AI Applications
National Supercomputing Mission: Launch Date, Phases and What It Has Actually Built
The National Supercomputing Mission was approved by the Union Cabinet in March 2015 with a budget of around Rs 4,500 crore over seven years, jointly run by DST and MeitY and implemented by C-DAC and IISc Bengaluru. The first system, Param Shivay, went live at IIT BHU in February 2019. This article gives the full timeline, explains why the launch date is often misquoted, lists what the mission has deployed, and connects it to the compute question every AI learner and researcher in India now runs into.
CSEWhy ·
The National Supercomputing Mission was approved by the Union Cabinet in March 2015 with a budget of around Rs 4,500 crore over seven years, jointly run by DST and MeitY and implemented by C-DAC and IISc Bengaluru. The first system, Param Shivay, went live at IIT BHU in February 2019. This article gives the full timeline, explains why the launch date is often misquoted, lists what the mission has deployed, and connects it to the compute question every AI learner and researcher in India now runs into.
The short answer: March 2015
The National Supercomputing Mission (NSM) was approved by the Union Cabinet in March 2015, with an outlay of roughly Rs 4,500 crore spread over seven years. It is run jointly by the Department of Science and Technology and the Ministry of Electronics and Information Technology, and implemented on the ground by C-DAC in Pune and the Indian Institute of Science in Bengaluru. The first supercomputer built under the mission, Param Shivay, was dedicated at IIT (BHU) Varanasi in February 2019.
So if the question is "when was it launched", the honest answer has two dates. The policy launched in 2015. The hardware arrived in 2019.
Why people get the date wrong
Four years between cabinet approval and the first working machine is a long gap, and that gap is the reason you will see 2015, 2016 and 2019 all quoted as the launch year in different places.
The mission was also structured in three phases rather than as a single build. Phase I was about assembling systems in India using imported components. Phase II moved some manufacturing in-country. Phase III targeted indigenous design, which is where C-DAC's Rudra server boards and the Trinetra interconnect come in. Each phase slipped, and the original seven-year window has been extended more than once. A mission approved in 2015 to finish by 2022 was still commissioning new systems in 2024.
That is not unusual for infrastructure of this kind, but it does mean "launch date" is a slightly soft question. Pick the one your context needs: exam answers want 2015.
| Milestone | When | What happened |
|---|---|---|
| Cabinet approval | March 2015 | NSM sanctioned, approx Rs 4,500 crore over seven years, DST and MeitY jointly |
| Implementation setup | 2016 onwards | C-DAC Pune and IISc Bengaluru named implementing agencies |
| First system | February 2019 | Param Shivay, about 833 teraflops, commissioned at IIT (BHU) |
| Expansion across institutes | 2019 to 2022 | Param Shakti, Param Brahma, Param Sanganak, Param Pravega, Param Ganga and others at IITs, IISERs and NITs |
| AI-focused compute | 2020 to 2023 | Param Siddhi-AI entered the global Top500 in late 2020; AIRAWAT at C-DAC Pune ranked 75th in May 2023 |
| Indigenous phase | September 2024 | Three Param Rudra systems dedicated at Pune, Delhi and Kolkata, built on Indian-designed servers |
What the mission has actually delivered
Around thirty systems have been installed across IITs, IISc, IISERs, NITs and national laboratories, with the count varying depending on which official statement you read and when. They are not all large. Many sit in the tens or low hundreds of teraflops and serve a single campus for weather modelling, molecular simulation, genomics or computational fluid dynamics.
The two that matter most for AI are different. Param Siddhi-AI and later AIRAWAT at C-DAC Pune were built around GPU clusters rather than general-purpose CPU nodes, which is what deep learning training actually needs. AIRAWAT's Top500 entry in 2023 was the first time in years India had a system placed that high.
The Param Rudra systems dedicated in September 2024 matter for a different reason. They run on hardware designed in India, which was the whole point of Phase III. Whether that scales is a separate question, but the capability exists now where it did not in 2015.
NSM is not IndiaAI Mission, and the difference is the point
People conflate the two constantly. NSM is a 2015 high performance computing programme aimed at scientific research: climate models, drug discovery, aerospace simulation, seismic analysis. The IndiaAI Mission, approved in March 2024 with an outlay of over Rs 10,000 crore, is a separate thing built specifically around AI, and its compute pillar is about empanelling tens of thousands of GPUs for startups, researchers and students to rent cheaply.
Why does the distinction matter to you? Because if you are training models, NSM's CPU-heavy systems are mostly not what you want, and access usually runs through your institution's research allocation. The IndiaAI compute route is the one designed for AI workloads and for people outside elite labs.
The honest read on nine years of NSM is this. India decided in 2015 that depending entirely on foreign compute was a strategic risk, acted slowly, and ended up with a modest but real base of machines and a small pool of engineers who know how to design and run them. That decision looks a lot smarter in 2024 than it did in 2016.
The gap nobody budgeted for
Hardware is the easy part of this story. Rs 4,500 crore buys machines. It does not buy people who know how to use them well.
Talk to anyone running an HPC facility at an Indian institute and you hear the same complaint: utilisation is uneven. Some clusters run at capacity, others idle because the researchers around them were never trained to parallelise a workload, write a job script or think about where their bottleneck actually is. C-DAC runs HPC training precisely because of this, and the NSM programme funded a training component from the start. It has not been enough.
The same pattern is now repeating with AI compute. GPUs are being subsidised at the national level, and plenty of the people who could use them have never trained a model beyond a Colab notebook. If that describes you, the fix is not reading more policy summaries. It is building things end to end, repeatedly, with someone experienced telling you where you went wrong, which is what the eight-week AI Creator Fellowship is structured around for students and early-career professionals. For colleges that have compute or are planning for it and want their students actually prepared to use it, our AI workshops and certification programmes are built for that institutional gap.
The mission's launch date is a fact worth knowing. What you do with the capability it created is the part that affects your career.
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. Which was India's first supercomputer under NSM?
Param Shivay, commissioned at IIT (BHU) Varanasi in February 2019 with a capacity of about 833 teraflops. It was the first system built and deployed under the mission, nearly four years after cabinet approval.
3. How can students or researchers get access to NSM supercomputers?
Access normally runs through your host institution. If your IIT, IISER, NIT or national lab hosts an NSM system, you apply for a compute allocation through the facility's internal process, usually with a faculty supervisor attached to the project. C-DAC also runs national facilities with application-based access for external researchers.
4. Is the National Supercomputing Mission still running?
Yes. The original seven-year window from 2015 was extended, and new systems including the three Param Rudra supercomputers were dedicated as recently as September 2024. Phase III work on indigenously designed servers and interconnects continues.
5. What is the difference between NSM and the IndiaAI Mission?
NSM, approved in 2015, builds general high performance computing capacity for scientific research across Indian institutions. The IndiaAI Mission, approved in March 2024, is a separate programme focused on artificial intelligence, with a compute pillar that subsidises GPU access for startups, researchers and students.
6. What is Param Rudra?
Param Rudra refers to three supercomputers dedicated in September 2024 at Pune, Delhi and Kolkata, deployed for radio astronomy, nuclear physics and materials science research respectively. They are notable for running on server hardware designed in India under Phase III of NSM.