India’s Pursuit of AI Innovation: Is the Country Lagging Behind?

Getty Images India has yet to develop its own foundational language model similar to DeepSeek, which powers chatbots and other AI applications.

Two years after ChatGPT took over the world, China’s DeepSeek has made waves in the tech industry by drastically reducing costs for developing generative artificial intelligence (AI) applications. However, as the global competition for AI supremacy intensifies, India seems to have fallen behind, especially when it comes to creating its own foundational language model that powers chatbots and other systems.

The government claims a homegrown equivalent to DeepSeek is not far away. To accelerate development, they are supplying startups, universities, and researchers with thousands of high-end chips in under 10 months. Additionally, global AI leaders have been praising India’s capabilities recently.

After initially dismissing the potential of Indian tech talent, OpenAI CEO Sam Altman now acknowledges that India should play a leading role in the AI revolution. Currently, India is OpenAI’s second-largest market by users. Microsoft has also invested significantly – committing $3 billion for cloud and AI infrastructure. Nvidia’s Jensen Huang speaks highly of Indias “unmatched” technical talent as crucial to its future potential.

With 200 startups working on generative AI in the country, there is considerable entrepreneurial activity underway. However, experts warn that without basic structural reforms in education, research and state policy, India risks lagging behind.

China and the US already have a significant advantage having invested heavily in research and academia as well as developing AI for military applications, law enforcement, and large language models over the past four to five years. While ranking fifth globally on Stanford’s AI Vibrancy Index based on metrics such as patents, funding, policy, and research, India is far behind these two superpowers.

Between 2010 and 2022, China was granted 60% of the worlds total AI patents while the US received 20%. In contrast, India secured less than half a percent. Furthermore, private investment in Indian AI startups is a fraction compared to what’s allocated to similar initiatives in the US and China.

India’s state-funded AI mission amounts to only $1 billion; this pales in comparison with the staggering $500 billion that has been earmarked by the US for its Stargate initiative aimed at constructing massive AI infrastructure. Meanwhile, China reportedly plans to invest over $137 billion toward becoming an AI hub by 2030.

Although DeepSeek’s success demonstrates that AI models can be developed on older and less expensive chips C something India can take comfort in C a lack of “patient” or long-term capital from either the industry or government is a major obstacle, according to Jaspreet Bindra, founder of a consultancy focused on building AI literacy within organizations. While it cost $5.6 million for DeepSeek’s model development, there was much more support and resources behind it.

Another problem is the absence of high-quality India-specific datasets required for training AI models in regional languages like Hindi, Marathi, or Tamil due to linguistic diversity in the country.

The significant success of Indias payments revolution serves as a positive example. It was largely driven by strong collaboration between government, industry, and academia through initiatives such as the Unified Payment Interface (UPI), which transformed digital transactions.

However, while Bengaluru’s $200 billion outsourcing industry should have been at the forefront of India’s AI ambitions given its large pool of coders, IT companies have failed to shift their focus from cheap service-based work toward developing foundational consumer AI technologies. This gap has largely fallen on startups and government missions to fill.

Many experts share the view that it will be several years before India can produce anything like DeepSeek. In the interim, they believe leveraging existing open-source platforms such as DeepSeek could help leapfrog progress in their own development efforts. But in the long term, creating a foundational model is critical to securing strategic autonomy within the AI sector and reducing dependency on imports.

Additionally, increasing computational power or hardware infrastructure would require manufacturing semiconductors C an industry that has yet to take off in India. Much of this needs to be achieved before meaningful progress can be made toward closing the gap with China and the US.

In conclusion, while there are significant opportunities for Indias AI sector, addressing key challenges related to funding, research infrastructure, and collaboration will prove crucial if it is truly to achieve its potential in a competitive global landscape.