India’s AI moment is slowly taking shape with IndiaAI Mission, Indian IT announcing billions of dollars of investment, and big-tech firms from the West setting up shop in India. But, is the last one really good news?
It’s déjà vu, if only AI is actually being built for India, and not just in the country.
Indian engineers, developers, data scientists may be powering the next wave of AI, yet most of it isn’t for India. It’s for the West. Even though there is a push for sovereignty in AI and tech, given the recent examples of the push for Zoho and MapMyIndia, the pattern seems to be a puzzle.
OpenAI, Anthropic, Perplexity and ElevenLabs are setting up offices, hiring talent, running pilot projects here. But the question is if they are actually building anything for Bharat, or just sourcing talent from here, while the Indian startups continue to struggle?
“Global AI companies are increasingly establishing their presence in India not merely to tap into data, but to leverage its vast talent base, scalability, and vibrant developer community,” Jaspreet Bindra, co-founder, AI&Beyond, told AIM.While he is optimistic about what this means for the country, not everyone is convinced that all is good.
Sridhar Vembu, co-founder and chief scientist at Zoho, also called this out in a recent interaction with Moneycontrol, saying the Indian government “needs to ensure foreign companies comply with Indian laws and register IP here. Our brains are being harvested, why shouldn’t we register the IP here?.”
A Paradox?
Last week, Nandan Nilekani hosted Anthropic CEO Dario Amodei at his Bengaluru home. According to a report by Moneycontrol, the guests included Rishad Premji, Kunal Shah, Vidit Aatrey, Tarun Mehta, Vivek Raghavan, and discussed positioning India as Anthropic’s R&D hub for mutual growth.
Anthropic’s Bengaluru office is expected to open by early 2026, possibly coinciding with IndiaAI’s launch of its first foundational model with Sarvam AI.
OpenAI is setting base at Delhi with a focus on enterprise, government, and localising products like ChatGPT Go. The goal isn’t India as a market. “It’s not about access to users. It’s about access to talent,” said Gaurav Vasu, CEO of UnearthInsight.
Roughly, one in four developers building agentic AI systems now works from India. India is ready, shaping up as a distributed R&D network for the West. The problem seems to be Indian customers’ reluctance to pay.
The same goes for Indian IT. Infosys trained 300,000 employees on its Topaz AI platform. TCS announced a $6 billion AI investment over six years for an AI data centre for startups, and HCLTech partnered directly with OpenAI and has generated $100 million AI revenue in this quarter.
Cognizant and Hexaware are experimenting with vibe coding, though through their partnerships with western firms like Lovable and Replit. Similarly, TCS is pushing an internal hackathon for 275,000 employees, echoing vibe coding programs.
The playbook is clear: scale, train, experiment. But, most of it still serves Western clients and platforms, and the companies they train with are also Western startups, not the Indian ones.
For example, similar to Replit and Lovable, India is also rising in the vibe coding market. Rocket.new, a Surat-based vibe coding company recently raised $15 million in seed funding led by Salesforce Ventures, with Accel and Together Fund also participating. India’s vibe coding moment seemed imminent.
Despite this, CEO Vishal Virani told AIM that the company generates 26% of revenue from the US, 15–20% from Europe, and only about 10% from India.
There’s also a geopolitical dimension. Ashutosh Singh, CEO of RevRag.ai, said that as the US and China face increasing volatility, companies are diversifying their presence across stable, high-growth markets like India. At the same time, India’s own ecosystem is maturing. “IT firms are collaborating with Western startups, and homegrown AI ventures such as Rocket.new and Emergent are attracting global capital,” he told AIM.
The real question Singh also asks is whether India can evolve from being primarily a delivery centre for Western markets into a creator of AI products that solve India-first problems. “With abundant data, a strong engineering base, and a rising AI policy focus, India is well-positioned but capturing that value will depend on how quickly domestic innovation scales beyond outsourcing,” he added.
Is this a pattern?
These are not isolated cases. Several startup founders have earlier told AIM that most of their market is in the US. Most of the use cases they build only earn money in the English language. Indic language is just not paying enough yet.
Similarly, Infosys’s history also shows the pattern. In 2015, it donated $1 billion to OpenAI, then a nonprofit. Later, OpenAI went for-profit with Microsoft backing. Infosys didn’t invest. Today, Infosys partners with Sarvam for building small language models, but hasn’t invested in the company yet.
Indian IT has the cash—over $20 billion collectively—but risk aversion keeps big bets and investments at bay. Indian IT seems more aligned with training its employees for services to the West, instead of investing in Indian AI.
This is one of the reasons guiding Indian startups to rely on foreign platforms or overseas enterprise contracts. Some founders admit a “Skip India” approach as Indian customers are stuck with POCs. Global customers first, local adoption later.
Some even argue that Indian enterprises take startups seriously only if their products are backed by international validation.
To make things possibly better, IndiaAI Mission has selected startups and deployed tens of thousands of GPUs. But it is still in the process of building AI foundational models, and the adoption is slow. Till then, Indian companies would rely on Western models. Even after Indian models are released, adoption by companies here can’t be predicted to be immediate.
The result: India builds AI for the West, American companies come to India for the talent. All of this while innovation for the country lags. And that leaves room for OpenAI and Anthropic to move in, which all-in-all is not wrong on their part.
The post Is India Building AI for the West Again? appeared first on Analytics India Magazine.
