Why India will become the nerve centre for all things AI

There is a view in the global markets that the AI revolution is passing India by. Investment attention remains concentrated on countries developing advanced semiconductors, manufacturing GPUs and building frontier AI models. By these measures, India may appear to be a follower rather than a leader.

But this overlooks a fundamental distinction: who creates AI technology and where that technology will be deployed, consumed, and delivered at scale.

The AI revolution will not be defined by chips and models alone. It will also be shaped by countries that build the infrastructure, platforms, applications, and ecosystems through which artificial intelligence becomes accessible to billions of people and businesses.

India has an extraordinary opportunity to become one of those countries.

Our ambition should extend beyond becoming a large consumer of AI. India can become a global nerve centre for AI infrastructure, sovereign cloud, model development, inference and the export of AI-powered services.

Changi did not build the aircraft

Singapore never manufactured commercial aircraft. It built Changi Airport, developed a world-class aviation ecosystem and made itself indispensable to international aviation.

There is an important parallel with artificial intelligence.

Today, advanced GPUs and frontier models come from a relatively small number of global technology companies. India must certainly develop indigenous capabilities in semiconductors, models and foundational technologies. Technological self-reliance remains an important national objective.

But we need not wait until every component is manufactured domestically before building a globally competitive AI ecosystem.

We can bring the world’s best technologies onto Indian infrastructure, combine them with Indian engineering capabilities, operate them under Indian jurisdiction, and deliver AI services globally.

The winners of the AI revolution will not necessarily be only those who invent intelligence. They will also be those who industrialise it, democratise it and deliver it at scale.

India can do all three.

Sovereignty is about control, not isolation

As AI becomes central to governance, national security, financial systems and enterprise operations, sovereignty will increasingly determine where critical workloads are deployed.

However, sovereignty is frequently confused with data residency.

Simply storing data within India’s geographical boundaries does not automatically make an AI or cloud platform sovereign.

True sovereignty requires control over infrastructure, encryption keys, administrative access, operations and the legal framework governing the platform. It also requires operational resilience against unilateral decisions by foreign technology providers or governments.

This distinction is particularly important for government workloads, regulated industries, and strategic national data.

India must develop genuinely sovereign cloud and AI infrastructure operated under Indian control and jurisdiction.

But sovereignty should not mean technological isolation. We should leverage the best global GPUs, networking technologies, software and AI models while retaining control over the infrastructure and operating environment supporting critical workloads.

Our ambition should be global technology, Indian infrastructure, Indian operational control and global competitiveness.

Such capabilities can also become an export opportunity, serving international customers seeking trusted, secure and jurisdiction-sensitive AI infrastructure.

From the world’s back office to its AI engine room

India’s greatest advantage is that it is not starting from scratch.

Over three decades, we have built a globally competitive technology services industry, an enormous engineering talent pool, thousands of global capability centres, and a thriving digital economy.

The natural progression is from exporting human intelligence through IT services to exporting machine intelligence through Indian AI platforms.

India can host global AI workloads, train and fine-tune models, deliver inference services, and develop enterprise AI applications for customers worldwide.

The opportunity is not merely to consume AI but to become a global producer and exporter of intelligence.

International customers are equally important to this ambition. They bring substantial computing demand, long-term commercial commitments, advanced technological requirements and global operating standards.

Serving these customers strengthens the infrastructure and expertise available to Indian enterprises.

There is no contradiction between serving global AI companies and building India’s sovereign AI capabilities. The two can reinforce each other.

The missing ingredient is bankable demand

India’s AI ambitions require enormous investments.

Deploying advanced GPUs involves billions of dollars in computing hardware, power infrastructure, cooling, networking, storage and specialised facilities.

Unlike conventional IT infrastructure, GPUs face rapid technological evolution, relatively short economic cycles and significant financial risks.

Globally, large AI infrastructure investments are supported by long-term customer commitments, frequently spanning several years. These provide predictable revenues against which infrastructure providers can raise debt and equity capital.

India cannot build globally competitive AI capacity through fragmented demand and short-duration procurement contracts alone.

A one-year compute contract cannot ordinarily support infrastructure whose investment economics extend across several years.

This is where government policy becomes decisive.

Government must become India’s anchor AI consumer

I have consistently argued that the Government of India must play three complementary roles in accelerating our AI ambitions.

First, government must become a major consumer of AI.

AI should be deployed extensively across public administration, healthcare, education, agriculture, transportation, public-sector enterprises and citizen services.

