IndiaAI Mission tender seeks 10,000 more GPUs; technical bids open October 30

The Centre for Development of Advanced Computing (C-DAC), which works under the Ministry of Electronics and Information Technology (MeitY), has invited AI service providers to supply up to 10,000 AI compute units for the IndiaAI Mission, with technical proposals due to be opened online on October 30, 2026. The tender follows Union Electronics and IT Minister Ashwini Vaishnaw's announcement on October 11, 2026 that the government would add 10,000 graphics processing units to the mission's shared pool. According to reports of the announcement, 3,000 of the new units will be kept for government departments and public-interest work, and 7,000 will be offered to companies building AI.

Building and running modern AI, from chatbots to Indian-language speech tools, depends on specialised chips. GPUs were designed for graphics but turned out to be very good at the many small calculations AI needs; TPUs are chips built specifically for AI. They are expensive and in short supply worldwide, which puts them out of reach for most Indian start-ups, students and researchers.

The IndiaAI Mission's answer has been to pay private data-centre companies to make this hardware available on a shared government platform, where approved users can rent computing time at subsidised rates instead of buying chips. Each new round adds capacity. This tender matters because it sets aside a large block, 3,000 units, specifically for government and public-interest uses such as health, agriculture or public services, while giving the bulk to companies building products and models.

The bid conditions show what the government wants: hardware physically located in India and switched on quickly. The requirement to deploy at least 512 units within six months, backed by a Rs 10 crore guarantee, is meant to filter out bidders who cannot deliver. The offer of usage guarantees of up to five years gives providers more certainty that the capacity they install will be used, which can lower the prices they quote.

The tender also comes as India's own foundation models, the large base models on which AI applications are built, move from the shortlist stage towards release. Twelve teams have been shortlisted under the mission to build such models, and their training needs are among the biggest demands on the shared pool. Alongside the hardware, the promised consultation paper on AI regulation will shape the rules these models and their users must follow.

Short-term: Data-centre and cloud companies have until the bid process closes to decide whether to offer capacity on these terms. Once technical bids are opened on October 30, C-DAC will evaluate them before price bids are considered.

Long-term: If the round succeeds, India's shared public AI computing pool will grow well past its current size, making it cheaper for start-ups, universities and government bodies to train and run AI models in the country rather than abroad.

Who is affected: Indian AI start-ups and researchers who rent computing time, data-centre and cloud providers that may bid, government departments planning AI projects, and the teams building India's foundation models.

Questions and Answers