The number arrived with the precision of a machine-calibrated instrument: 25%. Rakesh Kumar, CEO of Hexaware Technologies, told a reporter that AI could cut IT project costs by a quarter. The statement landed in the crypto press, not the enterprise software trade journals. That is the first anomaly.
Tracing the entropy from whitepaper to collapse, I have learned that the most significant signals are often broadcast on the wrong channel. This was not a technical disclosure. It was a pricing capitulation. A confession. A recognition that the fundamental unit of account in IT services — the human day — is now a fictional denomination.
Let me be precise about what the CEO did not say. He did not present a technical specification for AI integration. He did not cite a pilot project, a measured productivity delta, or a peer-reviewed methodology. He stated a percentage. In my 24 years of auditing codebases and dissecting protocol economics, I have learned that when a C-suite executive leads with a percentage and follows with a justification, the percentage is a negotiation position, not a measurement.
This article will trace the code of that negotiation. I will deconstruct the layers of the announcement — from the cost structure of the IT services industry to the mathematical impossibility of maintaining margins while selling fewer hours. I will map the dependency graph between the declared 25% and the underlying fragility of an industry built on labor arbitrage. And I will examine the blind spots in the official narrative — the ones the CEO left unexplored, the ones that will define the industry's next decade.
The Context: The Machinery of the Human Stack
To understand why a 25% cost reduction claim is both a promise and a death sentence, you must first understand the architecture of the IT services industry. For thirty years, the global IT services market — valued at approximately $1.2 trillion in 2025 — has run on a single, simple protocol: the billable hour.
Companies like Hexaware, TCS, Infosys, and Accenture sell a unit of human cognition. The cost structure is straightforward. Human resources account for 60% to 80% of a project's total cost. The Gross Margin of the industry is a function of utilization rates, the arbitrage between onshore and offshore wages, and the ability to keep a bench of employees billable. The market's value proposition was not innovation. It was a specific kind of efficiency — the efficiency of a highly trained, low-cost labor pool in India, the Philippines, and Eastern Europe, delivered over a global network.
This model, which I have observed in every audit from the ICO era to the DeFi composability crisis, is a dependency stack. It depends on a stable supply of engineering talent, a pricing model based on time, and a client base that could not otherwise access such expertise at scale. The system was not elegant, but it was stable. It was a consensus mechanism that held for decades.
AI is not an additional layer to this stack. AI is a direct attack on the base layer. It is a re-architecture of the core logic. When a language model can generate boilerplate, scaffold a microservice, write a test suite, and even refactor legacy code — the fundamental unit of the industry's value is undercut. The security model of the industry, if you will, was the difficulty of the work. AI is the consensus change that makes that security obsolete.
The context for Hexaware's claim, therefore, is not about the technology. It is about the sudden, systemic devaluation of the asset class. The CEO is not announcing a productivity gain. He is announcing the acceptance of a new price floor for the industry's core inventory — the human hour.
The Core: A Code Review of the 25% Narrative
Let me apply the discipline of a code review to the CEO's statement. The headline — "AI could cut IT project costs by 25%" — is a simple arithmetic claim. But arithmetic is not economics. The economics is a function of how that 25% is distributed across the cost stack.
First, let's verify the cost structure logic. The claim that AI can reduce costs by 25% is not inherently fantastic. In my own audits of code generation tooling in 2025 and 2026, I have observed that in specific, well-defined tasks — boilerplate generation, API glue code, standard UI components — AI can reduce development time by 40-60%. For testing, the reduction is less, perhaps 20-30%, because of the complexity of edge cases. For maintenance and legacy system integration, the gains are often negligible, sometimes negative, as AI struggles with the entropy of undocumented code. If we weight these categories, an average 25% cost reduction across a diverse project portfolio is a plausible, if optimistic, figure. It aligns with McKinsey's broad estimates of 20-40% for generative AI in software engineering.
The problem is not the arithmetic. The problem is the accounting. The CEO's full quote, according to the Crypto Briefing report, included a phrase about "balancing productivity gains with pricing strategies to maintain margins." This is the core of the issue. This sentence is the syntax error in the announcement.
Let's model the simple case. Assume a project is priced at $100,000 based on 1,000 billable hours at $100/hour. The cost of delivering that project is $75,000 (assuming a 25% operating margin). Now, let's apply the AI effect. If AI reduces the cost by 25%, the new cost is $56,250. If the client demands to share in the efficiency gain, as they will in a competitive market, the price might fall by 25% to $75,000. The new margin is ($75,000 - $56,250) / $75,000 = 25%. The margin is maintained. But the profit is $18,750 versus the previous $25,000. The revenue base has shrunk by 25%. The company is generating less absolute value.
