Tata Consultancy Services, India's largest IT-services company, is reorganizing part of its workforce around a single bet: that artificial intelligence will create new outsourcing work rather than destroy it. In an interview with Reuters published around July 12, 2026, CEO K Krithivasan said TCS will build a corps of up to 8,900 "forward-deployed" AI engineers and — in a sharp strategic turn — start shopping for acquisitions.
The forward-deployed model
Forward-deployed engineers, or FDEs, sit inside a client's operations to build and wire up AI systems on-site — a hands-on delivery model popularized by firms like Palantir and now spreading across enterprise software. Krithivasan said TCS would convert "1% to 1.5% of our associates" into FDEs, which against its end-June headcount works out to roughly 5,900 to 8,900 people.
Shopping for AI
The bigger break with history is on the deal front. TCS has for years grown organically, largely avoiding M&A until late 2025. Now, two executives told Reuters, the company is evaluating acquisitions in AI, data security and cybersecurity — a signal it wants to buy capability rather than wait to build all of it. TCS spends about $1 billion a year on talent development and internal AI enablement.
The existential question
The context is an industry staring at its own disruption. Investors worry AI could hollow out India's roughly $315 billion IT-services sector by shortening project timelines, squeezing prices and letting clients demand a share of the productivity gains AI unlocks. If a task that once took a 20-person team now takes five, the headcount-based billing model that built the Indian outsourcing giants comes under direct pressure.
The bet
TCS's answer is to run toward the threat: retrain a slice of its people into embedded AI specialists clients still pay humans to deploy, and acquire the niche skills it lacks. It is a wager that AI becomes a new line of billable work — implementation, integration, security — rather than a deflationary force that erases it. Rivals Infosys, Wipro and HCLTech are making versions of the same calculation.