Ask a talent head in Bengaluru how AI hiring is going and you tend to get the same answer in different words. The requisitions are open. The budget cleared. The people are not there.
They are not imagining it. India’s talent shortage reached 82% in 2026, above the global figure of roughly 72%, and for the first time AI skills topped the list of what employers cannot find. AI model and application development ranked as the hardest capability to source at 39%, with general AI literacy close behind at 38%.
What makes this different from every other skills crunch the Indian tech industry has absorbed is that the standard fix does not work.
Why the old playbook stalls
For two decades the industry answered scarcity with volume. Hire large, train on the bench, deploy in six months. It worked for Java, for testing, for cloud migration, for most of what the services model was built on.
GenAI does not bend the same way. Nasscom and Deloitte India put the AI talent pool at 600,000 to 650,000 in 2022 and project it past 1.25 million by 2027, with the market itself compounding at 25 to 35 percent a year. Supply is growing. Demand is growing faster. And the roles that hurt most are not the ones a twelve-week programme fills. Someone who can fine-tune a model is findable. Someone who has shipped a retrieval system into production, watched it fail in a way nobody documented, and fixed it, is a different search entirely.
One widely quoted estimate puts it at roughly one qualified engineer for every ten open GenAI roles. Treat the precise ratio carefully, since methodology varies a lot across these studies. The direction is not in dispute.
The market is repricing in real time
AI hiring within Indian IT rose 16% year on year in June 2026 even as overall IT job listings fell 3%. Two lines moving in opposite directions, in the same sector, in the same month. That is the market in one sentence.
Pay followed. AI roles carry a wage premium starting near 20% at entry level and widening to about 38% for professionals with more than ten years behind them. Inside global capability centres the gap runs wider. Zinnov’s India GCC analysis shows capability centres paying 12 to 20 percent above IT services firms for comparable roles, and 30 to 50 percent more for AI and data specialists.
A good deal of this hiring has moved outside the traditional recruiting stack. Specialist firms have multiplied, and a growing number of Indian companies now work with an AI recruitment agency rather than routing these roles through a general talent function. The reasoning is practical. Assessing a GenAI engineer means knowing which claimed experience is real and which is vocabulary, and most internal recruiters have no way to tell the difference.
What is actually working
Build alongside buy. TCS trained around 350,000 employees and Wipro around 220,000 on AI capabilities during 2023-24. Those figures get quoted as press releases, but the logic underneath is sound. An experienced engineer who already understands the domain and then learns the tooling frequently outperforms an AI specialist who has never seen the business.
Hire adjacent. Strong data engineers, backend engineers with real production discipline, and analysts who have worked near ML convert faster than most teams expect. The mathematics is rarely the bottleneck.
Slow the loop down and make it concrete. Screening on certificates stopped meaning anything the moment certificates became easy to collect. Ask what the person built, then ask what broke and what they did about it.
Fix retention before you fix sourcing. Every AI engineer who resigns is effectively two open roles, because backfilling at today’s rates costs considerably more than keeping them would have.
The uncomfortable timeline
Industry estimates suggest it could take another three to five years to build an enterprise-ready AI workforce at scale in India. Some projections show a talent surplus by 2030. That may well come true. It does nothing for anyone holding a headcount plan for this quarter.
The organisations handling this sensibly have stopped arguing with the constraint. Fewer AI initiatives, staffed properly, instead of nine pilots each short two people. Less impressive as a slide. Considerably more likely to ship.
Author Bio
Nikhil Vaidya
Nikhil Vaidya is the CEO of Prism HRC, a leading recruitment services company in India. Nikhil’s expertise in talent acquisition and has been instrumental in connecting hundreds of top-notch clients with exceptional IT talent over the last 15 years.