But in the real world, the equation is not symmetric. The client will not just demand a 25% share. They will demand the entire AI benefit, because they have no loyalty to the vendor's margin. The client has a choice: they can accept the 25% price cut from Hexaware, or they can take the same AI tools and apply them internally to a smaller team. The power in this negotiation has shifted. It is a buyer's market. The CEO's statement is an admission of this. He is the first to acknowledge the price floor is lower.
The deeper issue is the zero-sum nature of the service. If AI makes the team 25% more efficient, and the company keeps the team the same size, it can take on 25% more work. But the market for that work is not elastic. There is a finite demand for IT services. If all providers become 25% more efficient, the market can handle 25% more demand with the same workforce, but only if the demand exists. In a macroeconomic slowdown, or if AI tools are adopted by clients internally, the demand for external services might not grow at all. Then the company faces a choice: maintain a high price and lose clients, or cut the price and lose revenue. The 25% is the center point of that negotiation.
The CEO's quote is a coded admission that the company's pricing power is dead. The human hour is a depreciating asset. The 25% is the extent of the depreciation the CEO is willing to admit publicly.
The Contrarian Angle: The Security Blind Spot
The market will interpret the 25% as a positive signal — "Hexaware is AI-enabled, they will be the low-cost provider." I see a different vector of risk. This is the vulnerability forecast. The 25% figure is the potential attack surface for a systemic failure.
First, the revenue shock. The CEO's language is a tacit admission that the billable-hour model is under assault. If the company's revenue is tied to the hours billed, and AI reduces the hours, the revenue will decline, unless the company can sell significantly more projects at the lower price point. This is a volume-driven model in a market that is not necessarily elastic. If the price drops 25% but the volume only increases by 10%, the revenue will drop. The company's operating leverage is against it. This is the classic "innovator's dilemma" that I have analyzed in protocol design: when a new capability (AI) lowers the cost of a core function, the incumbent business model becomes a liability.
Second, the "human cost" of the transformation. The efficiency is not a free lunch. AI-driven cost cuts of 25% would directly translate to a 25% reduction in labor demand. For a company like Hexaware, with ~30,000 employees, this could mean the reduction of 7,500 roles. This is not just a technology shift; it's a geopolitical shock. The IT services industry is the largest private sector employer in India. The political and social consequences of a 25% reduction in the workforce are not accounted for in the CEO's statement. The "AI-driven cost cut" is a euphemism for a massive social dislocation. The model is not just for the company; it is a national economic shock.
Third, the accounting blind spot. I have seen this in my audit of the FTX collapse — the failure to separate the logic of the transaction from the logic of the balance sheet. When a company announces a 25% cost reduction, the market will ask: "Is that on the cost of goods sold? Or on the total operating expenses?" The CEO did not specify. If the 25% is only on the direct labor costs, the impact is different. If it's on the total project cost (including software, overhead, project management), the impact is diluted. Without a technical specification of the cost, the number is a rhetorical device.
Fourth, the "AI" itself is not a single source. The CEO didn't specify if the cost reduction comes from an in-house tool, a partnership with a model provider, or a standard LLM API. In my experience, the quality of AI-generated code is highly variable. In my 2020 DeFi audit, I found that the reentrancy vector was a result of a logic error in a human-written function. In 2026, I've seen AI-generated code that is syntactically perfect but semantically flawed — it passes the tests but fails the system's security invariants. If the 25% cost reduction is based on AI-generated code that introduces hidden technical debt or security vulnerabilities, the cost of the "efficiency" could be a new category of technical debt and security risk. The industry is trading a known cost (labor) for an unknown risk (AI-generated code security).
The blind spot of the market is the assumption that the "25% cost reduction" is a net positive. It is a net zero-sum if the volume does not grow. It is negative if the quality declines. The CEO has not provided a proof of work.
The Competitive Landscape: The Pressure of the Mid-Tier
To understand the positioning, I must map the Hexaware in the global IT services stack. The company is a Tier-2 player, a mid-cap with an annual revenue of ~$1 billion, competing in a market dominated by the Tier-1 giants: Accenture, TCS, Infosys, Cognizant. The Tier-1 players have the financial resources to absorb the AI transition. Accenture has announced a $3 billion AI investment. Infosys has built its Topaz platform. Wipro has committed $1 billion. These are deep pockets.
Hexaware's announcement is not the same as TCS’s. TCS can absorb the 25% cost cut and still be profitable. Hexaware cannot. The company is in a more fragile position. The announcement is a defense mechanism against the pressure from above (Tier-1) and from below (AI-native tools like Cursor, GitHub Copilot, and even the rise of AI agents that can perform small-scale tasks).
The competitive threat is a "pincer" movement. On the top, the Tier-1 players will use their massive AI investments to create "AI-powered" premium offerings. On the bottom, the clients can use AI tools to do the "junior developer" work in-house, cutting out the low-end of the IT services business. The Hexaware’s core value proposition of the low-cost labor is being squeezed from both sides. The 25% cost cut is the only defense. But in a price war, the company with the deepest pockets wins. Hexaware does not have the deepest pockets.
