India’s technology industry is estimated to have grown 6.1% in FY26, reaching $315 billion, while its headcount grew by about 2.3%, according to NASSCOM’s Annual Strategic Review 2026. Revenue is rising faster than the workforce that earns it. For a founder, that one gap holds both the opportunity and the problem, and which of the two it becomes depends on decisions that are still yours to make.

Consider a composite: a founder-director whose IT services firm is a little over six years old and employs 150 people. The firm has passed the usual early tests. Key clients stayed, the delivery team works, the accounts are healthy. Now the questions are changing. A long-standing client asks what AI has done to the cost of the work. A new graduate finishes in a week what used to take a month. A competitor announces an AI practice. None of these needs an answer today. All of them need a position.

What the Wider Picture Shows

The picture is mixed, and it is worth reading plainly. Crisil Ratings expects growth in IT services to stay muted at 1–3% over this fiscal and the next, citing AI disruption and weak client spending. It says AI-native solutions are intensifying pricing pressure and triggering deal renegotiations. Its analysis covers the 26 largest firms, so it does not describe a 150-person company directly.

The same release says mid-tier firms could prove to be nimble. And ICRIER, in a survey of 651 Indian IT firms, found no evidence of large-scale job losses, with the strongest demand growth in the technical roles AI touches most. The squeeze and the opening are arriving together. Five issues sit inside that picture, and each has a practical response.

Issue 1: The Pricing Conversation Is Coming to You

Most Indian IT services are still sold by time: people, months, a rate card. When AI shortens the time, a rate card hands the saving to the client automatically, and it arrives as a renegotiation. Crisil describes this pattern at the larger firms.

The workable response is to get there first. Choose one service line where the output can be measured, such as a defined testing package, a maintenance bundle or a migration, and offer it at a fixed price for a fixed scope. Keep part of the efficiency gain as margin and show the client the rest as a better outcome. The firm that proposes the new model sets its terms. The firm that waits is handed them.

Issue 2: The Shape of Your Team Is Changing

In the ICRIER survey, entry-level employment over the past 24 months fell at 55% of firms and rose at 35%. Mid-level employment rose at 42% of firms and fell at 25%. Senior employment was unchanged at 82%. The survey records the direction of change, not its size, and ICRIER calls the overall moderation modest. Among divisions most affected by AI, those where team size fell while output rose outnumbered the reverse by about 3.5 to 1.

The response is a deliberate decision about what a fresher is for. If routine tasks are now partly automated, the entry role can become an apprenticeship: real client work, reviewed by a capable mid-level colleague, with a clear path to the next grade. That makes the middle layer the one to invest in, since the people who can judge whether AI-assisted work is right are the ones the firm cannot do without.

Issue 3: Skills Are Being Built Slower Than Tools Are Being Bought

ICRIER found that only 4% of firms have trained more than half their workforce in AI-related skills, and it points to cost pressure, a shortage of qualified trainers and limited organisational readiness.

Training that sits outside the work rarely holds. A better pattern is to train inside live projects, starting with team leads. Pick one recurring task, record how long it takes today, introduce an AI-assisted method, and record the time again. That gives the firm its own evidence, not a vendor’s.

Issue 4: Adoption Is Still Mostly Reactive

NASSCOM’s Tech SME report (2025), based on interviews with founders and CEOs of more than 30 Indian tech SMEs, found that 64% adopt AI reactively, driven by client demand or internal efficiency, and 36% do so from a strategic vision. Over 80% named unclear use cases and uncertain returns as the main barrier, and only 32% keep a dedicated AI budget. The sample is small, so treat the figures as a signal, not a census.

Reactive adoption is understandable. It also means the firm’s AI direction is set by whichever client asked last. The response is to choose three use cases, each tied to an outcome a client already pays for. Give each an owner, a baseline measure, a budget line and a review date 90 days out. Three chosen use cases do more than ten tools bought one at a time.

Issue 5: Every One of These Needs an Owner

Pricing models, team structure, training plans and AI use cases all produce new decisions: which tools are approved, what client data may go into them, which pricing exceptions are allowed, which projects use AI-assisted methods. If every one of those still waits for the founder, the firm gives up the speed AI was supposed to provide.

The response is the same one that applies to growth in general. List the AI-related decisions that come up in a month, set a threshold below which someone other than the founder decides, and write down who that someone is. Our earlier article on the founder decision bottleneck sets out how to begin.

Five Issues at a Glance

A Word on Kerala

Kerala starts from a solid base. Technopark recorded software exports of ₹17,092 crore in FY 2025–26, up 17.3% from ₹14,575 crore, across 540 companies, according to official figures reported by IANS in September 2026. That is in rupees, so it is not directly comparable with NASSCOM’s dollar figures. It does show that the local ecosystem is growing, and a growing base is where well-timed decisions pay off most.

Where to Start

Diagnosis comes before prescription. Before choosing a tool or a vendor, find out which of the five is the real constraint in your firm. A founder usually has an instinct. A leadership team often holds three different instincts, and the disagreement itself is the first finding. Settling it is cheaper than buying the wrong fix.

The Real Question

Of the five, which one have you been postponing, and what would your firm gain if you decided it this quarter?

If you are not sure which of the five is holding your own firm back, a free 45-minute Organisational Health Conversation is a low-commitment place to find out.

Frequently Asked Questions

Will AI reduce the number of people an IT firm needs?

The evidence so far points to a shift in the mix more than a sharp fall in numbers. ICRIER found a modest, broad-based moderation in hiring, concentrated at entry level. Looking two years ahead, 27% of the surveyed firms expect headcount to fall, 28% expect it to rise and 44% expect no major change. These are expectations, not outcomes.

Should a 150-person firm move all its work to outcome-based pricing?

No. Start with one service line where output is easy to define and measure, learn from it, and extend only where the model works.

How much should a firm spend on AI?

There is no reliable benchmark for a firm of this size. NASSCOM’s SME report found that most firms allocate only 4–9% of their technology budgets to AI, and that is a description of current practice, not a target. Tying spend to three named use cases gives a better basis than a percentage.

Does this apply to firms outside Kerala?

Yes. The data above is national. The Kerala points are about the local starting position, not a different set of issues.