‘AI Can Take India From a $5 Trillion to a $20 Trillion Economy’
By Pallavi Chakravorty ·
Artificial intelligence is no longer just another workload running on the cloud—it is reshaping the very economics of computing, from energy grids and semiconductor design to enterprise software and agriculture. As the summit conversation moved from job disruption to responsible AI, Rajiv Kumar, Managing Director and President at Microsoft India Development Center, outlined how three forces—the shift from CPU to GPU computing, the rise of AI agents, and India’s policy push—could redefine the next decade of technology.
AI, he argues, is transforming the technology stack “right from energy to the app layer”, describing a systemic shift rather than an incremental upgrade. The cloud is no longer CPU-bound. It is GPU-bound—and that changes everything. “It’s moving from a traditional cloud to an AI cloud.”
Large language models have transformed graphics processors, once designed primarily for gaming, into the backbone of global AI infrastructure. Companies like Microsoft are racing to expand GPU capacity on Microsoft Azure, yet demand still exceeds supply.
“No matter how much we do, it’s not enough.”
But scaling is not simply a matter of plugging in more GPUs. AI workloads require a comprehensive rethinking of the full stack—encompassing data lakes, networking, storage, cooling systems, and sustainable energy. A single AI supercomputer today can require hundreds of thousands of GPUs, a scale few anticipated even three years ago.
“People could not have anticipated that just one company would need hundreds of thousands of GPUs. Multiply that by 50 companies—and the world is behind,” Kumar says.
India’s Leapfrog Moment
Global shortages, he believes, also create opportunity, particularly for India.
Drawing parallels with India’s telecom revolution—where the country moved directly from limited landlines to widespread mobile adoption—he sees AI as a similar inflexion point.
“We will jump-frog from old CPU creation to GPU creation. This is not a 25-year head start race. Everybody is near the beginning.”
With semiconductor incentives, deep talent pools, and policy support, India could move from being a consumer of compute into a designer of it. Microsoft is simultaneously building its own AI chips, the Maia series, while working with global foundries. It is part of the company’s broader strategy spanning silicon, infrastructure, models, and applications.
Yet, he stresses that the AI economy is not defined by chips alone. “It’s about sustainability.” Microsoft plans 100% renewable energy for new data centres in India, net-zero water replenishment by 2030, and net-zero carbon by 2050.
The AI cloud, in other words, must scale responsibly.
Trust is the Real Infrastructure
If compute is the visible race, governance is the invisible one.
“Responsible AI is not a principle sitting on paper,” he insists. “It is built into the rails.”
On Azure, more than 1,100 LLMs now operate within guardrails built around fairness, privacy, security, and accountability. He emphasises that Microsoft does not use customer data to train its models—a critical concern as enterprises deploy increasingly autonomous AI systems.
As AI shifts from assistant to decision-maker, enterprises must rethink trust, not just compliance.
“This is not one country’s job. AI doesn’t sit in Hyderabad or Seattle. It is everywhere, all the time.”
From Apps to Agents
Over the next five years, he predicts the most visible disruption will occur at the application layer, though not in the way most people expect.
“We are moving from building apps to building agents.”
Historically, employees used software as a tool to complete tasks. In an agent-driven future, digital systems perform the tasks themselves. Project managers may delegate reporting, formatting, data reconciliation, and even decision analysis to AI collaborators.
“You will have digital teammates. You delegate the drudgery, and you focus on the cream of your job.”
The impact, he says, is already visible in enterprises deploying hundreds of internal agents for HR, finance, and operations.
Jobs: Evolution, Not Extinction
Few topics generate more anxiety than AI and employment. He acknowledges the concern, especially in software development, where AI can now generate code.
However, he rejects the idea of displacement. “LLMs can help write code. They cannot decide what to build.”
Customer empathy, problem framing, deployment strategy, and evaluation remain deeply human tasks. AI may automate routine coding and testing, but it elevates developers into system designers and problem solvers.
Beyond technology, he sees even greater impact in agriculture and healthcare.
In partnership with healthcare providers like Apollo Hospitals, AI copilots are helping doctors analyse clinical data faster, improving both efficiency and quality.
“Doctors are getting 20-25% efficiency gains. But more importantly, quality goes up. AI reads every line; you may miss one.”
In agriculture, AI-driven precision farming projects have reportedly increased yields by 20-25% while reducing water usage—spreading specialised knowledge beyond local experts.
“It takes knowledge locked in one expert’s brain and makes it available to any farmer.”
India’s Language Challenge and Opportunity
Building AI for India’s linguistic diversity is expensive, but he argues that layered architectures can make it viable. Smaller language models can interface with larger ‘brain’ models, then render outputs in regional languages.
“You don’t need one giant model for everything. You orchestrate.”
This orchestration, he believes, will define next-generation AI systems, especially in culturally diverse nations.
Fear and the Future
Is he worried about AI?
“Fear, no. Worry, yes.”
His concern is not automation itself but misuse—rogue actors, irresponsible deployment and lack of guardrails. Yet, he remains optimistic, drawing parallels to past technological disruptions from calculators to cloud computing.
“Every technology wave has expanded opportunity. I believe AI can move India from a $4-5 trillion economy to $20 trillion.”
AI may change how we compute, how we build software, and even how we farm. But in his view, its deepest impact may be subtler: moving humans away from repetitive execution towards higher-order thinking.
“We may get lazy in small things. But we will do bigger things.”
In the global GPU race, India may be only a few meters behind. In an agentic future, however, it may have the chance to help define the rules, he concludes.