Hook: The Number That Means Nothing
Twenty-five percent. That is the number Hexaware's CEO offered the market last week when asked about AI's impact on IT project costs. A clean, round, media-friendly figure. It aligns neatly with McKinsey's projections. It matches Gartner's range. It is, on its face, entirely plausible.
That is precisely why it deserves suspicion.
In my eighteen years auditing technology claims โ from ICO whitepapers to Layer-2 fraud proof mechanisms โ I have learned that the most dangerous numbers are the ones that sound reasonable. The 25% figure is not a technical finding. It is a strategic communication. And the gap between what it says and what it means is where the real story lives.
Context: The Business Model Under Siege
The global IT services industry is a $1.2 trillion market built on a single, fragile premise: that human labor, measured in billable hours, is the unit of value delivery. Indian IT services firms โ TCS, Infosys, Wipro, HCL, and mid-tier players like Hexaware โ have spent three decades perfecting the art of labor arbitrage. Hire inexpensive engineers in Bangalore, bill them out at Western rates, and pocket the spread.
Labor costs constitute 60-80% of project expenses. The entire pricing architecture โ time-and-material contracts, fixed-bid engagements, managed services retainers โ assumes that headcount equals output.
AI breaks that assumption at the foundation.
When a language model can generate boilerplate code, write test suites, and draft documentation in seconds, the billable hour loses its meaning. The industry's core metric โ the "person-month" โ becomes an anachronism. This is not a marginal efficiency gain. It is a structural dislocation of the industry's entire economic logic.
Core: The Cost Reduction Paradox
Let me walk through the arithmetic, because the CEO's statement contains a contradiction that nobody in the coverage has flagged.
If AI reduces project costs by 25%, and the company bills on a time-and-materials basis, then revenue also falls by 25%. The margin โ the ratio of profit to revenue โ remains unchanged. The company simply does less work for less money. That is not a transformation. That is a shrink.
For the 25% cost reduction to translate into improved profitability, one of two things must happen. Either the company shifts to outcome-based pricing, capturing the efficiency gain as margin rather than passing it to the client. Or it maintains pricing while reducing delivery costs โ a position that requires the client to be unaware that AI has made the work cheaper.
The first path requires a fundamental renegotiation of the client-vendor relationship. The second is a temporary fiction. Clients are not stupid. They read the same McKinsey reports. They know what AI can do. The CEO's statement โ "balancing productivity gains with pricing strategy to maintain margins" โ is an admission that the company expects to be forced into sharing the efficiency dividend.
This is the core tension that the 25% headline obscures. The real question is not whether AI cuts costs. It is who captures the value of those cuts. And on that question, the IT services industry is structurally disadvantaged.
Consider the power dynamics. In the traditional model, the IT services firm holds information asymmetry โ the client does not know how much effort a project truly requires. AI dissolves that asymmetry. When a client can generate a functional prototype with Cursor or Copilot in an afternoon, the vendor's mystique evaporates. The negotiation shifts from "how many engineers do you need" to "why do I need you at all."
This is the efficiency-ethics friction that defines the moment. The industry's historical profitability was not a reward for technical excellence. It was a rent extracted from information opacity. AI is the audit that exposes the ledger. And ledgers do not lie, only their auditors do.
Contrarian: The Blind Spot Nobody Is Discussing
The coverage of Hexaware's announcement has focused on the obvious questions: Is 25% realistic? Will margins compress? Which companies are best positioned?
Nobody is asking the question that matters most: what happens to the 2.5 million people employed by Indian IT services firms?
The industry's entire employment pyramid is built on a broad base of entry-level engineers performing routine coding, testing, and maintenance. These are precisely the tasks that AI automates first. IDC projects that AI will eliminate 15-20% of junior programmer roles by 2028. For India's IT sector โ which employs roughly 5 million people directly and supports an estimated 15 million indirectly โ this is not a headcount adjustment. It is a social and political event.
The CEOs who tout AI-driven cost reductions are, in effect, announcing the obsolescence of their own workforce. The 25% figure implies a corresponding reduction in labor hours. That means hiring freezes, layoffs, or โ at best โ a massive reskilling effort that the industry has no track record of executing successfully.
There is a deeper irony here. The Indian IT services model succeeded because of its human capital advantage. The country produces hundreds of thousands of engineering graduates annually, creating a deep, low-cost talent pool. AI does not just reduce the cost of labor. It reduces the strategic value of that labor pool. The industry's comparative advantage โ cheap, abundant, skilled workers โ is precisely what AI commoditizes.
This is the blind spot in every optimistic AI-transformation narrative I have audited. The same logic applies to crypto protocols that promise to "democratize" finance while concentrating power in validator cartels, or to DAOs that claim decentralization while governance tokens concentrate in founding teams. The stated benefit is efficiency. The unstated cost is structural displacement. Code is law, but human greed is the bug.
Takeaway: The Vulnerability Forecast
The 25% figure will be validated or falsified within eighteen months. The signal to watch is not Hexaware's marketing materials but its quarterly earnings โ specifically, the relationship between revenue per employee and operating margin. If the company achieves cost reductions while maintaining revenue, the transformation is real. If revenue falls in lockstep with costs, the 25% was simply a pre-emptive concession to inevitable client pressure.
My forecast: the latter outcome dominates. The IT services industry will experience a period of margin compression as clients capture the AI dividend, followed by a consolidation wave in which mid-tier players like Hexaware become acquisition targets for larger firms seeking AI capabilities and client relationships.
The deeper question โ the one that will define the industry's next decade โ is whether IT services can evolve from selling labor to selling outcomes. That transition requires a fundamental rethinking of pricing, delivery, and value creation. It requires building bridges in the storm, not after the rain.
The 25% number is not the story. The story is what happens to the people, the pricing models, and the industry structure when that number becomes real. Yield is the interest paid for ignorance. And the IT services industry has been collecting interest on its ignorance of AI's true impact for far too long.