MeitY Secretary sets a 24-month target for an Indian GPU and plans to commercialise SHAKTI and VEGA processors
MeitY Secretary S. Krishnan says India aims to build its own GPU within about 24 months under ISM 2.0 and to commercialise the SHAKTI and VEGA RISC-V processors. No builder or chip specifications have been named yet.
India wants its own graphics processing unit, the chip at the heart of the AI boom, and it now has a timeline for it.
In an interview on November 4, 2026, S. Krishnan, Secretary of the Ministry of Electronics and Information Technology (MeitY), said India plans to develop an indigenous GPU under the second phase of the India Semiconductor Mission (ISM 2.0) and that the effort will take about 24 months. He also said the government wants to take India's home-grown SHAKTI and VEGA processors, both based on the open RISC-V architecture, to commercial markets.
The statement is ambitious, and so far it has few details.
What was said
According to reports of the interview:
- The GPU target: Krishnan said the indigenous GPU project would take approximately 24 months and sits under ISM 2.0.
- A target, not a launch date: He presented the timeline as a development target, not a firm commercial launch date. He did not name a manufacturer, chip supplier, design house or builder, and no architecture, performance figures or process node have been published.
- CPUs as well: Separately from the GPU, India wants to commercialise SHAKTI and VEGA. These are CPU families, not graphics chips, and they are distinct from the GPU effort.
- The Y2K analogy: Krishnan described the AI boom as a possible repeat of the Y2K era, when demand for software remediation helped launch India's IT services industry onto the global stage.
ISM 2.0, also called Semicon 2.0, was approved by the Union Cabinet with a fiscal outlay of ₹1,27,500 crore, according to MeitY. Two days after the interview, on November 6, Electronics and IT Minister Ashwini Vaishnaw launched the application portal through which companies can apply for support under the programme.
Why a GPU, and why now
GPUs have become the main currency of AI computing. Training and running large language models needs huge numbers of these parallel processors, and the market is dominated by a handful of mostly American companies. Reports on Krishnan's remarks note that global supplies of AI GPUs have been tight and prices have risen.
For India, the dependence is concrete. The IndiaAI Mission's shared compute pool, which subsidises access for Indian start-ups and researchers, is built on imported accelerators. India's sovereign foundation-model efforts, from Sarvam AI to the models released under IndiaAI this month, run on the same foreign hardware. A domestic GPU, even one well behind the frontier, would give India some leverage over supply, pricing and security, especially for government and defence workloads.
SHAKTI and VEGA: the foundation India already has
India is not starting from nothing. The SHAKTI programme, run by the Reconfigurable Intelligent Systems Engineering (RISE) group at IIT Madras, has spent years building open-source RISC-V processors with support from MeitY's Digital India RISC-V initiative.
VEGA is C-DAC's family of RISC-V processors, developed under the same national push. RISC-V is an open instruction set, so Indian designers can build processors without licensing proprietary CPU architectures. That matters for a country that wants control over the chips in its strategic systems.
Commercialising these processors means moving them from demonstration chips into products that companies actually design in, such as microcontrollers, embedded controllers, networking gear and defence electronics. That requires reliable foundry access, software toolchains, developer ecosystems and customers willing to commit. All of these have historically been weaker in India than chip-design talent.
How realistic is 24 months?
Building a competitive GPU is one of the hardest problems in semiconductors. It involves the silicon itself and, just as important, a software stack, drivers and programming frameworks that AI developers will actually use. China's domestic GPU makers have shown that progress is possible with heavy state support and long timelines, but also that matching leading products is a multi-year effort.
Several things will show whether India's target is credible:
- Who builds it. No company or institution has been named. Whether the project goes to an IIT-led team, a public lab like C-DAC, private design houses, or a consortium will shape its feasibility.
- What "GPU" means. A first-generation chip for inference, edge AI or graphics would be far more achievable than a data-centre training accelerator.
- Where it is made. The process node and foundry have yet to be specified in the interview.
- Software. Without a usable programming stack, even good silicon will not get used.
What it means for India's chip ambitions
The announcement shows how ISM 2.0's priorities are shifting. The first phase of the India Semiconductor Mission focused on bringing in fabs and assembly-and-test plants, several of which are now in commercial production. The second phase is putting more weight on design, intellectual property and products, which is where most of the value in the chip industry sits.
A credible Indian GPU programme, paired with commercially deployed SHAKTI and VEGA processors, would move India from building other people's chips toward owning chip designs, and that is the larger prize in Krishnan's Y2K analogy. For now, the target is a statement of intent. The real test will be the first named partners, budgets and tape-out milestones.
Sources
- https://www.tomshardware.com/tech-industry/semiconductors/india-targets-creating-homegrown-gpus-within-2-years-and-plans-to-commercialize-indigenous-shakti-and-vega-risc-v-processors-sees-ai-boom-as-a-potential-repeat-of-the-y2k-opportunity-for-its-it-industry
- https://cio.eletsonline.com/artificial-intelligence/meity-indigenous-gpu-chip/76737/
- https://cryptobriefing.com/india-indigenous-gpu-chips-24-months/
- https://www.tokenpost.com/news/technology/48699
- https://www.indianweb2.com/2026/11/india-approves-12-new-electronics.html
- https://en.wikipedia.com/wiki/SHAKTI_(microprocessor)
- https://www.freepressjournal.in/education/iit-madras-isro-develop-indigenous-microprocessor-for-space-applications
- https://government.economictimes.indiatimes.com/news/digital-india/india-plans-indigenous-gpu-chips-in-24-months-says-meity-secretary-s-krishnan-on-ai-strategy/135508818
- https://www.pib.gov.in/PressReleseDetailm.aspx?PRID=2333716&lang=2®=48