🗞️ Why in News Prime Minister Narendra Modi inaugurated “Param Pragya,” an AI-powered high-performance computing facility at IIT Delhi’s Sonipat campus, while addressing the institute’s 57th Convocation on August 8, 2026, where he also conferred degrees on over 3,000 graduating students, including 587 PhD scholars.
Two Different Machines Wearing One Name
The word “supercomputer” covers two technically distinct things, and the distinction matters for placing Param Pragya correctly. Traditional high-performance computing (HPC), the kind India’s National Supercomputing Mission (NSM) was built around in 2015, is optimised for large-scale numerical simulation: weather and climate modelling, computational fluid dynamics, molecular simulation, astrophysics. These workloads are CPU-intensive and reward tightly coupled processors working on a single large problem.
AI compute is a different technical requirement. Training and running large language models and other deep-learning systems is overwhelmingly GPU-dense work, parallel matrix operations at massive scale, and the infrastructure that supports it looks different from a classical HPC cluster even when both get called a “supercomputer” in public reporting. Param Pragya is explicitly described in official coverage as an AI-powered facility intended to advance IIT Delhi’s capabilities in AI, data science, advanced computing and interdisciplinary research, which places it closer to the GPU-dense AI-compute category than to classical scientific-simulation HPC, though public reporting has not disclosed its specific GPU count or petaflop rating.
Where Param Pragya Sits in the PARAM Lineage
The National Supercomputing Mission, a joint initiative of the Department of Science and Technology and the Ministry of Electronics and Information Technology (MeitY), implemented through the Centre for Development of Advanced Computing (C-DAC), has built out the PARAM series of machines since 2015 with a stated goal of near-complete indigenisation of high-performance computing by 2030. Two prior systems illustrate the trajectory. Param Siddhi-AI, commissioned with Nvidia’s support, delivered 5.267 petaflops of peak (Rpeak) and 4.6 petaflops of sustained (Rmax) performance, ranking 63rd globally on the TOP500 list of the world’s most powerful non-distributed computing systems in November 2020. Param Rudra, built under NSM’s Phase 3 using indigenously designed Rudra servers, was dedicated by the Prime Minister as three separate systems in Pune, Delhi and Kolkata in September 2024, part of a mission whose cumulative compute target has grown well past its original 2015-era 64-petaflop benchmark, with NSM reporting roughly 40 petaflops already deployed across 37 systems nationally and a live target of around 90 petaflops. IIT Delhi has separately been an NSM node, and Param Pragya’s naming places it in this same lineage, though it is sited at the institute’s Sonipat campus specifically and framed around AI workloads rather than general scientific HPC.
NSM and IndiaAI Are Not the Same Programme
It is easy to conflate Param Pragya with India’s other major compute initiative, but the two are administratively distinct. The IndiaAI Mission, approved by the Union Cabinet on March 7, 2024 with an outlay of Rs 10,371.92 crore, is built around a “Making AI in India and Making AI Work for India” mandate and operates through seven pillars: subsidised compute access, foundation-model development, startup financing, datasets, applications, safe and trusted AI, and future skills. Its compute pillar has scaled from an initial target of 10,000 GPUs to over 38,000 GPUs already deployed, with 20,000 more in the pipeline, offered to researchers and startups at a subsidised rate of roughly Rs 65 per GPU-hour.
NSM, by contrast, is the older programme (2015), MeitY and DST-run, delivered through C-DAC, and historically centred on HPC infrastructure at academic and research institutions rather than commercially accessible GPU rental. Param Pragya, as an IIT Delhi facility inaugurated by the Prime Minister with NSM-lineage naming, reads as an NSM-track institutional facility rather than a new IndiaAI Mission compute node, though official coverage has not stated the funding source or explicit programme linkage in detail. The two programmes are complementary rather than duplicative: NSM builds institutional compute capacity embedded in universities and national labs, while IndiaAI’s compute pillar democratises GPU access for a wider pool of startups and researchers who do not have their own infrastructure.
Why “Sovereign Compute” Is the Underlying Policy Goal
Both programmes are answers to the same strategic question: training and running large AI models requires large-scale, sustained GPU access, and a country without domestic compute capacity depends on foreign cloud providers for a capability that increasingly touches data sovereignty, research independence and, in defence and governance applications, national security. India’s framing of sovereignty in this space typically rests on three controllable layers, compute physically located in-country and under domestic governance, data kept resident under Indian law, and models trained on Indian languages, context and data. Param Pragya, an in-country facility at a public institution, and the IndiaAI compute pool, GPUs domestically deployed but centrally allocated, both build toward the first of those three layers.
