
Nvidia's leadership crisis has caused nervousness among India's data centre companies that deal with neocloud players as questions emerge whether Jensen Huang's successor would continue with the aggressive financing model championed by the ousted co-founder. Doubts have surfaced as the incoming interim lead at Nvidia, local executives told ET, has reportedly opposed Huang's aggressive financing of the AI ecosystem in the past. The uncertainty around Nvidia's leadership transition continues to create ripple effects across the broader AI infrastructure market, with investors reassessing their exposure to companies dependent on Nvidia's technology and financial backing. Greater US scrutiny amid Huang's 'spygate' exit has put a question mark over whether potential shifts in graphics processing unit (GPU) supply chains could disrupt project timelines and investment returns in India.
India's artificial intelligence ambitions are increasingly tied to Nvidia, with the American chipmaker powering a significant share of the country's AI computing infrastructure. According to Greyhound Research, 35,985 of the 44,077 compute units listed on the IndiaAI Compute Portal are Nvidia-based, accounting for roughly 82% of the listed units. Counterpoint Research puts Nvidia's share of India's high-end AI training and inference capacity at an estimated 85-90%. More than 4,000 GPUs work in tandem to train an AI model that is designed for the many languages spoken across India, while Bengaluru-based AI startup Sarvam AI scaled its training infrastructure on Nvidia's H100 GPUs. A Noida-based data centre operator Yotta Data Services has deployed 20,736 Nvidia Blackwell Ultra GPUs at its facility through a project exceeding USD 2 billion in investment, and AM Intelligence has placed firm orders for 29,000 Nvidia Rubin GPUs representing approximately 100 MW of committed AI computing capacity.
Nearly ₹5 billion in capital commitments have been made across India's Nvidia-powered AI infrastructure ecosystem, according to Counterpoint Research estimates. The IndiaAI Mission, approved in March 2024 with an outlay of ₹10,371.92 crore, has onboarded more than 45,000 GPUs through its subsidised computing ecosystem as of September 2026, with 237 projects approved for subsidised compute support and 93.18 lakh GPU hours sanctioned. Approximately 50-60% of the capital is expected to go towards physical infrastructure, including data centres, power, liquid cooling and networking, while the remaining 40-50% will be directed towards AI server racks, GPU silicon, memory and associated components. Based on these estimates, the investment pipeline could translate into approximately 150-200 MW of fully populated, live AI computing capacity over the next two to three years.
The October 7, 2022 U.S. announcement of sweeping restrictions on exports of advanced computing chips to China has created additional challenges for Nvidia's ecosystem, according to Investing.com analysis. Nvidia has supported an ecosystem of both suppliers and customers, with customers particularly neoclouds that rely on Nvidia's backing being the most affected. The departure of Jensen Huang has made it less likely that Huang will remain involved in another capacity, further complicating the company's strategic direction. AMD and custom-chip providers such as Broadcom are natural beneficiaries of this supply chain diversification, while networking is another potential beneficiary that the market underappreciates, as a more diversified mix of AI chips could reduce Nvidia's product lock-in and boost demand for independent networking suppliers.
Despite current challenges, experts maintain a positive long-term outlook for AI growth, according to Livemint analysis. Ives stated that 'the game is going to change for Big Tech. It will slow it down. But there's no overbuild, no bubble concerns', which he views as a positive development. The analysis suggests that there's no overbuild, no bubble concerns in the current AI infrastructure market, indicating sustainable growth potential despite policy changes and corporate disruptions. With this ecosystem push led by Jensen and his team, it is estimated that India enterprise is looking to generate over USD 300 billion by 2030 powering agentic AI applications not only for India but for the globe, according to Counterpoint Research. However, analysts expect Nvidia's share of India's AI compute market to moderate to around 65-75% over the next three to five years even as business continues to grow, as competing platforms like AMD Instinct GPUs, Google TPUs, and AWS Trainium become credible alternatives for inference workloads once models are trained and deployed at scale.