Edge data centres, anticipated to expand India’s cloud services market into smaller cities three years ago, have remained an unrealised ambition as applications requiring high-speed connectivity have not developed as forecasted.
Indian AI startups, particularly those involved in training complex AI models, require large-scale computational power and storage capacities. These demands are met more effectively by hyperscale data centres in metropolitan hubs like Mumbai and Chennai, rather than smaller edge data centres in tier 2 and 3 cities.
Edge Computing in India
Brijesh Patel, founder & CTO of SNDK Corp, an IT solutions and support services company, told AIM that emerging trends in AI, such as Gen AI and reinforcement learning, require substantial computing resources and low-latency processing. In India, this transition is leading to a heightened demand for edge computing infrastructure. Unlike conventional cloud configurations, edge computing positions processing nearer to the data source, facilitating quicker and more effective real-time decision-making.
Experts see reduced latency as the primary benefit of edge computing, especially for gaming, autonomous vehicles and automated traffic management systems. Edge computing faces challenges due to variability in processing and data retrieval, which can negate latency advantages. Increasing computing resources to address this variability contradicts the aim of smaller data centres, a source familiar with the matter said.
Moreover, applications must be tailored for edge computing, which adds to the developers’ workload. Currently, edge computing is also more expensive than traditional data computing at scale, and its success requires the entire ecosystem to mature alongside it, the source said.
India is seeing massive investments in data centre capacity to meet AI demand, which is forecast to reach about $100 billion by 2027, with major expansions in Mumbai, Chennai, and Noida by global and domestic players.
Additionally, AI startups usually operate in metropolitan cities where hyperscale data centres provide reliable and cost-effective infrastructure. Consequently, there is less incentive to invest in edge data centres in smaller cities, which face challenges like higher costs and limited talent.
“For most startups, the simplicity and flexibility of the cloud outweigh the complexity of distributed edge infrastructure. Edge holds promise, but cloud economics and maturity still make it a niche choice in India,” said Vaibhav Poonekar, CTO at Decimal Point Analytics.
Startups also rely on hyperscale providers for large-scale model training, while edge computing is suitable only for latency-critical or niche inference workloads. The demand for burst compute and extensive storage capabilities gives hyperscale elasticity a significant advantage, he explained.
Digital Connexion’s market analysis highlights the industry shift: India is now pivoting toward hyperscale data centres in response to rising AI workloads and increasingly sophisticated backbone networks. Edge facilities have become less central, as the expanded capacity and improved latency of hyperscale data centres, enabled by network upgrades, are fulfilling demands once expected of the edge.
Edge data centres offer low-latency and localised processing, but they lack the scalability of hyperscale cloud providers like AWS and Azure, which are essential for AI training. However, Patel asserts that edge data centres excel in real-time data processing for localised operations, such as autonomous systems and smart devices.
Scale and Cost Efficiency
Dheeraj Chaudhary, director of technology at Verge Cloud, told AIM, “The digital edge infrastructure in India isn’t saturated yet; it’s simply delayed by fundamentals.”
Hyperscale cloud providers’ scale and cost efficiency create challenges for smaller edge data centres, particularly in cost-sensitive Indian industries. AWS and Azure offer more scalable and cost-effective models, delivering better returns on investment, Patel highlighted.
Decentralised AI-driven automation in manufacturing and logistics reduces dependence on distant data centres. “With the push for faster responses and more intelligent processing, edge data centres are becoming a strategic necessity, supporting critical AI applications in industries like retail, healthcare, and smart cities,” he added.
Edge data centres currently generate limited revenue from AI startups. The demand for these facilities is primarily internal, driven by large companies such as Bharti Airtel, rather than coming from independent AI startups.
Chaudhary said, “Until monetisation drivers such as low-latency SaaS, AI workloads, and immersive applications mature, enterprises will remain reliant on centralised clouds.”
Nonetheless, Poonekar believes that Cloud meets immediate needs with elastic growth, while edge remains more tactical than strategic. In India, most startups still view the cloud as the scalable long-term path.
Future of edge computing
Chaudhary and Poonekar both noted that high capital costs and uncertain returns hinder the adoption of edge deployments in tier-2 and tier-3 cities, where latency and connectivity are paramount. Conversely, Tier-1 cities require compute-heavy AI workloads that are best served by hyperscale data centres. They emphasise that edge may still find relevance even as hyperscale continues to dominate at scale.
Companies with large data centres have also announced expansion plans, with Princeton Data announcing plans to establish 230MW of net capacity in India by the end of next year. Equinix also revealed a major expansion last year. Yotta’s Greater Noida facility, announced in October 2022, can expand to 250MW by next year. However, there are no specific plans for edge data centre expansion.
Chaudhary also believes that the future lies not in a competition between edge and cloud computing, but in their orchestrated convergence. As latency-sensitive applications in payment processing, over-the-top (OTT) services, and AI inference become more prevalent, the demands for compliance and the need to include underserved areas of India will make ultra-local compute and storage essential.
“This is where VergeCloud is positioning itself, building smaller city Points of Presence (PoPs) that extend beyond Content Delivery Networks (CDNs) to deliver true edge capabilities,” he said. This approach also aims to bridge connectivity gaps while aligning with India’s Digital Personal Data Protection (DPDP) Act and facilitating the next phase of digital growth.
The limited scale of latency-sensitive AI applications and operational challenges in tier 2 and 3 cities have further weakened the correlation between AI startups and edge data centres in India.
The post Edge Data Centres Take Back Seat as Indian AI Startups Flock to Hyperscale Hubs appeared first on Analytics India Magazine.
