Key Terms & Concepts — UPSC Mains
National Supercomputing Mission (NSM)
"A 2015 joint DST-MeitY mission, implemented through C-DAC, building India's PARAM series of supercomputers toward near-complete indigenisation of HPC capacity."
The National Supercomputing Mission (NSM) is a joint initiative of the Department of Science and Technology (DST) and the Ministry of Electronics and Information Technology (MeitY), launched in 2015 and implemented through the Centre for Development of Advanced Computing (C-DAC). Its stated goal is near-complete indigenisation of India's high-performance computing (HPC) capacity by 2030, built around the PARAM series of supercomputers deployed at academic institutions and national laboratories. Traditional HPC, the kind NSM was originally built around, is optimised for large-scale numerical simulation, weather and climate modelling, computational fluid dynamics, molecular simulation and astrophysics, workloads that are CPU-intensive and reward tightly coupled processors solving a single large problem. This differs technically from AI compute, which is overwhelmingly GPU-dense, parallel matrix-operation work suited to training large language models and deep-learning systems. Key NSM milestones include Param Siddhi-AI (commissioned with Nvidia support, ranked 63rd on the global TOP500 list in November 2020) and Param Rudra (dedicated as three systems in Pune, Delhi and Kolkata in September 2024, built on indigenously designed Rudra servers). By 2026, NSM had deployed roughly 40 petaflops across 37 systems nationally against a live target of around 90 petaflops. Param Pragya, inaugurated at IIT Delhi's Sonipat campus in August 2026 and explicitly framed around AI workloads, extends this lineage toward the GPU-dense AI-compute category.
Distinguishes traditional HPC from AI-specific compute, a technical distinction UPSC increasingly rewards, and situates it against the separate IndiaAI Mission.
- 1 Launched 2015, joint DST-MeitY initiative, implemented via C-DAC
- 2 Goal: near-complete HPC indigenisation by 2030
- 3 Builds the PARAM series of supercomputers (Param Siddhi-AI, Param Rudra, Param Pragya)
- 4 Param Siddhi-AI: 5.267 petaflops peak, ranked 63rd on TOP500 (Nov 2020)
- 5 Param Rudra: dedicated Sept 2024 (Pune, Delhi, Kolkata), indigenous Rudra servers
- 6 ~40 petaflops deployed across 37 systems by 2026, target ~90 petaflops
- 7 Distinct from the IndiaAI Mission (2024), which democratises GPU access via a subsidised compute pool
- 8 Param Pragya (IIT Delhi, August 2026) extends NSM toward AI-specific compute
Param Pragya's inauguration was best read as India adding an institutional AI-research node under NSM's lineage, not as a claim to frontier AI compute competitiveness.
