I trace the balance sheet, not the press release. When R Srikrishna, CEO of Hexaware Technologies, told the world that AI could slice 25% off IT project costs, he wasn't making a prediction. He was issuing a defensive memo. The statement, picked up by Crypto Briefing and other outlets, is a masterclass in narrative management—but beneath the optimistic headline lies a structural confession: the traditional IT services model, built on the arbitrage of human labor, is cracking. The 25% figure is not the story. The story is what that number reveals about the industry's inability to adapt without cannibalizing itself.
Hype is the only asset in a vacuum mint. In the IT services vacuum, the hype is "AI efficiency." But unlike a crypto whitepaper, we can audit the claims. The core question isn't whether AI can reduce costs—it can. The question is who captures that value, and at what human cost. This is not a technology story. It is a story about the redistribution of margin and the quiet erasure of the entry-level workforce.
The Cost Structure Confession
Let's start with the math. IT services projects allocate 60-80% of their budget to human labor. Coding, testing, documentation, and project management are the four pillars of delivery. AI tools, whether generative coding assistants or automated testing suites, demonstrably reduce the time spent on these tasks. McKinsey and Gartner have projected efficiency gains of 20-40% in certain development workflows. So, a 25% cost reduction claim is not implausible. It sits comfortably within industry consensus.
But here is the tension the CEO's statement glosses over. If you reduce the cost of delivery by 25%, and your revenue model is based on time-and-materials billing, your revenue also shrinks by 25%. The margin remains flat, but the absolute dollars decline. This is the core contradiction of AI adoption in legacy IT services. Srikrishna mentioned the need to "balance productivity gains with pricing strategy to maintain margins." That is corporate code for: we are about to face relentless pressure from clients to share the efficiency dividend, and we are not sure we can survive it.
I have seen this dynamic before. In the 2020 DeFi summer, I watched protocols offer absurd yields to attract liquidity, all while ignoring the structural fragility of their collateral models. The collapse was inevitable because the incentives were misaligned. Here, the misalignment is between the promise of efficiency and the necessity of revenue preservation. The CEO is trying to have it both ways: embrace AI to appease investors, while signaling to clients that prices will drop to retain their business. That is a fragile equilibrium.
The Pricing Paradox and the Power Shift
When a CEO publicly quantifies a cost reduction, they are not just informing the market. They are arming their clients with a negotiating lever. Every CIO reading that statement will now demand a discount. The 25% number becomes a baseline for renegotiation. This is a strategic error, unless the intent is to preemptively reset expectations. By saying "AI could cut costs," Hexaware is admitting that the value of their existing workforce is about to depreciate.
The balance of power in the client-provider relationship is shifting. Clients are no longer dependent solely on the provider's talent pool. They can deploy AI tools internally for routine tasks. They can use AI-native platforms for code generation. The moat of "we have 30,000 engineers ready to serve you" is evaporating. The new moat is either proprietary AI integration or a willingness to move to outcome-based pricing.
Hexaware, a mid-tier player with roughly 30,000 employees and annual revenue in the low billions, is caught in a squeeze. The giants—Accenture with its $3 billion AI investment, Infosys with its Topaz platform, Wipro with its $1 billion AI commitment—are building scale in AI capabilities. Meanwhile, AI-native startups are attacking from below, offering specialized tools that replace entire swaths of junior developer work. Hexaware has neither the capital of the giants nor the agility of the startups. Its CEO's statement is a signal to the market: we see the threat, and we are positioning ourselves as AI-adapters.
The Human Cost Ledger
Let's be cold about the human element. The IT services industry, particularly in India, has been the primary engine of the country's white-collar middle class. TCS, Infosys, Wipro, and HCL collectively employ over a million people. Hexaware adds tens of thousands more. The 25% cost reduction, if realized across the industry, does not just disappear from a ledger. It represents the elimination of a corresponding percentage of billable hours. That translates to fewer hires, frozen promotions, and significant layoffs at the entry level.
