We Need New Infosys and TCS

India’s IT services story is entering a new and uncomfortable phase. Headcount is shaky. AI is cannibalising long-held revenue streams. The comfort of being the world’s back office is no longer enough, and three decades without innovation is not helping either. 

The bigger shock is that the AI companies that were supposed to eat into services are now building services themselves. 

Anthropic joined hands with Blackstone, Hellman & Friedman, and Goldman Sachs to launch an enterprise services arm—or what it calls “new AI services company”—even as it continues to work with Infosys, TCS, and other Indian firms. OpenAI is doing something similar. 

Interestingly, Sarvam, the AI superstar of India, despite a possible investment from HCLTech, is building its enterprise AI ecosystem with the recent partnership with YCP India.

That should be read as a warning, not a footnote. The moat around Indian IT won’t disappear overnight, but it is getting thinner, more crowded, and far less predictable.

That is why the industry needs a new kind of Infosys and TCS. Not a copy of the old pyramid. Not another labour arbitrage machine dressed up with AI jargon. What India needs now is a services company that can own delivery, product thinking, data, and workflow redesign simultaneously. Maybe even build some models of its own.

The Contradiction

The urgency—and threat—is being felt across. AI labs are no longer content to sell models from the sidelines. They want deeper enterprise capture. 

But their move also exposes a contradiction. 

Even as Anthropic and OpenAI chase enterprise relationships, they still need the very Indian IT firms they seem poised to displace. Infosys has already partnered with Anthropic. TCS is building AI-optimised data centres with OpenAI as an anchor tenant for Stargate India. 

Gaurav Vasu, Founder and CEO of cognitive intelligence platform UnearthInsight, believes it’s only logical that native AI companies build services around AI products or solutions. 

“They cannot remain pure-play AI solutions only,” he recently told AIM, adding that this “validates that services companies, whether consulting services, technology services, or engineering services, are still relevant and still important.” 

That is key. The AI labs may own the model, but they do not automatically own the enterprise rollout. The last mile still needs human systems, trust, and delivery muscle.

Praveen Bhadada of consulting firm Neovay Global made a similar point. “It’s in their interest for services companies to know these capabilities and then leverage them for a three-part transformation—protect your current business, launch newer services, and then drive towards outcomes,” he earlier told AIM. 

That captures the real challenge for Indian IT. 

Protection alone will not be enough. Neither will hype. The companies that survive this structural transformation will have to do all three things at once, even as margins, hiring models, and client expectations shift beneath their very feet.

The market has already started pricing in that anxiety. The Nifty IT index has been battered this year. The concern is not only that AI will reduce the need for junior work. It is that the old pricing architecture itself is under pressure. If AI compresses L1 and L2 work, then the pyramid model that powered Indian IT for decades collapses. 

That old logic is exactly what Nandan Nilekani also targeted. The Infosys co-founder argued that India does not need to lead the world in building the most advanced AI models, but must ensure AI benefits are widely shared. His larger point was that India should focus on usage, not try to outgun the US and China in frontier model training. It needs to be the perfect distributor.

The Consistent Problem of Not Building Enough

But there’s an undercurrent of deeper frustration with India’s service-first identity. 

That is visible in the current AI services race, too, as AI labs move into enterprise execution, implementation, and transformation. This has traditionally been the turf of Indian IT. The real fight is not model versus services. It is who sits closest to the customer when AI moves from demo to deployment.

As Vishnu Subramanian of Jarvislabs AI noted in his blog, the deeper problem is not whether AI is a threat, but whether enterprises can absorb change fast enough. 

Training a model is one thing, he argued, but turning it into a usable, reliable product is where the real value sits. “The framing I keep coming back to is this: Indian organisations need to create IP and solve these problems so that we do not become cheap infrastructure for the world, just like we became cheap labour for the IT sector,” he wrote.

He said that IT is not inherently a bad business. “The problem is what we did not build enough of.”

The question, then, is what kind of Indian company can come out of this moment. The traditional answer would be another outsourcer with a larger headcount and a few AI slides. That will not be enough. The next Indian giant has to be willing to break the old model before the market breaks it for them. 

It has to invest in engineering depth, create products that can travel globally, and build capability in data, deployment, and workflow ownership. It has to think more like a platform company and less like a body shop. And it has to stop pretending that cheaper labour is a strategy.

IT companies are already bracing for this shift. Many are now focusing on digital labour, pricing AI agents into their services.

The New Indian IT?

But it is, arguably again, not going to be enough.

Srikanth Velamakanni of Fractal Analytics sees the current moment less as a disruption and more as a validation. In a recent LinkedIn post, he said that the last 12 months of AI progress have led to three clear shifts inside enterprises. 

Every workflow is now a candidate for AI-led transformation. Legacy systems are being rebuilt around data and ontologies. Even the structure of work itself is being rethought. 

But the bigger takeaway, in Velamakanni’s view, is that models alone are not enough. Enterprise AI only works when services step in to handle messy data, integrate with complex systems, and drive organisational change. That is where the real bottleneck sits today. 

Demand for AI transformation is scaling far faster than the capability of people who can actually deploy it.

This debate has hit a raw nerve. One post on X argued that clients will still return to traditional relationship managers for “AI-enabled labour.” 

The cynicism is not baseless. It comes from years of watching IT majors promise transformation while delivering the same outsourcing model in a shiny wrapping. If the sector does not evolve, the criticism will only get louder.

As Abhishek Upperwal, Founder and CEO of Soket AI Labs, said on X, “I hope Indian IT service companies now realise they need to work and invest in Indian startups.” If not, then “we [Soket AI] will be Anthropic one day and start our own service firm.”

So yes, India needs new Infosys and TCS. Not to replace the old icons, but to build a new generation of companies that can survive the AI reset.