The Compute Paradox: Why India's global AI leadership will be built on constraints, not abundance

The global narrative around Artificial Intelligence is dominated by an obsession with accumulation, more GPUs, more parameters, more data centers. This narrative, driven by the resource-abundant economies of the West and China, perpetuates the illusion that the AI revolution is merely a resource war.

It is not. While compute is the engine of AI, the true differentiator is not the raw power we possess, but how intelligently we deploy it.

India's AI ambition will not be realised by mimicking the Silicon Valley paradigm of centralised, brute-force infrastructure. That model assumes abundance: abundant capital, abundant energy, and abundant access. India operates in a reality of constraints. Recognising this is not a disadvantage; it is our greatest strategic opportunity.

The conversation in India must pivot immediately from chasing access to demanding optimisation. This is the essence of GPU orchestration, the new control layer of intelligence. But this layer must be built with a robust, "India-first," constraint-focused philosophy.

The rise of 'Intelligence per Watt'

The world is hitting a wall. The environmental and economic costs of the current AI boom are unsustainable. Electricity demand from global data centres is projected to more than double by 2030 according to IEA. The era of brute-force AI is ending. The future belongs to efficiency.

For India, this isn’t a future prediction; it's a present necessity. We cannot afford fragmented, underutilised compute capacity. The prevailing approach, importing Western infrastructure models, fails the Indian builder. These systems optimize for performance above all else, often ignoring the economic viability of the solution.

This is the Indian builder’s dilemma: they are forced to use sledgehammers for surgical tasks.

A robust GPU orchestration platform designed for India must reject generalized efficiency and prioritize maximising "Intelligence per Watt." It needs to dynamically allocate resources, manage workloads across diverse environments (from sovereign cloud to the edge), and balance cost, latency, and compliance in real-time.

The "India-First" builder imperative

The IndiaAI Mission's plan to deploy thousands of GPUs is a necessary step. True AI sovereignty means having the control to apply this compute to specific Indian problem statements, optimising agricultural yields with localised data, delivering healthcare diagnostics in low-connectivity environments, and managing financial inclusion across diverse linguistic landscapes.

We need platforms designed for the "constrained builder." This requires an infrastructure layer that understands the trade-offs inherent in the Indian market:

1. The cost of intelligence: For Indian enterprises, infrastructure cost is not a back-end metric; it is a primary determinant of viability. Startups training Indic LLMs or enterprises running sensitive financial models cannot afford opaque billing and unpredictable costs.
2. Data sovereignty and governance: Ensuring sensitive workloads in finance, healthcare, and government are processed within accountable, local environments, adhering strictly to data regulations.
3. Frugal innovation: Enabling breakthroughs without requiring venture capital equivalent to a small nation's GDP.

Constraints as a global competitive advantage

India’s success with Digital Public Infrastructure (DPI), such as UPI and Aadhaar, provides the blueprint. The "India Stack" succeeded because it was designed from the ground up for massive scale, extreme low cost, and open interoperability. It solved an Indian problem so effectively that it became a global standard.

We must apply this same design philosophy, frugal engineering at scale, to AI infrastructure.

When Indian builders are forced to innovate within tight constraints, they develop hyper-efficient models and architectures. By solving the hard problems of deploying AI in a resource-scarce, diverse environment, India is pioneering the future of sustainable AI.

This is our competitive advantage. As the rest of the world grapples with the unsustainable economics and energy demands of current AI models, the solutions forged in India, optimized, efficient, and constraint-aware, will become highly desirable globally.

If you can make AI work efficiently and affordably in India, you can make it work anywhere.

The moment of design

Every technological cycle reaches a moment when design choices become destiny. For AI infrastructure, that moment is now.

India has the opportunity to bypass the pitfalls of the acquisition arms race and focus directly on building a distributed, sovereign compute fabric unified by an intelligent orchestration layer.

This is not just about reducing dependence on imported technology. It is about designing a system that aligns with the specific economic and social goals of the nation. If we design our platforms to prioritize the Indian builder, focusing on optimisation and embracing our constraints, we will not just be participants in the AI revolution. We will be the architects of its next, more sustainable phase.

Karan Kirpalani, Chief Product Officer, Neysa