Artificial intelligence (AI) is no longer a peripheral technology in the information and communications technology ecosystem. Across the sector, AI has moved from experimental deployment to operational necessity, reshaping how operators build, manage and monetise their assets. In India, this shift is accelerating rapidly. Indian telcos are embedding AI across network operations, customer experience and enterprise services, while data centre players are scaling AI-ready compute infrastructure at a pace that few markets can match. Simultaneously, the government’s IndiaAI Mission and the draft National Telecom Policy 2025 are beginning to set the regulatory and investment frameworks needed to sustain this momentum. What is emerging is not a single transformation but a layered one, which is playing out across network infrastructure, enterprise operations, AI-native services and the silicon that powers them all.
A look into the key themes driving the intelligence shift for the sector…
Networks that think for themselves
The most immediate expression of AI in Indian telecommunication is inside the network itself. Operators are deploying AI across planning, operations, maintenance and security, helping them replace manual processes with autonomous, real-time decision-making at a scale that was simply not possible before.
Reliance Jio has arguably gone furthest in this direction. The operator has adopted hyperautomation from the ground up, deploying digital twins that allow engineers to plan network installations without site visits and generative AI tools that enable field staff to query the status of millions of network elements in real time without escalating to a centralised operations centre. Predictive, preventive and corrective maintenance across its network is now handled through AI-driven observability and event correlation, a model that has become essential given the scale of managing a nationwide 5G network.
Bharti Airtel has taken a parallel path, building AI capabilities that span its consumer and enterprise operations. Its AI-driven fraud detection system has flagged over 71 billion spam calls and nearly 3 billion spam messages, blocked more than 800,000 malicious links, and stops approximately 30,000 fraud attempts every day.
Similarly, Vodafone Idea Limited (Vi) has deployed AI-powered self-organising networks across its 4G and 5G infrastructure, alongside an AI energy optimisation module that the company claims reduces annual carbon emissions by the equivalent of planting 13 million trees.
Even Bharat Sanchar Nigam Limited, which continues to focus on 4G expansion, is using AI tools for network management, churn-trend analysis and customer satisfaction monitoring. The operator is also developing in-house small language models for telecom-specific tasks, which is a notable step towards reducing dependence on foreign AI providers.
Operators as AI platforms
Beyond their own networks, Indian telcos are increasingly positioning themselves as AI platforms for enterprises. The shift is significant, as operators that have traditionally sold connectivity are now selling intelligence.
Airtel’s digital arm, Xtelify is the clearest example of this evolution. Launched in August 2025, it offers a sovereign, telco-grade cloud platform that handles 1.4 billion transactions per minute internally and already serves external clients such as Singtel, Globe Telecom and Airtel Africa across 14 countries. Airtel is also building six individual AI agents spanning buying, billing, payments and customer care, with the ambition of transforming its consumer app into a conversational, AI-first experience.
Vi, on the other hand, has signed a five-year agreement for an AI-powered business support system overhaul, targeting intelligence, automation, personalisation and faster product launches. It has also enabled AI-powered conversational payments, positioning itself among the first merchants in India to offer agentic payment capabilities.
Similarly, Reliance Industries has incorporated a dedicated AI subsidiary and committed approximately Rs 7 trillion in investment across Gujarat over five years, with a significant portion directed at AI and digital infrastructure. The strategy spans gigawatt-scale data centres, global technology partnerships and AI services across sectors such as healthcare, education and agriculture.
The infrastructure beneath the intelligence
India’s data centre sector is undergoing a transformation that is directly tied to the demands of AI. The shift from conventional IT workloads to graphics processing unit (GPU)-intensive AI training and inference is rewriting the design, economics and geography of data centre investment across the country.
Airtel’s data centre arm, Nxtra, secured a $1 billion investment in March 2026 from institutional investors, valuing the business at approximately $3.1 billion. The capital will fund expansion from roughly 300 MW to 1 GW of capacity, with Nxtra already operating 14 core data centres and over 120 edge facilities across more than 65 cities. Nxtra also became the first data centre company in India to deploy AI for predictive maintenance and energy optimisation at scale, targeting a 10 per cent increase in asset life, a 15 per cent boost in equipment performance and a 25 per cent improvement in workforce productivity.
Yotta Infrastructure holds an estimated 60-70 per cent of India’s GPU capacity and is building a $2 billion AI hub. In February 2026, it announced the deployment of over 20,000 liquid cooled NVIDIA Blackwell Ultra GPUs, forming one of Asia’s largest AI superclusters, with a road map to 80,000 GPUs by the end of FY 2027. Yotta’s sovereign AI cloud platform, Shakti Cloud, is already hosting BHASHINI, the government’s national multilingual AI platform, which was migrated entirely from foreign hyperscalers to Indian infrastructure in early 2026.
