India's National Supercomputing Mission Reaches 40 Systems and 68 Petaflops
India’s National Supercomputing Mission now has 40 systems and 68 petaflops of capacity, according to October reporting, compared with the official August 2025 baseline of 37 systems and 40 petaflops. Domestic servers and software remain central to the programme.
India's public supercomputing network has increased its computing power in about a year. An update released in early October, ahead of the government's ESTIC 2026 science conclave, says the National Supercomputing Mission (NSM) has now deployed 40 supercomputers with a combined capacity of 68 petaflops. In the government's previous update, released in August 2025, the mission had 37 systems totalling about 40 petaflops.
One petaflop is a quadrillion floating-point operations per second. Sixty-eight of them, spread across institutes nationwide, is still small next to the largest machines in the United States, China or Japan. It does, however, make NSM the main source of shared high-performance computing (HPC) for Indian academia and government research.
What the network looks like
According to Insights on India’s account of the latest figures, the 40 systems fall into three tiers:
- 13 high-end systems above 1 petaflop each.
- 12 mid-range systems between 500 teraflops and 1 petaflop.
The source does not give a verified capacity breakdown for the remaining systems.
That spread is deliberate. NSM was designed to put computing capacity in many IITs, IISERs, national labs and universities, so that researchers outside a few elite centres can run climate models, molecular simulations, materials discovery and, increasingly, AI training.
Built in India, increasingly
The mission was launched in 2015 with an outlay of about ₹4,500 crore. It is run jointly by the Ministry of Electronics and IT (MeitY) and the Department of Science and Technology (DST), and implemented by C-DAC, Pune and IISc, Bengaluru. The early phases relied heavily on imported systems that were assembled in India. Later phases have moved more of the stack onshore.
The clearest example is PARAM Rudra. These systems are built on C-DAC's indigenous Rudra server boards and run an Indian system-software stack. Coverage of the latest update says more than 6,000 Rudra server units are now active. C-DAC has transferred the board designs to domestic manufacturing partners. Some PARAM Rudra installations also use Trinetra, an indigenous high-speed interconnect that provides 100–200 Gbps links between nodes. Servers and interconnects are where much of the strategic value lies, because they are the parts most exposed to export controls and supply shocks.
The gap that remains
More installed capacity does not by itself establish the availability or utilisation of every machine. The August 2025 official release recorded 37 systems and 40 petaflops; the October 2026 IANS report supplies the latest 40-system, 68-petaflop figures. The mission’s current end date and next expansion timetable require a fresh official update.
Why it matters now
Demand for computing in India is shifting. Weather and climate forecasting, drug discovery and engineering simulation have always needed HPC. Training and fine-tuning AI models has now joined them, and the IndiaAI Mission is separately subsidising tens of thousands of GPUs through private cloud providers. NSM's role differs from IndiaAI's. It provides publicly owned, research-grade machines and builds the domestic ability to design them.
That ability is the long-term payoff. A country that can design its own server boards, interconnects and system software can upgrade or expand its computing capacity without depending entirely on foreign vendors. For India's semiconductor and electronics ambitions, Rudra and Trinetra are also early proof that Indian-designed data-centre hardware can be built and deployed in numbers.
Watch for official announcements on the mission’s next phase, funding and coordination with other public compute programmes.