AI for Automation
Back to AI News
2026-03-20AI infrastructureYottaIndia AIGPUIPONVIDIA

India's AI chip king just filed for a $4 billion IPO

Yotta Data Services controls 70% of India's GPU capacity and is deploying 20,000+ NVIDIA Blackwell chips. Now it wants a $4B valuation on the stock market.


A company most people outside India have never heard of is about to become one of the biggest AI infrastructure plays on the public market. Yotta Data Services, a Mumbai-based startup that controls an estimated 60–70% of all GPU computing power in India, is preparing to go public at a $4 billion valuation — and the numbers behind it are staggering.

Yotta hyperscale data center facility in India

From real estate empire to AI superpower

Yotta was founded in 2019 by Darshan Hiranandani, heir to one of India's largest real estate dynasties (the Hiranandani Group), and Sunil Gupta, known in the industry as the "Data Centre Man of India." Together, they made a bet: India would need massive local computing power for the AI era, and whoever built it first would own the market.

That bet is paying off. Yotta currently runs 10,000+ NVIDIA GPUs across facilities in Navi Mumbai and Greater Noida. Its Navi Mumbai campus alone spans 820,000 square feet — the size of 14 football fields — housing 7,200 server racks powered by 50 megawatts of electricity.

The numbers at a glance:

$4 billion — target IPO valuation
$500–600 million — capital being raised pre-IPO
60–70% — Yotta's share of India's total GPU capacity
20,736 — liquid-cooled NVIDIA Blackwell Ultra GPUs arriving by August 2026
$2 billion+ — total investment in the Blackwell supercluster
80,000+ GPUs — planned capacity by FY27

A $2 billion NVIDIA supercluster — one of Asia's largest

The headline move: Yotta is building one of the largest AI computing clusters in Asia. By August 2026, it will deploy 20,736 liquid-cooled NVIDIA Blackwell Ultra GPUs (the latest generation of AI chips) connected by ultra-fast 800 Gbps networking. The total investment exceeds $2 billion.

Yotta and NVIDIA partnership for AI infrastructure in India

This cluster will be powerful enough to train trillion-parameter AI models (the kind of models behind ChatGPT and Claude) and handle millions of simultaneous AI queries. For context, training a model like GPT-5 requires exactly this scale of hardware — and Yotta is bringing it to Indian soil.

NVIDIA is so invested in this partnership that Jensen Huang, NVIDIA's CEO, personally endorsed the deployment: "India is emerging as one of the world's most important AI markets. Yotta's deployment creates advanced infrastructure supporting India's sovereign AI ambition."

Why India is building its own AI backbone

India has 1.4 billion people, the world's largest tech workforce, and a government that's going all-in on AI. Prime Minister Modi's IndiaAI Mission has allocated 10,000+ NVIDIA B300 GPUs specifically for developing Indian-made AI models — and Yotta is the company hosting them.

The strategic logic is simple: if Indian companies, researchers, and government agencies have to send their data to American or Chinese servers to use AI, India loses both money and control. By building frontier-scale computing locally, Yotta lets India train AI models on its own terms.

Sunil Gupta, Yotta's CEO, put it plainly: "India's AI ambition requires sustained, high-performance compute at scale with open models enabling developers to build sovereign applications."

The bigger picture: GPU scarcity is global

The race for AI chips (GPUs) is one of the defining resource battles of the decade. The U.S., China, Saudi Arabia, and the UAE are all pouring billions into securing GPU supply. India was a late entrant — but Yotta's aggressive buildout is changing that. The company plans to scale beyond one million GPUs within 3–5 years, with its Navi Mumbai campus alone scalable to 2 gigawatts of power (enough to run a small city).

Yotta NM1 hyperscale data center in Navi Mumbai, India

Who's backing the IPO

According to Bloomberg, Yotta is in discussions with Goldman Sachs, Nomura, ICICI Securities, and Kotak Securities to manage the offering. The company plans to raise $500–600 million in a pre-IPO round and then a similar amount when it lists publicly. It has already received in-principle approval from India's securities regulator (SEBI).

NVIDIA itself is deeply embedded through a $1 billion+, four-year DGX Cloud contract — effectively making Yotta one of NVIDIA's most important infrastructure partners in Asia.

Who should pay attention

AI developers and startups: If you're building AI products for the Indian market (or want cheaper GPU access outside the U.S.), Yotta's cloud platform offers direct access to NVIDIA's latest hardware — potentially at more competitive rates than AWS or Azure.
Investors: This IPO could be one of the first pure-play AI infrastructure companies to go public outside of China and the U.S. At $4 billion, it's a fraction of the valuation of companies like CoreWeave ($35B+), which could make it an interesting entry point.
Anyone watching the AI race: The fact that a single Indian company now controls the majority of the country's AI computing power — and is scaling to compete globally — signals that the AI infrastructure war is no longer just a U.S.–China story.

The road ahead

Yotta's IPO filing is expected within weeks. If successful, it would be the largest AI-focused IPO in India's history and one of the few globally that's purely focused on GPU infrastructure. The company's trajectory — from a real estate scion's side bet in 2019 to a $4 billion AI infrastructure empire in 2026 — mirrors the explosive growth of the AI industry itself.

The question now isn't whether India needs this infrastructure. It's whether Yotta can scale fast enough to meet the demand.

Related ContentGet Started with Easy Claude Code | Free Learning Guides | More AI News

Stay updated on AI news

Simple explanations of the latest AI developments