AI Applications
What Is the Aim of the National Supercomputing Mission?
A direct answer to what the National Supercomputing Mission set out to do: build a national grid of supercomputers, move India from assembling machines to designing its own processors, and train the people who can use them. Includes the three-phase structure, what has actually been deployed since 2019, and how NSM differs from the IndiaAI Mission's GPU push.
CSEWhy ·
A direct answer to what the National Supercomputing Mission set out to do: build a national grid of supercomputers, move India from assembling machines to designing its own processors, and train the people who can use them. Includes the three-phase structure, what has actually been deployed since 2019, and how NSM differs from the IndiaAI Mission's GPU push.
The aim, stated plainly
The National Supercomputing Mission was set up to give India its own high performance computing capability instead of renting or importing it. Approved in 2015 with an outlay of around Rs 4,500 crore, it is run jointly by the Department of Science and Technology and the Ministry of Electronics and IT, with C-DAC Pune and IISc Bengaluru doing the building. The original target was a grid of more than 70 supercomputers installed at academic and research institutions across the country and linked over the National Knowledge Network.
That is the headline. Underneath it sit four distinct objectives, and they matter because the mission is usually judged only on the first one.
- Install supercomputing capacity at Indian institutions so researchers do not queue for foreign compute time
- Move from importing and assembling systems to designing indigenous processors, servers and interconnects
- Build applications in areas with national stakes: weather and climate, drug discovery, genomics, seismic imaging, flood forecasting, materials science
- Train roughly 20,000 people in HPC, because machines without skilled users are expensive furniture
Why it was deliberately split into three phases
The mission was never meant to just buy supercomputers. The three phase design is the actual strategy, and it reads like an import substitution plan written by engineers rather than economists. Each phase raises the share of the machine that is designed and made in India.
Phase 3 is where the real test sits. C-DAC's AUM processor, an ARM based high core count server chip, and the Trinetra interconnect are the pieces that decide whether India ends up with its own HPC stack or a well stocked collection of imported racks.
| Phase | What India does | Representative output |
|---|---|---|
| Phase 1 | Assemble imported subsystems in India | PARAM Shivay at IIT BHU, 2019, 837 teraflops |
| Phase 2 | Manufacture key components domestically | Rudra servers, Trinetra interconnect, systems at IIT Kharagpur, IIT Kanpur, IISER Pune and others |
| Phase 3 | Design the processor and system end to end | AUM processor development, PARAM Rudra systems deployed in 2024 |
What has actually been built
Roughly thirty systems have been commissioned under NSM since 2019, spread across IITs, IISERs, NITs, C-DAC centres and agencies like the Indian Institute of Tropical Meteorology. The headline machines are worth knowing by name. AIRAWAT, C-DAC's AI focused system in Pune, entered the Top500 list at rank 75 in June 2023, which remains India's best placing from this programme. In September 2024 three PARAM Rudra supercomputers were commissioned at Pune, Delhi and Kolkata at a combined cost of about Rs 130 crore, built substantially on indigenous hardware.
The original seven year timeline slipped. Seventy plus systems did not happen by 2022, and the mission has run well past its first deadline. That is a real gap, not a rounding error, and it is the fair criticism of NSM. The counter argument is also real: the share of Indian designed hardware inside each new machine has risen steadily, and a weather model running on an Indian system in Pune no longer depends on someone else's export policy.
The objective almost nobody quotes: 20,000 trained people
Read the mission documents and you will find a human capital target sitting right alongside the hardware target. NSM set up nodal centres for HPC and AI training at institutions including IIT Kharagpur, IIT Madras, IIT Goa, IIT Palakkad, IIIT Hyderabad and JNCASR, precisely because the bottleneck in Indian research computing was never only silicon. It was people who could parallelise code, size a job correctly, and think in terms of compute budgets.
That bottleneck has now moved outward from research labs into ordinary workplaces. Compute is cheap and abundant compared to 2015. What is scarce is someone in a marketing team, a hospital admin office or a 40 person manufacturing firm who can look at a messy process and see where a model or an automation actually fits. If you are a student or early career professional reading about NSM because you want to work near this kind of technology, the honest next step is not studying HPC architecture, it is building things with AI until you have a portfolio, which is what the AI Creator Fellowship is structured around over eight weeks. Institutions that want their own students trained on this inside the campus usually start with AI workshops and certifications before committing to a full lab.
NSM and the IndiaAI Mission are not the same thing
People conflate the two constantly. NSM is about high performance computing, the kind that runs fluid dynamics, climate models and molecular simulations, and it dates to 2015. The IndiaAI Mission, approved in 2024, is about GPU capacity for AI training and inference, with a target of more than 10,000 GPUs made available to startups and researchers through empanelled providers. Different workloads, different hardware economics, different ministries driving the push.
They do overlap at the edges. AIRAWAT was built as an AI system inside NSM, and C-DAC sits in both stories. The useful way to hold it in your head: NSM asked whether India can build its own computers, and IndiaAI asks whether Indian builders can get enough compute today. The first question is answered slowly, in fabs and chip design teams. The second is being answered right now, and it is the one that will decide what you can build this year.
FAQs
1. Who implements the National Supercomputing Mission?
It is steered jointly by the Department of Science and Technology and the Ministry of Electronics and Information Technology. Implementation is handled by the Centre for Development of Advanced Computing (C-DAC) in Pune and the Indian Institute of Science, Bengaluru.
2. How many supercomputers has NSM installed so far?
Around thirty systems have been commissioned since 2019 at IITs, IISERs, NITs, C-DAC centres and research agencies, against an original target of more than seventy. The most recent additions are the three PARAM Rudra systems at Pune, Delhi and Kolkata, commissioned in September 2024.
3. What is India's fastest supercomputer?
AIRAWAT, the AI oriented system at C-DAC Pune, is India's highest ranked entry in recent Top500 lists, debuting at rank 75 in June 2023. Other large systems include Pratyush and Mihir, operated by the Ministry of Earth Sciences for weather and climate modelling.
4. What is the AUM processor?
AUM is an indigenous ARM based server processor being developed by C-DAC under Phase 3 of the National Supercomputing Mission. It represents the mission's furthest ambition: Indian designed silicon at the heart of Indian supercomputers rather than imported CPUs.
5. Is the National Supercomputing Mission related to the IndiaAI Mission?
They are separate programmes with some overlap. NSM, from 2015, builds high performance computing capacity and indigenous HPC hardware, while the IndiaAI Mission, approved in 2024, focuses on making GPU compute available for AI training and inference.
6. What research actually runs on NSM supercomputers?
Typical workloads include numerical weather prediction and monsoon modelling, computational drug discovery (used heavily during Covid), genomics, seismic imaging for energy exploration, flood forecasting and materials simulation. Access is mostly through the host institution's allocation process.