India Launches NSM 2.0: Integrated HPC, AI, and Quantum Path to Exascale Computing
By R Anil Kumar
Bengaluru, December 17, 2025. India has launched NSM 2.0, marking a pivotal advancement in its National Supercomputing Mission, with the unveiling occurring at the Supercomputing India 2025 event held at MIT–MAHE Bangalore.
This initiative builds on the successes of NSM phase one, which deployed 37 indigenous supercomputers totalling around 39-40 petaflops, powering critical applications from COVID-19 drug screening to astronomical discoveries.
The new phase introduces a unified roadmap converging high-performance computing (HPC), artificial intelligence (AI), and quantum technologies towards exascale capabilities, aiming for strategic autonomy in compute infrastructure.
The launch featured 11 new Memorandums of Understanding (MoUs) signed by the Centre for Development of Advanced Computing (C-DAC) with leading institutions such as the Indian Institute of Astrophysics (IIAP), Indian Institute of Astrophysics (IIA), National Geophysical Research Institute (NGRI), and IIT-Kanpur.
These partnerships focus on co-creation of IT systems, faculty exchanges, and establishing new HPC labs, fostering a national talent grid for next-generation computing. They are expected to add over 60 petaflops of capacity soon, concentrating resources in Bengaluru hubs like C-DAC, IISc, and CSIR-4PI.
Key hardware announcements included VEGA-powered development boards, such as the VSD Squadron Ultra board, leveraging India’s homegrown VEGA RISC-V ecosystem for sovereign compute stacks.
Other highlights encompassed the topsc.cdacb.in platform for supercomputing learning, and a Letter of Intent from Aster Quanta Neuro-systems blending deep tech with neuroscience applications. These elements signal a shift from India as a consumer of global HPC systems to a designer and manufacturer, emphasising indigenous processors, interconnects like Trinetra, and servers like Rudra.
NSM 2.0’s draft proposal targets pre-exascale systems—starting with 100, 200, and 500 petaflop machines from 2027—en route to 2 exaflops total capacity by around 2030, potentially ranking India among the global top four or five supercomputing nations.
The architecture integrates HPC as the foundation for AI ambitions, with quantum acting as an accelerator rather than a standalone solution, supported by unified software and global standards. Efficiency gains will prioritise architectural innovations over shrinking transistors, alongside advancements in cooling like direct liquid cooling for better power usage effectiveness.
For developers, researchers, and system engineers, this establishes a domestic ecosystem tailored for exascale workloads, backed by training over 27,000 professionals already and expanding domain-specific skills in AI, astrophysics, drug discovery, and more.
The mission addresses surging compute demands for large language models, protein folding, weather prediction, and sub-kilometre flood modelling, with Param Siddhi AI—once world’s second in AI supercomputers—exemplifying prior leadership.
Flagged off by IT Minister Ashwini Vaishnaw, Supercomputing India 2025 from 9-13 December underscores execution over theory, with NSM’s Rs. 4,500 crore phase one fully utilised and phase two set for larger funding post-Cabinet approval.