The objective should not simply be to subsidise AI infrastructure but to use AI at scale to improve productivity, governance and public-service delivery.

Second, government must become a demand aggregator.

Individual ministries, departments, states, public-sector organisations and research institutions have different computing requirements.

Aggregating these requirements into coordinated national demand can create the scale and predictability necessary for large infrastructure investments.

It can also improve procurement efficiency, reduce costs and ensure better utilisation of national computing capacity.

Third, government must create long-term, bankable offtake commitments.

If India wants domestic infrastructure providers to compete for advanced GPUs in a globally competitive market, our procurement structures must reflect international commercial realities.

Multi-year contracts, ideally five years, with predictable capacity commitments, commercially sustainable terms and appropriate performance safeguards can enable providers to finance large-scale deployments.

This is not an argument for guaranteed payments without accountability. Contracts can incorporate phased capacity creation, service-level commitments and performance obligations.

But the principle is fundamental: India cannot expect globally competitive AI infrastructure without creating globally competitive conditions for financing it.

The IndiaAI Mission has made an important beginning by democratizing access to GPU capacity for startups, researchers and enterprises.

Encouragingly, the government’s evolving approach towards capability-based empanelment, longer-term commitments and demand-linked capacity creation represents a step in the right direction.

The next challenge is translating this policy direction into executable contracts.

Indian infrastructure providers have demonstrated their willingness to invest substantial capital in sovereign AI capacity. Government must make it financially viable for that capacity to serve Indian startups, enterprises and national priorities.

The next revolution is in token factories

Another fundamental shift is underway in the AI industry.

Much of the early excitement focussed on training increasingly powerful foundational models. Training remains important, and India must continue investing in indigenous models, particularly for Indian languages and sector-specific requirements.

But the next wave of adoption will increasingly be driven by inference: the actual use of trained models to generate answers, automate workflows, analyze information and perform tasks.

As millions of enterprises and consumers embrace AI applications, inference demand will expand dramatically.

This creates an opportunity for industrial-scale AI token factories.

Just as conventional factories convert raw materials into finished products, token factories convert computing infrastructure into usable intelligence.

They combine GPUs, optimised inference engines, model-serving platforms, orchestration systems and APIs to deliver intelligence efficiently and affordably.

Most enterprises do not want to purchase GPUs or manage complex computing clusters. They want to consume intelligence through APIs, tokens and applications, paying for what they actually use.

The industry must therefore evolve beyond merely renting GPU capacity towards delivering AI as a consumable service.

India can become a global leader not merely in supplying computing capacity but in industrialising the production and distribution of affordable intelligence.

This is also where public policy should increasingly focus: encouraging actual consumption of AI services and tokens, rather than measuring success solely through installed GPU capacity.

Infrastructure, power and execution will decide the outcome

AI infrastructure requires far more than GPUs.

Reliable electricity, high-density data centres, advanced cooling, high-speed networking and specialised operational capabilities are equally critical.

In several established global markets, power availability and grid connectivity have become significant constraints on AI infrastructure expansion.

India has an opportunity to develop large, purpose-built AI campuses, but aggregate national electricity-generation capacity does not automatically translate into AI-ready power.

What matters is reliable, competitively priced electricity at the locations where computing infrastructure is deployed.

We must accelerate grid connectivity, strengthen power distribution, enable renewable-energy integration and streamline approvals for large infrastructure projects.

We must also develop specialized skills in liquid cooling, electrical infrastructure, GPU cluster operations, networking and AI platform management.

The scale of the opportunity demands execution discipline comparable to that which helped India build world-class digital public infrastructure.

The next three years will be decisive

India has the essential ingredients to become a global AI powerhouse: engineering talent, a vast domestic market, entrepreneurial ambition, growing infrastructure capabilities and an increasingly supportive policy environment.

But these advantages do not guarantee success.

The global AI industry is moving at extraordinary speed. Investment decisions being made today will determine where computing capacity and AI ecosystems become concentrated over the coming decade.

Government must create predictable demand and long-term investment confidence. Industry must bring capital, technology partnerships, innovation and execution excellence.

Together, we must make AI accessible to Indian startups, enterprises, researchers and citizens while building globally competitive platforms that serve international markets.

In the 2000s, India became the world's back office. It subsequently became its software and engineering powerhouse.

Now it has the opportunity to become the world's AI engine room, where intelligence is computed, industrialized and delivered.

India should not aspire merely to participate in the global AI revolution. It should aspire to power it.

(The author is Co-founder, Managing Director & CEO, Yotta. Views are personal.)