Two years after ChatGPT took the world by storm, China's DeepSeek has sent ripples through the tech industry by collapsing the cost for developing generative artificial intelligence applications.
But as the global race for AI supremacy heats up, India appears to have fallen behind, especially in creating its own foundational language model that's used to power things like chatbots.
The government claims a homegrown equivalent to DeepSeek isn't far away. It is supplying startups, universities and researchers with thousands of high-end chips needed to develop it in under 10 months.
A flurry of global AI leaders has also been talking up India's capabilities recently.
At first, OpenAI CEO Sam Altman did not see India as a key player in AI. However, this month, he acknowledged that India should take a leading role in the AI revolution. The country is now OpenAI’s second-largest market in terms of users.
Major companies are also investing in India’s AI sector. Microsoft has pledged $3 billion (£2.4 billion) for cloud and AI infrastructure. Nvidia’s CEO, Jensen Huang, praised India’s exceptional technical talent and its potential for AI growth.
With around 200 startups focused on generative AI, there is strong entrepreneurial activity in the country.
However, despite having important resources, India risks falling behind. Experts believe the country needs major improvements in education, research, and government policies to stay competitive.
China and the US have a significant head start of four to five years. They have made large investments in research, universities, and AI applications for the military and law enforcement. They have also advanced in building large AI models, according to technology analyst Prasanto Roy.
India ranks among the top five countries in Stanford’s AI Vibrancy Index, which measures progress in patents, funding, policies, and research. However, it still lags far behind the US and China in critical areas.
Between 2010 and 2022, China secured 60% of global AI patents, while the US obtained 20%. India, on the other hand, received less than 0.5%.
Indian AI startups also attracted much less private investment compared to their counterparts in the US and China in 2023.
The Indian government has allocated $1 billion for its AI mission. This amount is significantly lower than the US’s $500 billion Stargate project, which aims to build large-scale AI infrastructure. It is also far behind China’s $137 billion plan to establish itself as a global AI leader by 2030.
DeepSeek has shown that AI models can be built using older and cheaper chips, which is something India can learn from. However, Jaspreet Bindra, founder of an AI consultancy, says the bigger issue is the lack of long-term investment from both the government and private sector.
"Although DeepSeek is said to have developed its model with $5.6 million, much more money was actually invested behind the scenes," he explains.
Another challenge is the lack of high-quality datasets for training AI models in Indian languages like Hindi, Marathi, and Tamil. India's language diversity makes this even more difficult.
Despite these challenges, India has a strong talent pool, with 15% of the world's AI workforce coming from the country. However, research from Stanford shows that many Indian AI professionals are moving abroad.
Mr. Bindra says this is because top AI innovations usually come from deep research in universities and corporate labs, and India lacks a strong research ecosystem. There are very few major AI breakthroughs from Indian academic or corporate institutions.
He believes that India's successful digital payments revolution was possible due to strong collaboration between the government, industry, and universities. A similar approach is needed to boost AI development.
For example, India's Unified Payments Interface (UPI), a government-backed system, has transformed digital payments by allowing instant transactions. In contrast, India’s $200 billion outsourcing industry, which employs millions of coders, has not shifted its focus from service-based work to AI innovation.
"They left a big gap, expecting startups to fill it," says Prasanto Roy.
However, he doubts that startups and government programs can catch up quickly. The government’s goal of building an AI model in 10 months, he says, is an overreaction to DeepSeek's success.
"I don't think India will be able to create something like DeepSeek in the next few years," he adds, a view shared by many experts.
For now, India can focus on improving AI applications using open-source platforms like DeepSeek. Bhavish Aggarwal, founder of India's early AI startup Krutrim, recently suggested that this could help India make faster progress in AI.
But in the long run, developing India's own AI model will be crucial for independence and security. Experts say this will require more powerful computing infrastructure, including semiconductor manufacturing, which India has yet to establish.
Until these elements come together, the gap between India and AI leaders like the US and China will remain significant.