IDC projects that by 2028, AI will replace 15-20% of junior programmer roles globally. The IT services industry is the canary in the coal mine for the broader knowledge economy. If a junior developer's job is to write boilerplate code, run tests, and fix bugs—all tasks that AI is increasingly competent at—then the demand for that role collapses. The industry's historical reliance on a pyramid structure, with thousands of low-cost juniors supporting a few high-cost seniors, is inverted. AI compresses the base of the pyramid.
The CEO did not mention this. The press release did not mention this. But the forensic reality of the cost model demands it. You cannot cut project costs by a quarter without cutting the people who deliver the projects.
What the Bulls Get Right
I am not here to argue that AI will not transform IT services. It will. And there is a compelling counter-argument to my skepticism: the industry has survived paradigm shifts before. The move from on-premise to cloud, the rise of agile methodologies, the offshoring wave—each was supposed to be the death knell, and each actually expanded the addressable market. AI may follow the same pattern.
Lower delivery costs could unlock demand for software that was previously cost-prohibitive. A company that could not justify a $2 million custom application might invest $500,000 if AI makes it feasible. This is the Jevons Paradox applied to software development: as the cost of a resource falls, its consumption increases. The total market for IT services could grow, even as the cost per project declines. The bulls argue that Hexaware's 25% figure is not a threat to revenue but a catalyst for volume.
They also point to the opportunity for outcome-based pricing. If AI enables a provider to deliver a project for $750,000 instead of $1 million, the provider can structure a deal where they take a percentage of the client's resulting revenue or savings. This aligns incentives and can generate higher margins than the old time-and-materials model. The provider becomes a partner, not a vendor. This is the genuine upside, and it is the only path to sustained profitability in the AI era.
The contrarian view, which I hold with some measured respect, is that the transition will be brutal but ultimately value-accretive for the winners. The problem is identifying the winners. It will not be the companies that simply adopt AI tools. It will be the companies that fundamentally restructure their pricing, their talent mix, and their client relationships.
The Accountability Imperative
When the yield is too high, the exit is rigged. When the cost-cut claim is too clean, the hidden costs are buried. The market needs to demand granularity. I want to know which specific components of the project lifecycle are seeing the 25% reduction. Is it in testing? Documentation? Code generation? Each has a different AI substitution rate and a different quality risk. I want to know if Hexaware has actually realized this on a live project, or if it is a projection from an internal model. A projection is not a fact. An audit is.
The industry also needs to address the accountability vacuum regarding workforce displacement. The CEOs are happy to tout efficiency gains to investors, but they are silent on the social cost. This is not a call for protectionism. It is a call for honest accounting. If 25% of the billable hours are erased, what happens to the 25% of the workforce? Are there retraining programs? Are there severance packages that reflect the profits AI is generating? The silence is deafening.
The Verdict on the Ledger
A profile picture is not a shield against fraud, and a headline number is not a shield against structural decline. The 25% claim is a useful data point, but it is not the truth. The truth is that the IT services industry is entering a period of violent margin redistribution. The value that was captured by labor is shifting to capital (the AI tool providers and the shareholders of firms that adapt quickly) and to clients (who will demand lower prices).
My analysis, based on a decade of watching technological narratives mask economic realities, suggests that the medium-term winners will be the AI-native tool providers like OpenAI, Anthropic, and the coding assistance platforms. The short-term losers will be the junior developers. The uncertain variable is the mid-tier IT services firms. They are too big to pivot quickly and too small to out-invest the giants. They will survive, but only by shedding their legacy model.
The forward-looking question is not whether AI will cut costs. It is whether the industry can rebuild its social contract. Can it train its displaced junior staff to become AI supervisors? Can it shift to pricing models that share risk and reward? If not, the 25% figure will be remembered not as a mark of progress, but as the moment the industry signed its own downsizing order. I will be watching the on-chain data of quarterly earnings, not the press releases, to see who actually survives. The exit is rigged for those who refuse to change the game.