Reliance is reportedly building a 1-3 GW AI data centre in Jamnagar, Gujarat, which could rank among the world’s largest single-site facilities, powered by green energy. Similarly, NTT Data is developing a Rs 105 billion AI data centre cluster in Hyderabad.
Energy consumption remains the central challenge for the sector. AI racks consume five to six times more power than conventional racks, and India’s data centres consumed approximately 0.5 per cent of national electricity in 2025, a figure that is projected to nearly double by 2030. Talent is the other binding constraint, with India needing approximately 500,000 AI professionals by 2027 and currently facing a shortfall of 300,000 semiconductor and data centre engineers.
GenAI and the AI-as-a-service layer
Generative AI (GenAI) is arriving in Indian telecommunication not just as an internal efficiency tool but as a commercial service layer. Operators and data centre players are building platforms that allow enterprises to consume AI the way they consume connectivity, which is on demand, at scale and without building it themselves.
Jio’s AI platform, JioBrain offers large language model (LLM)-as-a-service, speech and text translation, text-to-video generation and code generation via over 500 application programming interfaces, with a plug-in architecture designed for integration with existing enterprise networks. Airtel Cloud offers sovereign, India-hosted AI infrastructure with claimed cost savings of up to 40 per cent compared with global hyperscaler pricing.
Yotta’s Shakti Cloud provides AI workspaces, serverless inferencing and GPU clusters on a pay-per-use basis, and goes a step further by accepting equity stakes from start-ups in lieu of cash for compute access.
On the model side, India is also building its own AI stack. Sarvam AI, selected by the IndiaAI Mission to develop India’s sovereign LLM, released models in early 2026 that outperform global counterparts on Indian-language benchmarks. BharatGen, a government-supported initiative led by a consortium of IITs, unveiled a 17 billion-parameter multilingual model at the India AI Impact Summit 2026. Further, the IndiaAI Mission has onboarded over 38,000 GPUs across empanelled firms, including telcos and data centre operators, making subsidised compute available to start-ups and researchers across the country.
However, execution gaps remain visible. Of the Rs 103.72 billion five-year IndiaAI Mission outlay, only approximately Rs 4 billion had been released in the first two years as of early 2026. An EY survey further found that only 15 per cent of Indian enterprises have GenAI in production and only 8 per cent can measure their AI costs, underlining that the infrastructure investment narrative runs well ahead of enterprise adoption maturity.
The silicon question
Underpinning every dimension of this transformation is the semiconductor layer, and here, India’s position is more complex. The country aims to train 85,000 chip design engineers, contributing roughly 20 per cent of the global semiconductor design workforce and hosts large design centres for companies such as AMD, ARM, Intel and Renesas. Yet, it has, until very recently, had no domestic manufacturing capability for chips of any kind.
This is beginning to change. Micron’s assembly and test facility at Sanand, Gujarat, inaugurated in February 2026, is India’s first operational semiconductor plant, producing DRAM and NAND memory modules at commercial scale. The Tata Electronics-PSMC fab in Dholera is targeting production by 2027 at 28nm-110nm nodes. CG Semi and Kaynes Semicon have both begun operations at OSAT facilities in Sanand, adding further momentum to the domestic manufacturing effort.
For telecommunication and data centre applications, however, all high-performance AI accelerators and telecom chipsets remain entirely imported. India’s 5G networks run on foreign silicon, and its GPU clusters are built on foreign processors. The India Semiconductor Mission 2.0, announced in Union Budget 2026-27, prioritises semiconductor equipment and materials with a Rs 10 billion allocation. It is a step in the right direction, but it remains modest relative to the scale of the ambition. Bridging the gap between chip design strength and manufacturing capability will take time, requiring sustained investment and policy consistency that extends well beyond a single budget cycle.
In sum
What is unfolding in India’s AI and telecommunication landscape is a race on multiple fronts at once. Network intelligence, platform economics, compute infrastructure, sovereign AI models and indigenous silicon are all moving in parallel, and the progress made on each front will, to a large degree, determine how competitive India’s digital economy becomes in the years ahead. The foundations being laid today are substantial. Operators are embedding AI into the core of their businesses rather than layering it on top, and data centre players are investing at a scale and speed that few markets have seen. Those treating AI as core architecture rather than an incremental feature are already pulling ahead, and the competitive gap between those at the frontier and those still catching up is only likely to widen. For India’s telecom and data centre sector, the AI era is not approaching. It has already begun.
