India Sets 24-Month Deadline to Build Its First Homegrown GPU

India's Ministry of Electronics and Information Technology (MeitY) has set a roughly 24-month timeline to build the country's first indigenous GPU, according to Tom's Hardware, citing comments from MeitY secretary S. Krishnan. The announcement lands alongside plans to commercialize India's existing SHAKTI and VEGA RISC-V processors and comes as a new fab facility takes shape in Gujarat under a partnership between ASML and Tata Electronics.

What Happened

In a November 4 interview reported by Tom's Hardware, Krishnan said the GPU effort falls under India Semiconductor Mission 2.0 and will not happen overnight. He described the project as requiring approximately 24 months and outlined a five-layer stack the government wants to support domestically: energy, accelerator chips, computing infrastructure and servers, and AI models and applications. Krishnan did not name a manufacturer, chip supplier, or builder for the GPU project, leaving the most consequential detail, who will actually fabricate it, unresolved.

The manufacturing side of India's ambitions is further along than the GPU design itself. Tom's Hardware reported that ASML is partnering with Tata Electronics to equip India's first commercial semiconductor fabrication facility in Dholera, Gujarat. The plant is designed around 300mm wafers and is expected to eventually reach a capacity of 50,000 wafers per month. Intel has also announced a strategic partnership with Tata Group covering semiconductor manufacturing and packaging for the Indian market, according to the same report.

On the processor side, India's homegrown SHAKTI and VEGA chips, both based on the RISC-V instruction set architecture, are moving from development boards toward practical deployment. Tom's Hardware noted that at the September 26 SEMICOM India event, the government highlighted an autonomous water-cleaning robot built around a SHAKTI chip and a digital telescope project and a separate forest-monitoring project, both using VEGA processors. Indian companies including VVDN, Kaynes, and Avalon are already manufacturing servers, per Tom's Hardware., and the Centre for Development of Advanced Computing has developed its Rudra server range before transferring the technology to local manufacturers. MeitY's Design-Linked Incentive program has assisted 24 companies that subsequently raised venture capital or other private investment, working on products such as microcontrollers, drone flight controllers, and smart meter chips, per Tom's Hardware.

Krishnan also framed the AI buildout as a potential repeat of the Y2K-era opportunity that built India's IT services sector, which the report notes employs about 6 million people. A NITI Aayog paper cited by Tom's Hardware estimates roughly two million existing jobs could become redundant as AI adoption reshapes the industry, while around four million new jobs could emerge. India's semiconductor market is currently valued at about $50 billion and is projected to approach $100 billion by 2030, according to the India Electronics and Semiconductor Association figures cited in the report.

Why It Matters

MachineHours analysis: the gap between India's GPU ambition and its fab reality is the real story here. A 50,000-wafer-per-month, 300mm facility using ASML tooling is a credible step toward domestic chip capacity, but it says nothing about whether that fab will ever run advanced logic nodes capable of producing a competitive GPU design. SHAKTI and VEGA remain CPU architectures; VEGA's demonstrated use cases are embedded/IoT-focused, while SHAKTI's design classes per the source range from embedded and microcontroller applications to workstation and server workloads — neither is built for graphics or AI accelerator workloads., so the 24-month GPU target is effectively starting from a different base entirely. This pattern, governments announcing accelerator ambitions while actual manufacturing capacity lags years behind, is consistent with the broader semiconductor market concentration and bubble risk building around AI infrastructure spending globally. With Nvidia's market cap sitting at $4.90 trillion and the SOX index still elevated despite a recent pullback, the economic incentive for other nations to attempt entry is obvious, even if the manufacturing runway to compete is not.

Further Reading and Context

India's effort to build a domestic chip ecosystem touches on fundamentals that predate the current AI cycle. For background on how modern fabrication advances translate into usable compute, see MachineHours' explainer on TSMC's 2nm node roadmap, which outlines the process generations separating leading-edge foundries from newer entrants. The underlying device physics are covered in The Transistor: The Invention That Built the Modern World, and the architectural logic that governs how any new processor, including SHAKTI or VEGA, must organize memory and instruction flow is detailed in Von Neumann Architecture. For how GPU-style accelerators stack up against alternative designs like TPUs, see MachineHours' analysis Google TPU vs Nvidia GPU: AI Accelerators Compared (2026).

What to Watch

The most telling signal will be whether MeitY names a manufacturing partner or process node target for the GPU project in the coming months. Krishnan's 24-month figure is only credible if paired with a fab commitment, and the Dholera facility's 300mm, 50,000-wafer-per-month capacity is not yet confirmed to support anything beyond legacy-node production. Watch also for whether the Intel-Tata packaging partnership extends into accelerator-class chips, and whether India's Design-Linked Incentive program begins funding GPU-specific startups rather than the microcontroller and sensor projects it has backed to date.