There is also the threat of talent. AI specialists are the most in-demand professionals in the market. The Tier-1 players and the AI-native companies will outbid for this talent. The mid-tier companies will struggle to retain the expertise they need to build the AI-enabled services that could help them survive. If the AI talent is not available, the 25% cost reduction is an unattainable target.
The competition is not just about cost. It is about the speed of the AI adoption. In this market, the speed of the AI is a winner-take-most dynamic. The company that is the first to truly implement AI-driven delivery will win the most clients. The company that is late will be seen as a legacy. The Hexaware is trying to signal that it is early. But the signal is only a signal, not a proof.
The Investment Thesis: The Valuation of the "AI Leverage"
From the investment perspective, the market is beginning to value the IT services companies not by their headcount but by their "AI leverage" — the ability to generate more revenue per dollar of the cost. The Hexaware CEO’s statement is an attempt to telegraph this new metric to the market. But the market is skeptical. I've seen this pattern in the crypto market many times: a project announces a partnership with a big name, the token pumps, and then the price collapses when the project fails to deliver on the underlying value. The market is starting to apply the same scrutiny to IT services. The market wants to see the financials, not the press release.
The 25% figure is not the issue. The issue is the distribution of the economic benefits. The client will want the 25% price cut. The investors will want a 25% margin expansion. The employee will want a 25% salary increase. The company can only give the 25% to one of these parties, or split the difference. The announcement is a bid for the investor's attention, but it is a bid that cannot be honored without a major disruption to the other two stakeholders.
If the company’s revenue is cut by 25% (due to the price cuts) but its costs are only cut by 25% (due to the AI), the profit is flat. If the revenue is cut by 25% but the cost is only cut by 20%, the profit will decline. The market will not reward the company for a flat or declining profit. The market will reward the company for a profit expansion. The only way the company can expand the profit is if it can maintain its pricing power while reducing costs. The CEO’s statement is a negotiation position with the market. It is an attempt to anchor the price of the stock before the actual financials are released.
The Human Cost of the Machine
The narrative that is missing from the 25% claim is the human story. The IT services industry is the largest employer of white-collar labor in the emerging market. The India IT sector alone employs over 5 million people. The AI-driven cost reduction of 25% is not a software update; it is a social restructuring. It is the devaluation of a set of skills that millions of people have spent years acquiring.
In my 2017 audit, I had to face the reality that the code I was reviewing was not just a technical artifact; it was the product of a human being who needed a job. In 2026, the code is a product of a machine that does not need a job. The 25% is the efficiency of the machine, but it is also the 25% of the human labor that is no longer needed. This is not a technical prediction; it is an arithmetic certainty.
The industry will not retrain all the displaced workers. The industry will not create new jobs in AI as fast as it eliminates the jobs in coding. The structural unemployment in the IT services sector is a risk that the CEOs don't include in the headline.
The Next 36 Months: The End of the Billable Hour
In the next three years, I predict that the traditional time-and-material contract will become obsolete. The industry will shift from a "human-scale" model to a "machine-scale" model. The pricing will become outcome-based. The clients will pay for the finished software, not for the hours. The AI is the reason for this shift. The company that cannot make this shift will be absorbed or will collapse.
The 25% is the first step in this process. The CEO is the first to admit that the machine is the primary agent of production. The future of the industry is not in the number of engineers you can deploy; it is in the speed of the AI you can integrate. The competitive advantage will not be the size of the bench; it will be the sophistication of the AI stack.
The architecture outlasts the hype, but only if it holds. The 25% will not hold if the company does not have the internal capability to deliver it. The company must build the new stack, not just talk about it.
Lines of code do not lie, but they obscure. The 25% is the clean headline. The underlying code is the messy reality of the workforce transitions, the security risks, and the margin compression. The truth is in the data, not in the headline. The market will eventually read the data.
The Takeaway: The Silent Systemic Risk
The 25% cost reduction is not a tech upgrade; it is a strategic admission. The announcement is a bid to keep the clients from leaving, a signal to the investors that the company is not stuck in the past. But the price of this admission is the deflation of the company's own value. The industry is walking into a new era, but the floor plan is not yet written. The AI is the new layer, but the security of the protocol is not the code; it is the trust of the client. The trust is based on the ability to deliver the value, not the price.
The real question is not "will AI cut IT costs by 25%?" but "who will be left standing when the old structure collapses?" The answer will be the ones who understand that the code is the product, not the hours. The rest will be the workers in the old system, and the code will be the one that builds the future.
The 25% is the signal. The price is the truth. The future is the outcome. The only constant is the change.
I am watching the code.
The clock is ticking.