Today, Artificial Intelligence (AI) has become foundational to a nation’s strength and future, particularly because of its ability to act as a multiplier for economic productivity, modernise governance systems and the military, and solve complex social challenges.
At the recently concluded India AI Impact Summit in New Delhi, Union Minister for Electronics and Information Technology Ashwini Vaishnaw announced India’s plan to add 20,000 GPUs (Graphics Processing Units) to its current 38,000-unit GPU cluster, with a longer-term target of 2 lakh GPUs to stay competitive with the world’s AI leaders.
This indicates how computational power, often measured in GPUs, has become foundational to technological sovereignty in the age of AI. But what is a GPU and how does it work? How has it evolved into an important strategic resource with the arrival of AI? What are the efforts made by India to secure its position in a future driven by Large Language Models (LLMs) and other AI systems? Let us explore.
GPUs: From gaming hardware to a key AI prerequisite
A GPU is a specialised processor (it is a type of application-specific integrated circuit, or ASIC) designed for ‘massive parallelism’. It can process millions of calculations simultaneously rather than sequentially.
This characteristic of GPUs can be traced to its origins in gaming, where rendering millions of pixels requires large-scale concurrent processing. In contrast, the CPU (Central Processing Unit) is a general-purpose processor designed for high-speed sequential logic. In modern systems, the two units work in tandem, with the CPU managing complex system logic and delegating data-intensive, repetitive tasks to the GPU.
Modern AI models, particularly LLMs, rely on GPUs’ massive parallelism to carry out matrix multiplications, which form the foundation of AI neural networks. As GPUs can process these calculations in parallel at a scale impossible for traditional CPUs, they have now become indispensable for training and running AI systems.
Story continues below this ad
GPUs and the geopolitics of AI
In the AI era, computational power has become a key measure of a nation’s strength and future potential. This has created a new geopolitical hierarchy in which access to high-end silicon, used in GPU semiconductors, is as important as access to energy resources.
Currently, the GPU sector is highly concentrated, with over 90 per cent of all high-end GPUs being designed by the US-based tech company Nvidia and manufactured almost entirely in foundries in Taiwan. Notably, Nvidia’s dominance over GPUs in the AI era has made it the most valuable company in the world by market cap. This implies that the global AI ecosystem is potentially vulnerable to any disruptions in production or any policy shift by the US or Taiwan.
Already, the growing strategic value of GPUs has given rise to ‘chip wars’, with the US placing strict controls on the export of advanced GPUs to prevent China and other competitors from attaining equal AI capacity. Several nations, including India, are now attempting to counter this dominance by building sovereign AI, which is the capacity to create AI using its own data, technology, and workforce.
Creating sovereign AI would ensure that a nation’s governance, defence, and economic models are fully under its own jurisdiction. For India, sovereign AI is not only a technological and economic goal, but also a strategic hedge against a ‘compute divide’ created by nations that dominate AI.
Story continues below this ad
Key pillars of IndiaAI Mission
India launched the IndiaAI Mission in March 2024 with an outlay of Rs 10,372 crore for the development of the overall AI ecosystem in the country. It is guided by the vision of “Making AI in India” and “Making AI Work for India”.
The Mission is implemented by IndiaAI, an independent business division under the Ministry of Electronics and Information Technology. IndiaAI Mission targets innovation, startup support, increased data access, and use of AI for public good through its seven-pillared approach:
- IndiaAI Compute Pillar: It provides high-end GPUs to startups and researchers at a highly subsidised rate of Rs 65 per hour, allowing small-scale innovators to train AI models.
- IndiaAI Application Development Initiative: It aims to support the development of AI applications for India-specific challenges in various sectors.
- AIKosh Dataset Platform: It aims at developing large datasets for training AI models, integrating data from government and non-government sources to help developers focus on AI solutions instead of building modules from scratch.
- IndiaAI Foundation Models: It supports the development of Indian LLMs using Indian languages and data to ensure sovereign capability and global competitiveness in generative AI. The recently launched Sarvam AI is one of four startups selected in this programme.
