In a significant move toward technological self-reliance, Union Minister Ashwini Vaishnaw recently announced that India aims to develop its own high-end graphics processing units (GPUs) within the next three to five years. This declaration, made while launching several initiatives under the IndiaAI Mission, signals India’s determination to establish sovereignty in the critical domains of artificial intelligence and semiconductor manufacturing.

But what exactly are GPUs, and why do they matter so much for India’s technological future?
GPUs are specialised electronic circuits originally designed to accelerate the processing of images and videos. Unlike CPUs, which handle general computing tasks sequentially, GPUs excel at parallel processing — performing multiple calculations simultaneously. While they began as tools for rendering high-quality graphics in gaming, GPUs have evolved into essential components for AI, machine learning, scientific simulations, and autonomous vehicles.

In the 21st century, GPUs hold the same strategic importance that oil and steel held during the industrial age. Nations controlling their own supply of advanced chips — the United States, China, and Taiwan — maintain a significant geopolitical advantage. For India, relying on imported GPUs from companies like NVIDIA or AMD creates vulnerability to supply chain disruptions, trade restrictions, or even sanctions.

The global GPU landscape is dominated by a handful of players: American companies NVIDIA and AMD lead in design, while Taiwan’s TSMC and South Korea’s Samsung handle most of the manufacturing. China is racing to develop its own capabilities through companies like Huawei, despite facing US sanctions.
Where does India stand in this technological race? Currently, we are consumers, not creators. Our semiconductor industry remains nascent, focused primarily on design with limited manufacturing capabilities. This dependency represents a strategic liability that India, with its growing global aspirations, can ill afford. The ongoing US-China tech war has demonstrated how export controls on advanced chips can severely hamper a nation’s technological progress. The AI diffusion categorisation introduced recently by the US — albeit by the previous Biden administration — had placed India in a second category to receive technology in a restrictive form.

Building indigenous GPU capabilities will require a multifaceted approach. First, India needs a robust R&D ecosystem populated with world-class engineers and scientists skilled in computer architecture and semiconductor physics. While India produces talented professionals from premier institutions like the IITs, many seek opportunities abroad due to limited prospects at home. A “National GPU Mission” under MeitY could provide the necessary focus, similar to how Taiwan nurtured TSMC through government-backed research.

Second, India needs semiconductor fabrication plants capable of producing advanced chips. Modern GPUs require cutting-edge fabs that typically cost billions of dollars to establish. A pragmatic approach might involve starting with older process nodes that remain effective for less demanding applications, gradually scaling up as capabilities improve. Collaboration with global leaders like TSMC or Intel could expedite this journey.

Third, a skilled workforce is essential. Although India produces numerous engineering graduates annually, few possess specialised training in semiconductor design. Bain and Company recently reported that India will require a 2.3-million-strong workforce in AI by 2027 and there will be a shortfall of 1.2 million. Educational curricula must be revamped to include specialised courses, with industry collaboration providing hands-on experience. Programmes like IIT Madras’s M.Tech in Microelectronics are steps in the right direction, but we need to scale up training dramatically.

Financial backing represents another significant hurdle. The costs associated with GPU design run into hundreds of crores, while establishing a fab requires thousands of crores. Private sector giants like Tata, Reliance, Adani and HCL could contribute substantially with government support through subsidies and public-private partnerships. India’s Semiconductor Mission has allocated Rs 76,000 crore, but more targeted investment in indigenous innovation is necessary.

A robust supply chain for essential materials — ultra-pure silicon, rare earth metals, and specialised chemicals — is equally critical. Currently, India relies heavily on imports for these components. Developing domestic sources or securing long-term international agreements must be prioritised.

Finally, a focused policy framework with clear timelines and milestones would signal strong commitment to this initiative. Tax holidays for chipmakers, export incentives, and protection from cheap imports could help create a level playing field for domestic producers.

The path ahead isn’t easy. India currently lags decades behind established players, and the global market is fiercely competitive. Initial GPUs produced in India may not match the performance of leading models, potentially generating scepticism within the tech community.

Yet there are reasons for optimism. India possesses a vast domestic market with potential early adopters in gaming, AI startups, and defence. The success of ISRO demonstrates that India can excel in high-tech fields with proper focus and investment. Moreover, current geopolitical dynamics may favour India, as Western nations could support its tech ambitions to counterbalance China’s growing influence.
The critical question remains: As India’s tech talent develops new innovations, will intellectual property continue flowing abroad, or can the country retain and nurture this expertise within its own economy? The answer will determine whether India emerges as a true tech superpower or remains primarily a consumer in the global semiconductor race.

The first steps have been taken. The AIKosha platform with its 300 datasets and the AI Compute Portal promising access to thousands of GPUs represent promising foundations. Now the hard work of building a self-reliant GPU ecosystem must begin in earnest.
Link: https://indianexpress.com/article/opinion/columns/in-ai-race-what-india-needs-to-do-to-acquire-indigenous-gpu-capabilities-9880877/