Shifting Focus for AI Hardware Incentives

India's Ministry of Electronics and IT (MeitY) is reviewing its Production-Linked Incentive (PLI) scheme for IT hardware. This review addresses the changing economics of AI servers, which are different from standard computer hardware. The current scheme, launched in 2023 with a ₹17,000 crore budget over six years, was not designed for the unique structure and cost breakdown of AI servers. Sushil Pal, Joint Secretary at MeitY, noted that AI servers rely heavily on high-end chips and GPUs. These components make up most of the cost, leading to questions about what the incentive model truly encourages. Although AI servers are part of the IT hardware PLI, their specific makeup requires a closer look to ensure incentives lead to real domestic value creation.

The GPU Challenge in Value Creation

The main challenge for India's AI hardware sector is that the most value is tied up in specialized, expensive parts. Pal said GPUs or silicon currently make up about 70% of an AI server's cost, a figure expected to reach 90-92% with newer models. This reliance on imported, expensive parts raises a key question for India's turnover-based incentive scheme: how much value can be created in India if the main costs are from abroad? Globally, GPU servers dominate the AI market for training and running AI models. Leading chipmakers like NVIDIA, AMD, and Intel supply the core technology. Large cloud providers, such as Google and Amazon, are also developing their own custom chips (ASICs) that could offer cost benefits and change the market. India's IT hardware PLI aims to boost local manufacturing. However, a focus on total sales might encourage simple assembly instead of manufacturing key high-value chips locally. Previous electronics PLI schemes, like the one for mobile phones, saw large production jumps (146%). But analysis suggests moving beyond assembly to localize more components is needed to capture greater value.

Production Challenges: Energy and Competition

Despite ambitions to become a global AI hardware manufacturing hub, India faces significant structural challenges. Energy supply is a major hurdle. AI servers and data centers require vast amounts of power, putting a strain on current grids and possibly limiting growth. India's data center capacity is expected to grow rapidly, but expansion might outpace power generation and transmission, leading to a supply gap. While India has developed its electronics manufacturing sector, the success of PLI schemes has had mixed results. Some analyses show that foreign companies have often done better than domestic ones under PLI schemes, and while production has risen, the actual local value added needs careful review. Globally, countries like China are pushing hard for domestic AI chip capabilities. They offer incentives such as power subsidies for data centers using local chips and provide significant state funding for chipmakers. This global competition, along with supply chain risks and high import costs, highlights the need for the PLI scheme to encourage genuine technological advancement and value creation in India, not just production volume.

Future Strategy for AI Hardware

The government aims to build its own AI capabilities through initiatives like the India Semiconductor Mission and the upcoming Semiconductor Mission 2.0, which focuses on domestic AI chip design. Policy is also shifting towards helping researchers and startups access computing power directly, rather than subsidizing data centers, to ensure competitive AI compute pricing. While India has achieved success in electronics manufacturing, revising the IT hardware PLI scheme shows a more advanced strategy to develop a high-value AI hardware sector. This balances manufacturing goals with the complex economics of advanced technology and infrastructure limits.