Yotta's IPO Is Partly a Plan to Pay Off the Debt That Built It
India's newest AI-infrastructure unicorn is stacking family capital, bank debt, GPU-backed vehicles, and now a public listing to fund a $2 billion compute bet, and the IPO paperwork itself says some of that money is going straight to paying down what came before it.

Yotta Data Services wants to raise up to $1.5 billion in an IPO targeted for January to March 2027. Part of that money, according to Reuters reporting picked up this week, is earmarked to repay debt. That's an unusual thing to see spelled out plainly in pre-IPO commentary: a company going public partly to pay off the borrowing that got it to the point of going public.
Most coverage of Yotta this year has focused on the valuation headline and the GPU count. Both are real and both are large. But the more useful story is the financing stack underneath them, because it's more complicated, and more leveraged, than "AI infrastructure unicorn raises money" usually implies.
What Yotta has actually raised, and how
Yotta, backed by the Hiranandani Group and founded in 2019 by Sunil Gupta and Darshan Hiranandani, runs data center campuses in Mumbai, Gujarat, and near New Delhi. In February, Reuters reported the company planned to spend over $2 billion on Nvidia's latest chips to build an AI compute hub, and was seeking to raise up to $1.2 billion from investors ahead of a possible IPO later that year. A source told Reuters that Abu Dhabi's Mubadala sovereign wealth fund was in talks to participate in that pre-IPO round, though Yotta declined to comment and Mubadala didn't respond.
By July, the company had closed roughly $150 million in primary growth capital from non-institutional investors, at a valuation reported as ₹37,000 crore. What that converts to in dollars depends on which outlet you read: Dealroom put it at $4.4 billion, while a separate report citing CEO Sunil Gupta's own LinkedIn post put the same rupee figure at $3.9 billion. That's a half-billion-dollar gap on the same underlying number, which says less about Yotta and more about how loosely dollar figures get attached to Indian funding rounds when everyone's in a hurry to publish the valuation.
The company said the July round was entirely primary capital, with no promoter selling shares, and the funds are going toward scaling its AI cloud to more than 40,000 Nvidia Blackwell GPUs within months and around 85,000 by the end of the current financial year, a target that, if hit, would reportedly make it one of the largest AI compute platforms outside the US and China.
The part that doesn't fit a normal unicorn story
Here's where it gets more interesting than a straightforward capital-raise narrative. According to the most recent Reuters reporting, Yotta is also exploring financing structures where outside partners buy the GPUs themselves through special-purpose vehicles, share in the revenue those chips generate, and then transfer ownership to Yotta after four or five years. That's off-balance-sheet financing for hardware, a structure more familiar in aviation leasing or power-plant financing than in a venture-backed tech company's cap table.
Layer that against the rest of the stack: family capital from the Hiranandani Group, bank debt, the $150 million primary round, a planned $1-1.2 billion pre-IPO raise (though Gupta now says the IPO portion will likely be smaller than originally planned because much of the target has already been met through other channels), and now a public listing seeking up to $1.5 billion. That is at least five distinct financing types being run in parallel to fund one buildout. Each individually is a normal way to finance infrastructure. Run together, on a compressed timeline, for a single asset class as volatile as GPU capacity, it's a materially more leveraged structure than the "unicorn valuation" framing suggests.
Why the IPO timing matters
Gupta told Reuters that Yotta plans to file draft IPO papers in October, with the listing itself targeted for the January-to-March 2027 window. Data center operators globally are increasingly turning to public markets specifically because GPU costs and compute demand have outpaced what private capital alone can comfortably finance, and Yotta's own executives have framed this raise explicitly around that dynamic rather than around a liquidity event for early backers.
That framing matters for how to read the debt-repayment detail. An IPO that primarily cashes out early investors is one kind of signal. An IPO whose proceeds are earmarked to retire debt taken on to build the infrastructure being valued is a different one: it suggests the private financing stack, however creative, wasn't fully sufficient to carry the buildout on its own, and public markets are being asked to complete a capital structure that started under real time pressure to compete for GPU allocation before it's competitively too late.
What to watch
The clean test here isn't the valuation number, which will keep moving and getting reported inconsistently regardless. It's the draft IPO prospectus itself, expected in October. A prospectus that discloses exactly how much of the $2 billion-plus GPU capex has been financed through debt versus equity versus the GPU-revenue-share vehicles, and exactly what portion of IPO proceeds go to debt repayment versus new capacity, would settle whether this is a growth story asking the public markets to fund the next phase, or a leveraged buildout asking the public markets to help refinance the current one. If that disclosure shows debt repayment is a small, secondary use of proceeds, the current framing holds up fine. If it turns out to be a large share of the raise, that's a materially different story than "India's AI infrastructure champion goes public," and worth a direct follow-up once the DRHP is actually public.



