The Mumbai-based company is aiming to secure roughly $500 million to $600 million in capital and then file its formal listing documents within weeks, according to people familiar with the matter. It plans to raise a similar amount in the public offering.
Yotta has positioned itself as a domestic alternative to Western technology giants, as Prime Minister Narendra Modi advocates for his country to become an AI power that can compete with the US and China. Amazon.com Inc. and Alphabet Inc. have said they will each pour more than $100 billion into capital expenditures this year, including AI infrastructure in India.
Yotta is in discussions with several banks to manage the IPO, with contenders including the local arms of Nomura Holdings Inc. and Goldman Sachs Group Inc., along with ICICI Securities Ltd. and Kotak Securities Ltd., the people said, requesting anonymity due to the sensitivity of the talks. The company has already received in-principle approval for the listing and is awaiting final clearance from the Securities and Exchange Board of India.
Among the potential investors in the pre-IPO round are sovereign wealth funds, including Mubadala Investment Co., and several prominent Indian billionaires’ family offices have already signed on. Yotta had targeted $1.2 billion in the pre-IPO fundraising, Reuters reported in February, citing comments from Chief Executive Officer Sunil Gupta.
Yotta didn’t immediately respond to emailed requests for comment.
Companies from AI developers like OpenAI to data center specialists like CoreWeave Inc. and Nebius Group NV are investing record amounts of capital into computing capacity, as they strive for leadership in the emerging field. Just four of the so-called hyperscalers — Amazon, Alphabet, Microsoft Corp. and Meta Platforms Inc. — are planning to spend about $650 billion on capex this year.
At the heart of Yotta’s pitch is its build-out of AI infrastructure in India. The company currently operates about 10,000 of Nvidia’s H100 chips and plans to have thousands of Nvidia B200 units by May. An additional batch of more than 20,000 Nvidia B300 processors are expected to go live by August, part of a $2 billion investment announced last month.
These graphic processing unit chips, or GPUs, represent successive generations of Nvidia’s AI-focused architecture. The H100, or the Hopper, is widely deployed across global data centers, while the newer B200, the Blackwell, has improved performance and energy efficiency. The B300 extends those advances, enabling the training and deployment of extremely large-scale AI models.
Yotta had previously explored a US listing through a SPAC merger, but has since pivoted to its domestic market, where investor appetite for AI and digital infrastructure assets has strengthened. It’s positioned itself as a provider of sovereign computing services — independent from foreign players — and its customers include government bodies and private enterprises.
The timing of the listing is tied to Yotta’s planned spending on Nvidia hardware. CEO Gupta has said he intends to build one of the largest AI accelerator clusters in Asia.