India's Growth in Supercomputing and the NSM

#GS-3 #Economy #Infrastructure #Science & Technology #Artificial Intelligence #ICT #National

Key takeaways

As of September 2026, India has deployed 40 supercomputers with a combined capacity of 68 PF under the National Supercomputing Mission (NSM).
The NSM was launched in April 2015 with a total budget of ₹4,500 crore to achieve compute sovereignty.
India has developed indigenous hardware like Rudra servers and high-speed interconnects of 100 Gbps and 200 Gbps.
Despite progress, a huge gap exists as the US Frontier system alone (1,200 PF) is about 18 times more powerful than India's entire network.
The mission has supported over 16,000 researchers and helped in critical tasks like flood prediction for the Mahanadi River basin.

By September 2026, India successfully deployed 40 supercomputers with a total capacity of 68 PetaFLOPS (PF).
These systems were set up under the National Supercomputing Mission (NSM).
The government is now focusing on creating local servers, system software, high-speed connections, and cooling systems.

Supercomputers use a huge number of processors or nodes working together to solve complex problems much faster than normal computers.
They are often called parallel computers because many CPUs can work on one calculation at the same time.
Computing power is measured in FLOPS (Floating Point Operations Per Second).
1 teraFLOPS (TF) means one trillion operations per second, while 1 petaFLOPS (PF) means one quadrillion operations per second.
The fastest supercomputer in the world reached about 2.19 exaflops in 2026, which is 2.19 quintillion calculations per second.
These machines are vital for climate research, healthcare, space studies, agriculture, and engineering.
They help in predicting crop yields, tracking disease outbreaks, discovering new drugs, and forecasting floods.
India produces nearly 20% of the world's data, making high computing power essential for Artificial Intelligence, weather forecasting, and space research.
The government is providing this infrastructure to research institutes to support Aatmanirbhar Bharat.

The National Supercomputing Mission (NSM) started in April 2015 with a budget of ₹4,500 crore.
The Department of Science and Technology (DST) and the Ministry of Electronics and Information Technology (MeitY) lead this mission.
The main goal is to make India self-reliant in supercomputing and boost research and innovation.
Two main bodies implement the mission: C-DAC, Pune and the Indian Institute of Science (IISc), Bengaluru.
The mission uses a three-phase build approach: first assembly, then manufacturing, and finally full design and manufacturing support.

The mission aims to set up 50 supercomputers with a combined capacity of over 123 PF across Indian institutions.
India's journey started with PARAM 8000 in 1991, developed by C-DAC, which had a speed of 1 gigaflop.
Later systems like PARAM Yuva (54 teraflops) helped in weather forecasting and fluid dynamics.
As of September 2026, the 40 deployed systems include 13 high-end systems (above 1 PF), 12 mid-range systems (500 TF to 1 PF), and 15 systems below 500 TF.
The mission supports 11 SDGs by helping with forest-fire management, climate modelling, and industry collaboration.

In medicine, it helps in drug discovery. During COVID-19, it screened drugs and predicted side effects like cardiac risks.
For cities, it uses weather and air-pollution models to predict heavy rain and smog, helping in disaster planning.
In energy, it maps underground geological structures for oil and gas exploration, matching the quality of commercial tools.
The Early Warning System predicts floods up to two days in advance, specifically helping the Mahanadi River basin.
The Forest Fire Spread Model uses satellites and computing to predict fire movement, tested in the Sikkim Himalayas.
In chemistry and materials science, it simulates atoms and alloys to study their properties.

C-DAC developed the Rudra servers, which handle AI and complex simulations. These are now made by Indian Electronics Manufacturing Services (EMS) partners.
PARAM Rudra supercomputers use these local servers and a domestic software stack for research in astronomy and physics.
By September 2026, 6,000 Rudra servers were deployed, with 1,500 more being made.
India created its own high-speed interconnect networks with speeds of 100 Gbps (Trinetra-A) and 200 Gbps (Trinetra-B).
Local cooling technology is now being deployed to make systems more power-efficient.
A full High-Performance Computing (HPC) system software stack was developed for better management.
PARAM Shavak is a compact supercomputing-in-a-box for students and researchers in colleges.
Local applications are now used by the India Meteorological Department (IMD), Central Water Commission (CWC), Central Pollution Control Board (CPCB), and Ministry of AYUSH.

Over 16,000 researchers and 2,900 PhD scholars from 400+ institutions have used the NSM infrastructure.
The systems have run over 1.5 crore compute jobs and led to 1,990 research publications.
Workshops and hackathons teach students about High-Performance Computing (HPC), Deep Learning (DL), and Generative AI.
Faculty from non-CS backgrounds get training in AI and Machine Learning through AICTE collaboration.
Learning is available through the EduHPC Workshop and the SWAYAM platform under NPTEL.
The HPC Shiksha Portal provides lectures and materials for practical learning.
The NSM Users Forum allows experts to share knowledge and solve technical problems.
The National Knowledge Network (NKN) acts as the high-speed backbone connecting all these facilities across India.

India still imports the core silicon, such as CPUs and GPUs, from foreign vendors like NVIDIA and Intel.
There is a huge gap in capacity. The US Frontier supercomputer alone has 1,200 PF, which is about 18 times India's total capacity.
Private companies in India rarely use NSM nodes, often renting foreign clouds for tasks like virtual crash tests.
High heat in India makes cooling these machines expensive and energy-intensive.
Many Indian researchers are locked into foreign software like NVIDIA's CUDA, making it hard to switch to local architectures like RISC-V.

India should use the India Semiconductor Mission (ISM) to make its own multi-core processors, like the Shakti and Vega chips.
The supercomputing goals should align with the IndiaAI Mission to build GPU clusters for training Large Language Models.
New exascale systems should be powered by solar and renewable energy to keep power usage efficient.
The government should provide subsidized compute access through NKN for deep-tech and biotech startups.

India has come a long way from PARAM 8000 in 1991 to the PARAM Rudra ecosystem.
True technological independence now depends on making domestic silicon chips and reaching exascale computing levels.
This progress is key to achieving the goals of Viksit Bharat 2047.

PetaFLOPS (PF): A measure of computing speed equal to one quadrillion floating-point operations per second.
Exascale Computing: The highest tier of computing, performing quintillions of calculations per second.
PARAM Series: The family of indigenous supercomputers developed by C-DAC.
National Knowledge Network (NKN): The high-speed network connecting India's research and academic institutions.