The target and the programme
India wants to develop a homegrown GPU within about 24 months. The target was announced by S. Krishnan, secretary of the Ministry of Electronics and Information Technology (MeitY), in an interview published on 4 November 2026. Classification: STATEMENT — this is a goal declared by the government, not the announcement of a finished chip or a settled production plan.
What is still missing
The effort sits inside India Semiconductor Mission 2.0 (ISM 2.0), the second phase of the national semiconductor programme, approved by the Indian cabinet in July 2026 with roughly 1.27 lakh crore rupees — about US$ 15 billion — over 12 years. The new phase widens the original programme's scope and now explicitly includes AI accelerators, alongside fabs, advanced packaging, equipment, materials, research and talent.
The base already in place
What does not exist yet is what matters most to turn the target into reality: the government has disclosed no manufacturer, supplier, architecture, process node, production capacity or dedicated funding for the GPU. No interim milestones have been published either. The test of the plan will therefore be the release of those details — not the announcement itself.
Why now
India is not starting from scratch. The country already has homegrown processors, such as SHAKTI, developed by IIT Madras on the open RISC-V architecture, and VEGA, from C-DAC, and the government says it wants to commercialise both. The SHAKTI project, started in 2013, already targets a 7-nanometre processor for 2028, as announced in October 2025. India also accounts for about 20% of the global semiconductor design workforce, and five of the twelve projects approved in the mission's first phase have entered commercial production.
The context explains the urgency: the global supply of AI GPUs remains tight and chip prices have risen. On the software side, the IndiaAI Mission released the homegrown models EKA-1, from Soket Labs, with 26 billion parameters, and Dhīr 2B and Dhīr 5B, from ZenteiQ, on 3 November — a story Geekiki has already covered. The difference is that those models still depend on foreign hardware to be trained. The Indian government has also said public procurement of computing will remain technology-neutral, meaning a homegrown chip will have to compete with imported ones on price and performance.
Cover image: a silicon wafer during chip manufacturing. Credit: SEMI.
Sources: Crypto Briefing — India plans indigenous GPU chips within 24 months, says MeitY Secretary; MLQ — India Sets Two-Year Target for Indigenous GPU; The 420 — India Plans to Develop Homegrown GPUs Within Two Years.
Transparency: this article was produced with the help of artificial intelligence, based on public internet sources cited below, and the information is classified as an official statement — a target announced by the government, not a confirmed product. We recommend checking the original sources to confirm the details.
Muse
