From Quick Wins to Lasting Power
Not long ago, venture capital in India’s AI space was a race to back the next viral app. Think AI-powered photo editors, chatbots, and marketing tools. These 'application-layer' startups were attractive because they promised quick user adoption and a clear
path to market. However, the landscape is undergoing a seismic shift. While funding for AI startups has surged, with over $1.4 billion raised by September 2026, a growing portion is being channelled away from surface-level apps and towards the pickaxes and shovels of the AI gold rush: the infrastructure. This includes everything from data centres and specialised chips to the foundational models that power all AI applications. Investors are realising that true, sustainable value lies not just in using AI, but in owning the machinery that makes it possible.
The Need for 'Desi' AI
A major driver of this shift is the dawning realisation that Western-built AI models don’t fully grasp the complexities of India. With 22 official languages and countless dialects, cultural nuances, and regional contexts, models trained primarily on English-language data often fall short. This creates a massive opportunity for homegrown solutions. Startups like Sarvam AI, selected by the IndiaAI Mission, are building large language models (LLMs) from the ground up, trained on diverse Indian languages and datasets. Their mission is to create 'sovereign AI'—systems that understand and respond to the unique needs of Indian users, businesses, and governance. This isn't just about better translation; it's about creating AI for healthcare, finance, and education that is deeply embedded in the Indian context.
The Soaring Cost of Intelligence
Building and running advanced AI is incredibly expensive and energy-intensive. The lifeblood of modern AI is the Graphics Processing Unit (GPU), a specialised chip essential for training large models. Access to this computing power is a significant bottleneck for innovation. Recognizing this, a new class of startups and investors is focusing on the 'compute layer'. This involves building massive AI-ready data centres, developing more efficient processing units, and creating platforms that offer affordable access to GPU power. As one industry expert noted, compute capacity is now a strategic asset, determining how much AI activity a country can support. This move from being a consumer of foreign cloud services to a builder of sovereign compute infrastructure is seen as critical for India's technological autonomy.
Government as a Strategic Catalyst
The Indian government is actively fuelling this infrastructure-first approach. Initiatives like the IndiaAI Mission, with an initial outlay of over ₹10,300 crore, are designed to build a robust domestic AI ecosystem. A key part of this mission is creating a shared public compute infrastructure, making tens of thousands of GPUs available to startups, researchers, and academic institutions at subsidised rates. Furthermore, the government is exploring a massive ₹20,000 crore fund specifically for frontier AI and compute infrastructure, designed to provide the long-term risk capital that these capital-intensive projects need. This strategic push, combined with a planned $25 billion fund for the broader 'Deep Tech' ecosystem, sends a clear signal: India is serious about building foundational technology and reducing its reliance on foreign hardware and platforms.
A Maturing Ecosystem
Ultimately, the shift from apps to infrastructure reflects a new level of confidence and ambition within India's tech scene. The initial wave of AI was about application, proving what was possible with existing tools. This new wave is about creation—building the core engines, the data pipelines, and the physical hardware that will power the next decade of innovation. Investors are moving from making many small bets on apps to fewer, larger, and more strategic bets on infrastructure platforms. This transition is essential for moving up the global value chain, from being a nation that adopts technology to one that shapes and defines it. The focus is no longer just on creating the next app; it is on building the very foundation upon which all future apps will stand.