- IndiaAI FutureSkills: This pillar is designed to build AI-skilled professionals by supporting Indians at various stages of their educational journey, from undergraduates to PhD students. It also provides for the creation of AI and data labs across the nation’s educational institutions.
- IndiaAI Startup Financing: This provides funding to AI startups under the IndiaAI Startups Global programme launched in March 2025.
- Safe and Trusted AI: It ensures responsible AI adoption, focusing on machine unlearning, bias mitigation, privacy-preserving ML, explainability, auditing, and governance testing.
Furthermore, India is also investing in building large data centres through partnerships like L&T-Nvidia partnership, leveraging global expertise to ensure sovereign AI.
Challenges in building sovereign AI ecosystem
India faces several hurdles in its pathway to building a sovereign AI ecosystem. Large GPU clusters have a significant environmental footprint. Scaling up to lakhs of GPUs will create greater demand for several utilities, particularly electricity and water. A single rack of high-end GPUs consumes more than 100kW (equivalent to the peak energy demand of 80-100 Indian households), putting immense stress on the grids of cities that house data centres, as datacentres house several GPU racks.
Story continues below this ad
Data centres require enough amounts of water for cooling, particularly those using evaporative cooling systems. In a water-stressed nation like India, this has raised concerns. At the same time, building a sovereign AI system will require huge, multi-year investments with long gestation periods.
The training of AI models also requires datasets of high quality. While India has vast amounts of data, the unstructured nature of data collection and processing, and the fact that a considerable portion of Indian users’ data is processed on foreign servers, present another key challenge.
Moreover, the creation of sovereign AI models requires specialised AI experts with the ability to build these models from scratch. Such high-value experts often migrate to the US or Europe due to significant wage gaps and better research grants. This brain drain is another important hurdle.
Strengthening AI infrastructure to power India’s future
Securing India’s AI future with sovereign AI requires building a holistic chip-to-grid ecosystem to ensure self-sufficiency across the entire chain – from the silicon chips used in GPUs to the national electricity grid.Simultaneous investments across all stages of this chain will help prevent delays and bottlenecks.
Story continues below this ad
India needs data centres that run around the clock on renewable energy. The recent union budget’s measure to extend tax holidays to data centres with optimal PUE (Power Usage Effectiveness) is a step in the right direction.
There is also a need to promote R&D in domestic production of cooling technology that helps cool data centres with minimal water usage. Sustained efforts will be required to train and retain a workforce capable of mastering GPU technology and AI at large.
Most importantly, domestic production of GPUs and associated silicon chip technology would help reduce the dependence on Nvidia and the US. The India Semiconductor Mission (ISM) 2.0, focusing on the design of GPUs optimised for Indian languages and the creation of indigenous IP, if implemented effectively, will bring great benefits to the nation.
To sum up, India’s quest for building a GPU capacity of around two lakh units in the long run is not just about accumulating hardware, but also about strengthening the nation’s AI infrastructure to power its future.
Story continues below this ad
Post read questions
Computational power has become a key determinant of technological sovereignty in the age of Artificial Intelligence. Examine this statement in the context of global competition for GPUs and AI infrastructure.
What are Graphics Processing Units (GPUs)? Explain why they have become indispensable for modern Artificial Intelligence systems.
Discuss the significance of sovereign AI for India. How can it help the country avoid a global ‘compute divide’?
India has announced plans to significantly expand its GPU computing capacity. Examine the opportunities and challenges involved in building large-scale AI compute infrastructure in the country.
Story continues below this ad
Export controls on advanced semiconductor chips have emerged as a new tool of technological geopolitics. Analyse their implications for global AI development.
(Kannan K is a Doctoral candidate at the Centre for Economic and Social Studies, Hyderabad.)
Share your thoughts and ideas on UPSC Special articles withashiya.parveen@indianexpress.com.
Click Hereto read the UPSC Essentials magazine forFebruary 2026.Subscribeto ourUPSC newsletterand stay updated with the news cues from the past week.
Story continues below this ad
Stay updatedwith the latestUPSC articlesby joining ourTelegram channel–IndianExpressUPSC Hub, and follow us onInstagramandX.
