40% of the work, 70% of the people: Cognizant's CFO on Indian IT jobs

Cognizant's chief financial officer has said in public, with numbers attached, what I first argued in 2017: that Indian IT's headcount-led model was brittle. Jatin Dalal said AI agents could take 30–40% of client work, and that if the company grows only 15% in five years its human workforce "will shrink". Most of those people are in India. A TCS study out the same week shows executives more uncertain than ever about jobs. My suggestion: a reskilling-and-placement duty on the firms that gain, run through a National Employment Grid.

I want to thank Mr Dalal for saying it plainly, and with arithmetic, instead of in the usual language of "workforce optimisation".

"If you believe we will grow only 15% in five years in aggregate, in that case, that human workforce will shrink. There is no choice around that." — Jatin Dalal, CFO, Cognizant, fireside chat with BMO Capital Markets' Keith Bachman on 12 November 2026, as reported by Mint on 13 November.

He was not confessing one company's difficulty. He was reading out the price list of a model I have been writing about for years. In Can Indian IT Re-invent itself? (3 February 2017) I quoted Business Line on an industry being forced towards "a business model that is less reliant on headcount", and added only this: the question is not "Can?" but "When?", and the answer is "Now". I had no number then. He has supplied one.

The figures, from Mint's report of 13 November:

• AI agents could handle 30–40% of client work; roughly 70% would still need humans, Mr Dalal said.
• Cognizant employs 351,600 people, almost three-fourths of them in India.
• Revenue in 2025 was $21.1 billion, up 7%. Guidance for 2026 has been cut to 4.5–4.9% growth.
• He called 2026 the most intense year for pricing in three years.

The sum needs no help from me. Subtract 30–40% of the delivery from a payroll that is mostly Indian, add mid-single-digit revenue growth, and the pyramid cannot keep its old shape. Mr Dalal's answer is to sell inference and outcome-based contracts rather than seats, and he expects inferencing to be the industry's future. That is sensible for margins. But a rupee of token revenue employs far fewer Indians than a rupee of billed effort.

The same week, the TCS Enterprise AI Study — 1,333 leaders across 25 countries, reported by TechSabado on 14 November — found something more interesting than another forecast:

• 84% name productivity as the chief value from AI; only 10% name revenue growth.
• Only 10% allow AI to execute decisions without human approval across more than one function.
• The share saying it is "too early to tell" what AI does to jobs rose from 4% in 2024 to 19%.

Capability is arriving faster than confidence. That is a small mercy and a warning together: it slows the substitution, and it tells us the window for reskilling is open today and will not stay open.

So, three things I would suggest, knowing well how slowly such things move here:

1. Attach a duty to the gain. Every AI-enabled saving sold to a client should carry a stated reskilling share, the way CSR spend is mandated, so that a firm which saves a salary funds part of that person's next skill.
2. Build the National Employment Grid I sketched in my whitepaper of 17 July 2025: a resume uploaded once, AI matching both sides every day, alerts on the phone — the same idea expanded in the Bharat Job Alert Mission two days later, where jobs find people rather than the other way around.
3. Make 40% a national number, not a company secret. Ask listed IT firms to disclose what share of delivery is agent-delivered, so the Ministry of Labour plans against data instead of anecdotes.

I may be too hopeful about speed. But a forecast this specific is a gift to anyone who has to plan. The Ministry of Labour no longer needs to guess at the future of white-collar work in India. It can read it off an earnings call.

With regards,

Hemen Parekh
www.Iam-Immortal.ai / www.IndiaAGI.ai / www.YourContentCreator.in
14 Nov 2026