Can India's $1 trillion AI ambition and homegrown talent overcome deep tech gaps and global rivals?

Aadhaar. Unified Payments Interface. Digital India. Three innovations vaulted India onto the world stage as a digital powerhouse. Yet, in the high-stakes race for foundational artificial intelligence (AI), the technology poised to redefine global economies and geopolitics, India has been mostly on the sidelines.

Slowly, that’s starting to change. Propelled by the government’s IndiaAI Mission, the country is seeking to build its own AI capabilities. The government has tapped several domestic start-ups in the mission to develop local foundational models, a bold step signaling India is no longer content consuming global technology. It wants to shape what it consumes.

Jaspreet Bindra, founder of AI & Beyond, which trains corporate executives in AI, acknowledges India’s growing research and innovation in the field. “India is top 10 in terms of papers published, patents and corporate investment in AI,” he says. “India has software services giants such as TCS (Tata Consultancy Services) and Infosys which serve almost all Fortune 500 companies. They can build infrastructure and applications around Generative AI for these clients and dramatically lift their engineers’ productivity using AI,” he adds.

India arriving late at a high-tech party is not new. It was late for the Cloud revolution in computing, was initially tentative in the space race and has often erred on the side of caution when others dared to bet big. And yet, there’s one thing the country has never lacked: the resolve to play catch-up with—and sometimes leapfrog—rivals. With AI rapidly becoming the backbone of digital economies, the Indian government has set itself the goal of building a $1 trillion AI industry by 2030.

But winning in foundational AI—a form of generative AI models trained on vast amounts of diverse data, enabling them to be adapted for a wide range of applications—demands more than ambition. From graphic processing units (GPUs) and high-bandwidth memory to advanced chip fabrication units and specialised electronic design automation (EDA) tools, the tech stack underpinning AI is expensive and complex.

Consider the scale of investment: the US has poured about $10 billion into its AI ecosystem and China has invested between $15 billion and $30 billion. India’s AI investment is just $1.2 billion.

Morgan Stanley estimates that China’s AI investments could break even by 2028 and deliver a 52% return on invested capital by 2030, a projection that underscores the economic stakes of foundational AI.

A debate is under way, however, on whether investment alone guarantees success. Cost-effective launches of smaller AI models are proving that innovative thinking may sometimes matter as much as deep pockets.

Even so, the scale of the challenge confronting India is formidable.

More than catching up

Without building domestic capacity, India will run into “structural ceilings,” says Nitin Mittal, Global AI Leader at Deloitte. “In the short run—from 2025 to 2030—pooling demand, like MeitY’s bulk GPU procurement, and leveraging India’s design talent can drive growth,” says Mittal.

“But beyond 2030, without indigenous advanced-node fabs, high bandwidth memory (HBM) and 2.5-D/3-D packaging, India risks forfeiting potential AI-led GDP (gross domestic product) gains, ceding standards-setting power, and seeing its top talent migrate to countries that control the Silicon stack,” says Mittal. MeitY is short for the Ministry of Electronics and Information Technology. 2.5-D and 3-D are advanced semiconductor packaging technologies.

Under the IndiaAI Mission, the government has procured around 34,000 GPUs from giants like Nvidia Corp., Advanced Micro Devices Inc. Amazon Web Services Inc. and Intel Corp. This is just a fraction of what it needs.

The world is, meanwhile, racing ahead. Companies like Perplexity, the AI-powered search engine, are redefining the browser experience. Western enterprises are weaving AI into the fabric of business processes. Global attention is rapidly shifting towards the next frontier: Artificial General Intelligence (AGI) that possesses human-level cognitive abilities.

India has already missed the first wave of large language models (LLMs), as AI models from the US, China, France, and even the United Arab Emirates (UAE) surged ahead. The UAE, for instance, has emerged as a surprise AI hub with its powerful Falcon LLM, currently the most advanced, large Arabic-language model.

Yet, India may still hold a unique edge.

Small is beautiful

India’s advantage lies in small language models (SLMs) tailored for local contexts and languages, say tech experts. With its vast population and cultural and linguistic diversity, the country is fertile ground for specialised, context-aware AI.

“LLMs are very good for general-purpose enquiry, insights and information, but they are not necessarily purpose-fit or contextualized for a certain domain, industry, or client,” says Mittal. “This is where you need SLMs. India can be at the forefront, not by building monolithic foundation models, but by focusing on how fast and at what cost you apply AI, making it mainstream in enterprises,” he adds.

In other words, India’s AI future may not rest in matching Silicon Valley model-for-model, but in solving problems unique to its markets at a massive scale and lower cost.

One undeniable strength India brings to the AI race is talent. Since 2016, the country has seen a staggering 256% increase in AI talent—the highest growth rate globally, according to the AI Index Report 2025.

India has around 18 million software developers, making it the world’s second-largest talent pool after the US in terms of GitHub activity. GitHub is a proprietary developer platform for developers to create, manage and share code.

Secret weapon?

Perhaps India’s biggest opportunity lies in a uniquely homegrown concept: Digital Public Infrastructure (DPI).

Services like UPI for payments and Aadhaar for digital identity have transformed the nation, reducing friction in financial transactions, enabling financial inclusion, streamlining welfare distribution, and even facilitating the world’s largest vaccination programme, undertaken after the coronavirus pandemic.

Why not build and deploy AI at scale by contextualising SLMs to help companies move from experimenting with generative AI to mainstreaming it? “This can increase employee productivity, improve business processes, generate new revenue, and rethink cost structures,” says Bindra.

Puneet Chandok, President of Microsoft India and South Asia, says, “At Microsoft, we’re seeing Indian enterprises report up to 5X ROI (return on investment) on GenAI investments, and 90% of AI power users in the country now begin their day with AI. This is a shift in how work gets done, how decisions are made, and how value is created. India’s strength lies in its people.”

India cannot afford to rely solely on foreign AI systems, experts warn. Dr Vivek Raghavan, co-founder of AI venture Sarvam, argues that such dependence is strategically short-sighted. “India has all the ingredients to lead: linguistic and cultural diversity, a fast-digitising economy, ambitious digital public infrastructure, and deep technical talent. What we need now is the will to align these strengths into a long-term national capability,” he says.

“If India had its own cloud, we’d control our data. Without it, we’re dependent,” says Punit Pandey, founder and Chief Information Officer of Ojas Softech and Astrosage AI. “Some say, ‘We’re doing fine without it,’ but how? We have no global tech products despite having 30–50% of Silicon Valley’s engineers. This isn’t ‘fine’. Colonial history may have made us risk-averse, but we must change.”

Only by securing the full Silicon value chain can India safeguard its AI growth trajectory, reduce strategic vulnerabilities, and transform from a global back-office to an end-to-end AI powerhouse, say experts.

At the end of the day, the question remains: Can India’s bold AI ambition translate into real economic power or will it remain just another unfinished chapter in its long saga of potential waiting to be tapped.

@palakagarwal64