traditional data-centre capacity to include GPUs, high-density power and cooling, networking, and storage."
Banerjee's caveat cuts both ways. It validates Yotta's differentiation, but it also highlights the execution risk: GPU-intensive infrastructure carries technology obsolescence risk that conventional colocation does not. A Blackwell Ultra cluster bought at a premium in 2026 must earn its keep before the next generation renders it less competitive. In a market where every large Indian conglomerate is now building AI capacity, the pricing power that justifies Yotta's valuation target could erode faster than the debt used to finance it can be repaid.
The counter-thesis, in short, is that Yotta is not late to AI - it is early to commoditisation. If GPU supply normalises and Indian data centre capacity doubles as projected, the company's 60-70% share of GPU capacity becomes a shrinking share of a much larger pie, and its premium valuation multiple compresses toward utility levels.
The signal that would prove this counter-thesis right is concrete: if Yotta's GPU utilisation falls materially below its current ~80% level, or if new GPU capacity in India grows faster than contracted demand such that effective pricing per GPU-month declines for two consecutive quarters, the AI-premium thesis is broken and the stock should be priced as a conventional data centre operator.
What to Watch: The Falsifiable Path Through FY27
The near-term catalysts are well defined. The DRHP filing with SEBI is expected in the coming weeks, with the IPO targeted for the fourth quarter of fiscal 2027. Investors will watch three things: how much of the $1.5 billion closes in the pre-IPO round, which institutions anchor it, and whether the company can hold its $5.5-6 billion valuation target in that private round - because the pre-IPO price will set the anchor for the public offering.
Over the medium term, the key metric is not revenue growth but the revenue mix. A sovereign cloud run rate approaching $300 million from government customers alone would validate the domestic-listing thesis. Conversely, a failure to convert GPU capacity into contracted, take-or-pay revenue would expose the gap between the company's AI narrative and its utility-like fundamentals.
Over the long term, the question is whether India's data sovereignty regime creates a durable moat or merely a temporary shelter. If regulatory data-localisation requirements tighten and the government continues to favour domestic providers for sovereign workloads, Yotta's first-mover position in GPU capacity and its government relationships compound into a defensible franchise. If those requirements soften or if hyperscalers find compliant ways to serve the Indian market directly, the moat narrows quickly.
Scenarios for the listing:
- Base case: Yotta closes most of the $1.5 billion in the pre-IPO round at a valuation near the top of its $5.5-6 billion target, files the DRHP on schedule, and lists in Q4 FY27 with the IPO raising close to $400 million. The stock trades as a growth-oriented infrastructure name, supported by contracted sovereign cloud revenue.
- Upside case: GPU scarcity persists through 2027, take-or-pay contracts from overseas clients expand, and the market prices Yotta as an AI infrastructure platform rather than a data centre utility - supporting a valuation above the $6 billion target.
- Downside case: The pre-IPO round closes below target, forcing a larger IPO in weaker market conditions; GPU pricing normalises as Indian capacity doubles; and the listing is delayed or priced below the $5.5 billion floor, repricing the stock toward conventional data centre multiples.
The short-term trade will turn on IPO sequencing and anchor investors. The medium-term fundamentals will turn on the revenue mix. The long-term outcome will turn on whether data sovereignty proves to be a structural regime change or a cyclical policy tailwind.
Yotta's chairman is right that India's AI infrastructure moment is real. The harder question is whether a $1.5 billion raise at a $6 billion valuation is pricing that moment - or racing past it.
Explore more exclusive insights at nextfin.ai.
