India Sets 24-Month Target to Develop Homegrown GPU Chips

India has set a 24-month target for homegrown graphics processing unit (GPU) chips as its semiconductor program expands to cover AI accelerators amid tight global supply of the processors.

MeitY Secretary S. Krishnan tied the goal to India Semiconductor Mission 2.0, an updated effort to build the country’s semiconductor ecosystem. The Cabinet approved the mission in July 2026, with about $15 billion allocated over 12 years.

The updated program includes AI-specific accelerators and supporting semiconductor infrastructure. The effort is part of a broader push on sovereign AI that also covers data protection, domestic AI models and opportunities for India’s IT industry.

Global supplies of AI GPUs have been tight, and chip prices have climbed. India already has indigenous processors, including SHAKTI and VEGA, that the government aims to commercialize. The country accounts for about 20% of the global semiconductor design workforce.

Five of the 12 projects approved under the first India Semiconductor Mission had entered commercial production by late 2026. The new GPU target adds an AI hardware goal to the government’s semiconductor agenda.

Three indigenous AI models were revealed under the IndiaAI Mission on Nov. 3. They include EKA-1, a 26 billion-parameter model from Soket Labs, and Dhīr variants.

The government has said it will take a technology-agnostic approach to procuring AI computing resources. That means homegrown chips will need to compete for use as India acquires computing capacity. Krishnan’s stated timeline for the GPU effort is 24 months.