India bets on open-source AI to challenge US tech giants: Himanshu Tyagi

Sentient founder Himanshu Tyagi advocates for open-source AI to ensure India's technological sovereignty, as the country's AI talent pool grows rapidly and aims to challenge US and Chinese dominance

India is positioning itself at the forefront of the global race to build artificial general intelligence (AGI), with homegrown entrepreneurs like Himanshu Tyagi betting on open-source development that can challenge the dominance of American tech giants and offer developing nations a path to AGI leadership.

 

He argues that unless countries like India develop their own advanced AI capabilities, they risk being permanently locked out of the future economy as intelligence becomes increasingly centralised in the hands of a few Western corporations.

 

Tyagi is a professor at the Indian Institute of Science (IISc), Bangalore, and co-founded $1.2 billion artificial intelligence (AI) startup Sentient.

 

He said that ‘Closed AI’ represents the greatest heist of public knowledge in modern history. Tyagi co-founded Sentient along with Sandeep Nailwal last year due to his passion for the potential of open source AGI. He fears the monopolisation of AI by private, proprietary and closed-source vendors.

 

“Unless there is a third front where open source AGI is coming, countries like India will have a hard time more and more in the future, because AI has become essentially everything,” Tyagi told Business Standard.

 

AGI — AI systems that can outperform humans across cognitive tasks — could redefine national power in the decades ahead, according to experts. In that race, the US and China currently have the largest AI talent pool globally.

 

This is followed by India whose AI talent pool has grown to over 416,000 professionals in 2025, with projections exceeding 1.25 million by 2027, according to industry data.

 

India’s AI market is expected to triple to $17 billion by 2027, driven by its demographic dividend, vibrant startup ecosystem, and enterprising talent hungry to innovate and compete globally.

 

“Make India a compute hungry country by building reasoning models from India. There's no other option, if the country has to remain relevant,” said Tyagi.

 

This is where his venture Sentient is playing a role. It is fighting the privatisation of AI by developing a fully open, community-built alternative.

 

Backed by Peter Thiel’s Founder's Fund, Pantera Capital, and Framework Ventures, Sentient is committed to developing an ecosystem for ‘Loyal AI’: technology that is open to all, aligns with community values, and provides fair incentives to builders.

 

“Our mission is simple: make sure open source AI stays viable, monetisable, and decentralised. That means builders, not just big tech, can innovate, share, and profit from what they create,” said Tyagi.

 

At the centre of Sentient's mission is Roma (Recursive Open Meta-Agent), Sentient’s open-source orchestration framework that coordinates multiple lightweight AI agents into modular, composable “super agents.”

 

The company said Roma has already outperformed Big Tech’s proprietary models on key reasoning and search benchmarks. It is proving that open frameworks can compete with and surpass closed systems.

 

Sentient has developed ‘The Grid’, an open, developer-led network that unites AI agents, datasets, and tools into composable workflows. This enables creators and enterprises to build, coordinate, and monetise intelligence at scale.

 

“Think of The Grid like an app store, but for intelligence,” added Tyagi. “Anyone building a smart agent, whether it writes code, plans your day, or teaches a kid, should be able to offer it up, and get paid when it’s used. That’s the world we’re creating.”

 

Tyagi said most major players are building monolithic AI — large models in black boxes. The Grid is building an Internet of agents: small, specialised intelligences that talk to each other and evolve together.

 

“It’s a more modular, decentralised, and ultimately more open path to AGI,” he said.

 

He also cautioned against dependence on AI models originating in China, citing long-term strategic risks and warned that such models may not remain viable.

 

Tyagi advocates for globally accessible, state-of-the-art open-source reasoning models.

 

“We want to make sure that they all can have the open source models and contribute back to it,” he said, referencing countries like India, South Korea, Japan, and Australia.

 

He added that India poses “no regulatory hurdle on building AI,” and instead offers “encouragement, if anything.”

 

While domestic policy remains largely undeveloped, Tyagi notes that external pressures — particularly US regulations from the Trump era — are shaping where Indian firms can source or host AI models.

 

India is stepping up efforts to secure critical AI infrastructure, focusing on graphics processing units (GPUs) vital for advanced computing.

 

The government has unveiled a ₹10,000 crore fund-of-funds to spur innovation in AI and deep tech. Private players are also moving in — Reliance Industries is planning the world’s largest data centre, while Infosys and top universities plan to invest in GPU capacity.

 

“There's a lot of even strategic interest from the government to build reasoning models. One should expect a good, decent reasoning model to come out of India soon,” said Tyagi.