Every major Indian IT firm wants to wear the AI badge. Infosys has Topaz, Wipro has Wipro Intelligence, LTIMindtree has BlueVerse, Tech Mahindra talks about sovereign LLMs, and TCS is pouring billions into AI data centres. And yet, only one firm has disclosed to the market exactly how much money it is making from AI.
HCLTech revealed that its Advanced AI business earned $100 million this quarter. The rest are content to showcase platforms, agents, and slogans about “delivering AI right”. Everyone insists that AI is creating a significant impact on their deals and revenue, and clients are actively demanding it.
It is easy to dismiss this as AI washing, but the reality is more nuanced. Indian IT companies are not silent on AI revenue because it doesn’t exist. The real issue is that separating AI revenue from the rest of a service contract is nearly impossible, and doing so could create more confusion than clarity.
Too Much Pressure to Call Out AI Revenue
Gaurav Vasu from Unearth Insights explained that the silence on AI revenue might be intentional. “The management may believe that it is not important for them. It is all about management styles. For example, TCS doesn’t give revenue guidance [at all]…They focus on outcomes, not flashy announcements.”
Persistent Systems offers a stark contrast. The mid-sized company doesn’t run campaigns or declare hundreds of AI agents. Instead, it consistently delivers growth. Its 22nd consecutive quarterly growth included a 24% revenue jump and a 45% year-on-year profit rise.
There is also a structural advantage at play. Sanchit Vir Gogia, CEO of Greyhound Research, told AIM that mid-sized firms like Persistent do not carry the same internal complexity as the giants. They are not burdened with layers of governance or sprawling service lines.
“This allows them to act on client signals faster and adopt technologies like AI incrementally, without overcommitting to a single narrative. As a result, while Tier-1s are busy rearchitecting their delivery models and retraining vast workforces, Persistent has been busy signing new deals and expanding margins,” he said. “It is not ignoring AI. It’s using it without making it the centrepiece.”
Across the sector, Indian IT is now also moving beyond proofs of concept (PoCs). “Clients no longer want PoCs,” Wipro CEO Srini Pallia said during the company’s Q2 earnings call. The statement marks an important shift. PoCs don’t pay bills; production-ready solutions do. The industry is now under pressure to show tangible impact on revenue, cost and operational efficiency.
HCLTech is the first to make this transition visible by reporting $100 million from AI solutions this quarter. Others are expected to follow quietly. While analysts predict more disclosure in Q3 or Q4, the numbers are unlikely to be dramatic. That’s not due to negligence but the nature of embedded AI revenue.
“Most of these revenues are part of existing deals,” Vasu said. “None of the Indian pure plays has a separate AI business vertical. Even if you count it, it is embedded in engineering, application development or infrastructure. Counting it separately doesn’t make sense.”
The desire to declare AI revenue is partly driven by global peers like Accenture, which reported $1.4 billion in generative AI bookings in Q2 FY25. But Indian IT model is structurally different. Since most deals already integrate AI, reporting AI revenue separately might inflate expectations and confuse investors.
Counting AI Revenue May Not Make Sense
Meanwhile, Vasu pointed out that if AI revenue doesn’t increase revenue per employee or margins, it is simply part of existing growth. “The announcements are unlikely to change stock market perception. Even HCL’s $100 million is a fraction of the overall quarter.”
Vasu said that Unearth Insight forecasts overall growth of about 3-4% for the quarter.
This explains why, apart from HCLTech and smaller firms like Sonata Software, no major Indian IT firm has been reporting AI revenue separately. Infosys says it is tracking AI revenue internally but will disclose it when the “time is right”.
Wipro highlights over 200 AI agents, but does not reveal their financial impact. LTIMindtree and Tech Mahindra promote their AI ecosystems and trillion-parameter models, yet no numbers back them up. TCS is investing $6-7 billion in AI infrastructure, signalling a long-term play. However, analysts expect short-term revenue and profitability to take a hit.
Apoorv Iyer, global head of GenAI at HCLTech, calls it “vague”. In a previous interaction with AIM, he said, “Almost all deals now include AI. Distinguishing what constitutes AI revenue and what doesn’t is vague.” He compared it to cloud revenue five years ago. “Today, everything is cloud. AI is likely to follow a similar path. Almost all deals now include AI enablement and AI solutioning.”
While AI adoption across the industry is a reality, the pipeline isn’t large enough to meaningfully alter industry-wide revenue. HCLTech’s $100 million is an exception rather than a benchmark. Infosys, Wipro and TCS are quietly converting PoCs into production solutions, but progress remains gradual.
“The coming quarters will test whether AI pipelines convert into real revenue streams,” Vasu said. “Until then, calling out AI revenue is cosmetic. The numbers are either embedded or incremental.”
For now, AI should be viewed as a differentiator, not a revenue stream. TCS’ $6-7 billion investment in data centres positions it for the long term, and Tech Mahindra’s sovereign LLMs should not be interpreted as immediate revenue growth.
Indian IT’s AI journey resembles its early cloud phase. In the beginning, companies did not report cloud revenue separately because it was integrated into every service—consulting, applications and infrastructure. AI is following a similar trajectory. Distinguishing revenue from AI versus traditional services is tricky, and attempting to do so could create artificial narratives.
The post It Doesn’t Really Make Sense for Indian IT to Call Out AI Revenue appeared first on Analytics India Magazine.
