The deal, which pairs Anthropic’s Claude models, including Claude Code, with Infosys’s proprietary Topaz AI platform, is aimed squarely at sectors like telecommunications, financial services, manufacturing, and software development.
“There’s a big gap between an AI model that works in a demo and one that works in a regulated industry — and if you want to close that gap, you need domain expertise,” said Dario Amodei, CEO and co-founder of Anthropic. “Infosys has exactly that kind of expertise across important industries: telecom, financial services, and manufacturing.”
Infosys CEO Salil Parekh has identified six new service lines he believes AI will unlock — spanning AI strategy and engineering, agentic legacy modernisation, data readiness, process reimagination, physical AI, and AI trust — and is positioning the company to capture what it estimates is a $300–$400 billion market opportunity by 2030. “AI is not just transforming business — it is redefining the way industries operate and innovate,” Parekh said.
At the operational core of the collaboration is agentic AI — systems capable of independently executing multi-step tasks such as processing insurance claims, generating and testing code, or running compliance reviews, rather than simply responding to one-off queries. Using tools like the Claude Agent SDK, Infosys plans to build AI agents that can work persistently across long, complex enterprise workflows. Legacy modernisation, long a bread-and-butter revenue stream for Indian IT, also features prominently, with the two companies combining their platforms to accelerate migration away from aging infrastructure.
The deal kicks off with a dedicated Anthropic Center of Excellence focused on the telecom sector, before expanding into the other verticals.For analysts, the announcement offered some relief, even if questions remain. “This development is encouraging, as it suggests that next-generation AI applications are unlikely to disrupt Indian IT companies’ business models to the extent initially feared,” said Vinod Nair, Head of Research at Geojit Investments. “Instead, these solutions are expected to be incorporated into both existing and new projects, which should help ease concerns around long-term business sustainability.”
Nair was careful not to declare the all-clear. Uncertainties around deal sizes, pricing evolution, and the net margin impact, once employee-cost efficiencies are weighed against productivity gains, remain unresolved. He described the sector’s outlook for FY27-28 as “muted compared with the strong performance of the past two to three years,” though he noted that subdued valuations were beginning to present re-entry opportunities for long-term investors.
Sumit Pokharna of Kotak Securities was more direct in endorsing the strategic logic. “This is definitely a step in the right direction and the need of the hour,” he said, emphasising that regulated industries simply cannot adopt AI through trial and error. “They need governance, transparency, compliance, reliability, and security” — which is, in effect, the value proposition Infosys is now selling.
Also read: IT stocks attempt to get back on their feet. What Nomura, UBS and Citi are saying
The broader significance of Tuesday’s announcement extends well beyond one partnership. Indian IT has spent the better part of two years on the defensive, fielding investor concerns that generative AI would compress headcount, shrink project scopes, and erode the labour-arbitrage model that built the sector. The Infosys-Anthropic deal is a deliberate attempt to rewrite that narrative and position Indian IT not as a casualty of the AI wave, but as the indispensable layer between frontier models and the messy, compliance-laden reality of global enterprise.
Whether that repositioning holds will depend on execution, deal flow, and whether pricing power follows. For now, the market has given Infosys the benefit of the doubt.
(Disclaimer: Recommendations, suggestions, views and opinions given by the experts are their own. These do not represent the views of The Economic Times)
