Tata Consultancy Services plans to stand up a team of as many as 8,900 forward-deployed engineers, AI specialists who embed inside client operations to make AI systems actually work in production. CEO K Krithivasan disclosed the plan in an interview with Reuters, published July 12, framing the cohort as 1% to 1.5% of the company’s associates.
The move borrows a playbook made famous by Palantir and since adopted by OpenAI, Anthropic and Microsoft: put engineers next to the customer, not behind a ticket queue. For the world’s largest IT services firm, it is one of the more consequential enterprise business technology bets of the quarter, and a wager that AI does not gut the outsourcing model but rewires what buyers pay TCS to do.
What’s actually new
The number comes from arithmetic, not a job posting. Krithivasan told Reuters, “We would be … ensuring that we have as many as 1% to 1.5% of our associates who could be what you would call FDEs.” Applied to TCS’s end-June headcount of 593,798, that band works out to roughly 5,900 to 8,900 people. Crucially, he did not say whether these roles will be new external hires or existing staff retrained into the function, so this is not a hiring target. TCS spends about $1 billion a year on training, which makes internal conversion at least as plausible as recruitment.
One note on provenance: the “forward-deployed engineer” framing surfaced in the Reuters executive interviews, not in TCS’s official July 9 results release, so treat the label as attributed to those interviews rather than to the earnings call. NDTV Profit’s write-up carries the same detail.
The second signal is appetite for M&A. After growing almost entirely organically until late 2025, TCS is now hunting acquisitions in AI, data security and cybersecurity. CFO Samir Seksaria said the company is “looking at where we can find things which will help us enable or enhance our strategic positioning.”
What this means for enterprises buying AI integration
The FDE push is a direct answer to the fear that AI kills the outsourcer. Krithivasan’s argument is that TCS’s deep knowledge of messy customer environments is exactly what is needed to wire the multiple AI models clients now run into their existing systems. That is the same logic behind the enterprise buyer’s calculus on agentic AI: the hard part is not the model, it is integration and lock-in.
The financials show why TCS is moving now. In Q1 FY27 (quarter ended June 30, reported July 9), revenue hit $7,624 million, up 2.7% year over year and roughly flat sequentially, with net income near $1.46 billion and an order book of $9.5 billion, including an $800 million AI-led transformation deal with SKF. Annualized AI revenue reached $2.6 billion, but growth cooled to about 13.6% quarter-on-quarter from roughly 28% the prior quarter. Krithivasan targets around 25% sequential growth long-term and warns the path “won’t be linear.”
That deceleration matters more than the headline AI number, because margin, not growth, is the real scoreboard in AI services: operating margin already slipped about 130 basis points to 24.0% on wage hikes. The partnership signals point the other way, though. TCS this quarter became a premier partner of Anthropic with early Claude model access and 50,000 licenses, and Mistral’s first global systems-integrator partner. FDEs are how those licenses turn into billable delivery rather than shelfware.
What to watch
Three things. Whether TCS staffs the FDE cohort by hiring or retraining, which tells you how it reads its own talent bench. Whether AI revenue reaccelerates toward that 25% target or the slowdown persists. And what it buys: an acquisition in cybersecurity would confirm this is a capability grab, not just a headcount reshuffle. For an industry that spent 2025 fearing AI would automate it away, TCS is betting the opposite, the same reversal Altman and Amodei themselves walked back.
AI and business tech news, verified by a physician who reads the filings. One email a week, no noise.