The Argument
The case for this being a substantive step. Institutional AI compute embedded inside a top research university, rather than only accessible through a centrally allocated pool, gives IIT Delhi’s faculty and 587 newly graduated PhD scholars sustained infrastructure for AI and data-science research without competing for time on a shared national resource. Combined with the IndiaAI Mission’s rapid scale-up from 10,000 to 38,000-plus GPUs in roughly two years, the trajectory shows real capacity being added, not just announced, across both programmes.
The counter to engage. The scale gap between India’s public AI compute and what leading global AI labs operate is not a rounding error. Frontier AI training runs at major labs now routinely draw on compute measured in the hundreds of thousands of GPUs; India’s entire IndiaAI compute pool, at roughly 38,000-58,000 GPUs once the pipeline lands, plus institutional facilities like Param Pragya of undisclosed but almost certainly smaller scale, is not positioned to compete at that frontier. What this capacity realistically supports is applied AI research, model fine-tuning, domestic-language AI work and public-sector use cases, a legitimate and valuable goal, but a different one from frontier foundation-model training, and coverage of Param Pragya’s inauguration has not claimed otherwise.
Balanced verdict. Param Pragya is best read as India adding a node to its domestic AI-research infrastructure rather than a claim to frontier AI compute competitiveness. Its real significance for UPSC purposes is as an illustration of how India’s compute strategy is being built in layers, NSM-track institutional HPC/AI facilities and the IndiaAI Mission’s centrally allocated GPU pool, both aimed at the same underlying goal of sovereign compute capacity, at a scale appropriate to applied research and public-sector needs rather than frontier-model competition.
UPSC Relevance
GS Paper 3: Achievements of Indians in science and technology; indigenisation of technology; developments and their applications in everyday life; infrastructure for science and technology; awareness in the fields of IT and computers.
Prelims focus: National Supercomputing Mission (launched 2015, DST + MeitY, implemented via C-DAC), the PARAM series (Param Siddhi-AI, Param Rudra, Param Pragya), the IndiaAI Mission (approved March 7, 2024, Rs 10,371.92 crore outlay, seven pillars), and the distinction between NSM and IndiaAI Mission as separate but complementary programmes.
Mains angle: “India’s AI compute strategy runs on two administratively distinct tracks, the National Supercomputing Mission and the IndiaAI Mission.” Distinguish the technical difference between traditional high-performance computing and AI-specific compute infrastructure, and assess whether India’s current sovereign compute capacity is better understood as building foundational domestic capability or as a step toward frontier AI competitiveness.
📌 Facts Corner, Knowledgepedia
Param Pragya:
- Inaugurated by PM Narendra Modi at IIT Delhi’s Sonipat campus, August 8, 2026
- Occasion: IIT Delhi’s 57th Convocation
- Degrees conferred on 3,000+ graduating students, including 587 PhD scholars
- Medals awarded: President’s Gold Medal, Director’s Gold Medal, Shankar Dayal Sharma Gold Medal, Perfect Ten Gold Medals
- Described officially as an AI-powered high-performance computing facility for AI, data science and advanced computing research
National Supercomputing Mission (NSM):
- Launched 2015, joint initiative of Department of Science and Technology (DST) and Ministry of Electronics and IT (MeitY), implemented via C-DAC
- Goal: near-complete indigenisation of HPC capacity by 2030; original 2015-era cumulative target was 64 petaflops, since revised to approx. 90 petaflops (approx. 40 petaflops already deployed across 37 systems)
- Param Siddhi-AI: 5.267 petaflops peak, 4.6 petaflops sustained, ranked 63rd on TOP500 (Nov 2020)
- Param Rudra: dedicated as three systems (Pune, Delhi, Kolkata) in September 2024, built on indigenous Rudra servers under NSM Phase 3
IndiaAI Mission:
- Approved by Union Cabinet: March 7, 2024
- Outlay: Rs 10,371.92 crore
- Compute pillar: from initial target of 10,000 GPUs to 38,000+ GPUs deployed, 20,000 more in pipeline
- Subsidised compute rate: approximately Rs 65 per GPU-hour
- Seven pillars include compute, foundation models, startup financing, datasets, applications, safe AI, future skills
Other Relevant Facts:
- C-DAC: Centre for Development of Advanced Computing, the implementing agency for NSM under MeitY
- Sovereignty in AI compute policy is typically framed around three layers: compute (in-country GPUs), data (domestic residency), and models (trained on Indian languages/context)
Sources: PIB, IIT Delhi, IndiaAI
Source: Param Pragya and the Difference Between a Supercomputer and an AI Machine — Ujiyari.com | Free UPSC & State PCS Current Affairs
