MeitY Empanels TCS Kyndryl NEC India CoRover Innefu for Government AI Projects From 80 Bidders
India shortlisted six companies from over 80 bidders to deliver AI inside its central government. The list includes a railway chatbot startup, a surveillance intelligence company, and three global IT giants. What the government does with this panel will define whether Indian AI moves from announcements to actual delivery
For a very long time, getting the Indian government to adopt new technology followed a particular rhythm. A committee would be formed. A working group would produce a report. A policy would be announced. A portal would be launched. And then somewhere between the announcement and the actual delivery, the momentum would dissolve into procurement paperwork, inter-departmental coordination, and timelines that stretched across budget cycles. The empanelment that MeitY’s National e-Governance Division announced on June 4, 2026 is a quiet but meaningful attempt to break that rhythm, at least for artificial intelligence.
The Ministry of Electronics and Information Technology, operating through NeGD, its Digital India implementation arm, ran a competitive selection process for companies that would provide AI and machine learning services and personnel across central ministries, state departments, public sector undertakings, and affiliated organisations. More than 80 companies submitted bids. The names included in that pool were not small or obscure: Deloitte, EY, PwC, KPMG, and Fractal Analytics were all in the running. When the process concluded, six companies made the cut. Those six will now form a pre-approved vendor panel that any government body can draw from for AI projects, without running a fresh procurement exercise each time. The empanelment is valid for two years, with a possible extension of one more year.
The six companies are TCS, NEC Corporation India, Kyndryl Solutions, CoRover, Innefu Labs, and Cactus Technology Solutions. Reading that list carefully reveals something about how the government has chosen to think about what AI delivery inside the public sector actually requires. TCS needs no introduction. It is India’s largest IT services company, it has deep institutional relationships with government clients, and it carries the weight of a proven implementation track record across the most demanding public sector programmes the country has run. NEC Corporation India brings a different kind of credibility, a Japanese multinational with a long history in critical infrastructure, identity systems, and public safety technology. Kyndryl, the enterprise infrastructure services company spun out of IBM in 2021, has committed $2.25 billion to India over three years and is already running an AI Innovation Lab in Bengaluru with a specific mandate to build AI capabilities for governance, critical infrastructure, and cyber resilience.
Those three names make intuitive sense in a government vendor list. The other three are the ones worth paying closer attention to. CoRover is a Bengaluru-based startup that built BharatGPT, a conversational AI platform fine-tuned for 22 Indian languages and dialects. If you have ever used the IRCTC website or app to book a railway ticket and been greeted by an AI assistant, you have used CoRover’s technology. BharatGPT now serves over one billion users across IRCTC, government citizen services, banking, and enterprise customer support. It can handle text, voice, and video interactions in 12 Indian languages and is built with data sovereignty as a core design principle, meaning the data stays within India, which matters enormously for government deployments. CoRover quoted the highest rate in the empanelment at Rs 69 lakh, but it brings something none of the larger vendors can fully replicate: a working, production-tested multilingual AI system that has already operated at Indian government scale.
Innefu Labs is a different kind of specialist. It is an Indian cybersecurity and intelligence analytics company that works across defence, national security, and law enforcement, building AI tools for threat detection, open-source intelligence, facial recognition, and surveillance-grade data analysis. It bid the lowest amount in the entire empanelment process at Rs 40.67 lakh, below TCS’s Rs 42.89 lakh and NEC India’s Rs 48.98 lakh. That Innefu is in this panel at all tells you something about the range of AI use cases the government is preparing for, which extends well beyond citizen services and process automation into security, intelligence, and monitoring functions that require a very different kind of vendor.
What these six companies will actually do under the empanelment spans a wide range of work. The categories include AI model development and customisation, implementation of AI systems within existing government workflows, advisory and consulting on AI strategy, capacity building and training for civil servants, and project execution for new AI-driven initiatives. The framework is designed to be modular, meaning a ministry can pick the vendor and the scope appropriate to its specific need, rather than committing to a single large contract with a single vendor for everything at once. That flexibility is one of the empanelment’s more practical design features. It acknowledges that AI adoption in the public sector will not happen through a single master project. It will happen department by department, use case by use case, which means the procurement architecture needs to be able to move at that pace.
The numbers that sit behind the government’s urgency here are significant. The global AI in government market was valued at $22.41 billion in 2024 and is forecast to reach $98.13 billion by 2033, growing at a compound annual rate of nearly 18%. In India specifically, 47% of Indian enterprises reported several generative AI use cases live in 2025, with another 23% in pilot stages. Digital payment fraud alone, one of the most urgent governance problems the country faces, has grown nearly 40 times in value over four years. Government AI is not an aspiration. It is an operational necessity.
The honest question, and it is the one that empanelments alone cannot answer, is whether the procurement reform will translate into real delivery. India has announced AI ambitions before. The difference this time is that the six companies on this panel are not being asked to present ideas. They are being handed a cleared path to implement them inside actual ministries, with pre-agreed terms that remove the biggest single obstacle to AI adoption in Indian governance, the time and cost of procurement itself. Whether the departments that now have access to this panel are ready to use it with the clarity of purpose that delivery requires is an internal question that no vendor list can resolve. The bottleneck in Indian government AI has rarely been the absence of capable companies. It has been the absence of clear problem statements, clean data, and the institutional will to see a project through from pilot to production. Six very different companies are now on standby. The test is what the government asks them to build.
