An engineer with eight years of experience sits down for a review having heard that someone hired last quarter, comparable tenure, different job title, sits on a materially higher band. That conversation is happening across Indian tech right now, and almost no compensation framework in the country was designed for it.
The gap is well documented by now. Salary reporting in 2026 put AI roles at roughly ₹36 lakh against ₹26 lakh for comparable IT positions, a premium of about 38% at senior levels. It begins nearer 20% for entry-level hires and widens with experience. In BFSI, where a 42% skill gap in AI and data roles has been reported, some organisations are paying 1.5 to 2.5 times their normal bands to land specific people.
Global capability centres are pulling hardest. Zinnov’s India GCC work shows capability centres paying 12 to 20 percent above IT services firms for equivalent roles, and 30 to 50 percent more for AI and data specialists. Nearly 64% of newly created GCC roles now require AI, data analytics or automation expertise, concentrated in the four-to-ten-year experience band. That happens to be exactly the population Indian services firms rely on for delivery leadership.
The equity problem nobody wants to name in a meeting
Bands exist to keep pay defensible. When the market price for one skill moves 40% in eighteen months and the band moves eight, either the band breaks or the hiring stops. Most companies quietly broke the band and hoped it would not travel.
It travels. Salary information in Indian tech moves faster than any internal communication plan ever built, and it moves through WhatsApp groups that HR is not in.
Three ways companies are handling it
Separate the structure openly. Some organisations have built distinct bands for scarce skills with published criteria for what qualifies. Not elegant, but transparent and workable. The criteria matter far more than the structure, because an unexplained exception reads as favouritism regardless of how justified it was.
Pay in composition rather than base alone. Retention-linked components, delivery bonuses tied to specific outcomes, and equity where it exists move the discussion away from one number on offer day. It also gives you somewhere to go when the market cools.
Plan for the premium to expire. A scarcity premium paid as permanent base salary sits on the payroll long after the scarcity ends. Cloud engineers commanded something similar a few years ago and that gap narrowed considerably. Nobody clawed anything back, and nobody was going to.
Vendors selling AI recruitment solutions have multiplied alongside the shortage, and the better ones bring salary benchmarking that is more current than any annual survey can be. Ask what the benchmark actually sits on and how recently it moved. A number built from last year’s closed offers is describing a market that no longer exists, and in this particular market twelve months is a long time.
What the premium is really pricing
Two different things get conflated here and it is worth pulling them apart.
Some of the premium reflects genuine scarcity. People who have run models in production, at scale, under real latency and cost constraints, and who have debugged something that failed in front of customers. There is no shortcut to producing that person and there is no training budget that manufactures the experience.
The rest is anxiety. Organisations that decided they needed an AI team before deciding what the team would do, bidding against each other for the same few hundred profiles. That part corrects. It always corrects. Companies that overpaid for AI capability without a clear mandate will find out in a year or two that they bought expensive headcount for a project no executive ever really sponsored.
The question worth asking internally
For HR leaders the useful question is not whether the premium is fair. Markets are not fair and negotiating with one rarely goes well. The question is which of your roles genuinely sit in the scarce category, and which got labelled AI because it made the requisition easier to push through an approval chain.
In plenty of organisations that second group is larger than anyone will say out loud. Sorting it honestly saves more money than any negotiation tactic, and it stops you paying a scarcity premium for work that is neither scarce nor, if you are honest about it, particularly AI.
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.
