Part I: The Number That Says Nothing
Ravi Kumar S., the CEO of Hexaware Technologies, made a statement that would have been unremarkable a decade ago but now carries the weight of a structural confession: "AI could cut IT project costs by 25%."
The number is precise. It is plausible. It aligns with McKinsey's broad estimates. And it is entirely meaningless without context.
Here is what the CEO did not say: which stages of the software development lifecycle—the 25% cost reduction comes from. He did not specify whether this figure comes from live production environments or internal models. He did not disclose how the savings will be distributed between his company and its clients. He did not address the most critical question of all—whether a 25% reduction in cost will be accompanied by a 25% reduction in revenue, thereby leaving margins exactly where they started.
I have spent the better part of the last decade dissecting protocol architectures and auditing smart contracts, and I have learned one lesson that applies universally across every technology sector: when an executive quotes a number without disclosing the underlying model, that number is a marketing artifact, not a technical metric. The architecture of trust in a trustless system begins with the recognition that numbers without provenance are just creative writing.
The 25% figure appears at the intersection of several conflicting incentives. It is large enough to demonstrate to investors that Hexaware understands the AI revolution. It is modest enough to avoid triggering clients to demand even steeper price cuts. And it is vague enough to escape any possibility of verification. This is not a projection. It's a strategic signal.
The signal is this: the IT services industry, a global market worth approximately 1.2 trillion dollars and employing millions of knowledge workers, is now openly acknowledging that its business model—selling human hours at a markup—is facing an existential threat from the very technology it claims to deploy. When a company's CEO describes the way to cut your own costs, you are not hearing an announcement of progress. You are hearing a euphemism for adaptation to the market that no longer needs your core product in its previous form.
Part II: The Architecture of the Industry
To understand why this statement matters, we must examine the fundamental architecture of the IT services industry. This is a sector whose entire economic structure is built on a simple principle: time arbitrage.
The traditional model works something like this: a client in New York or London requires custom software, maintenance, or digital transformation services. The client hires an IT services firm. The firm assigns a team of developers, many of whom are located in India, where the average salary of a skilled engineer is a fraction of the cost in the West. The client is billed by the hour or by the "man-month," a unit of time that has become the industry's primary currency.
This model has proven remarkably resilient for three decades. It created the Indian IT juggernaut—TCS, Infosys, Wipro, HCL Technologies, and a second tier of players including Hexaware. These companies built their success on the back of a simple observation: the quality of code produced by Indian engineers could be equal to or better than Western engineers, at a fraction of the cost. The resulting margin—the difference between billing rates and labor costs—created enormous profits and funded the growth of a middle class across India.
The entire system is built on a fundamental assumption: that human labor is the scarce resource. That assumption has now been violated by the emergence of large language models capable of generating, reviewing, and refactoring code with reasonable accuracy and at near-zero marginal cost.
When the CEO of Hexaware says AI could reduce project costs by 25%, he is implicitly admitting that the fundamental input of his industry—human hours—is now being replaced by a cheaper alternative. This is not a cost reduction. It is a recognition that the raw material of the IT services industry has been commoditized.
But here is the problem the CEO did not mention. When AI reduces the cost of delivering an IT project by 25%, it also reduces the amount of human labor required to deliver that project. The price charged to the client is not necessarily determined by cost. It's determined by perceived value. In a competitive market, however, if your competitor can deliver the same project at 25% less cost, they can charge a lower price to the client—or keep the same price and pocket the margin. Either way, the industry's revenue pool for a given amount of work is about to shrink.
This is the fundamental tension of AI-driven cost reduction in a service industry. Efficiency gains do not automatically translate into profit gains. They translate into price pressure.
The CEO's phrase "balancing productivity gains with pricing strategies to maintain margins" is the closest thing to a confession we will get. It is the IT services equivalent of a store manager saying, "We'll need to figure out how to sell the same goods at lower prices without losing money."
Part III: Where the 25% Goes
Let me take you through what happens to that 25% in the real world of software delivery. As someone who has audited codebases, analyzed gas optimization in smart contracts, and watched the operational realities of software development, I can tell you where the fat is in the system.
The cost structure of a typical IT project breaks down roughly into these phases: requirements gathering and architecture (10-15%), development (40-50%), testing (20-30%), deployment and maintenance (10-20%). These percentages vary by project type, but the pattern holds.
Now let's apply the AI lens.
Requirements and architecture. AI can assist in this phase by analyzing existing codebases, documenting legacy systems, and generating preliminary architecture recommendations. But this phase requires deep business context and human judgment. AI's impact here is real but limited—perhaps 10-15% efficiency gains.
Development is where AI shines. Code generation tools like GitHub Copilot, Amazon CodeWhisperer, and more specialized systems can generate boilerplate code, implement standard functions, and even debug common errors. For junior-level tasks—CRUD operations, API endpoints, data transformations—AI can reduce effort by 30-50%. However, for complex business logic, legacy system integration, and security-critical code, AI still requires heavy human oversight. The average development efficiency gain is 25-35%.
Testing is a significant opportunity. AI can generate test cases, analyze test coverage, and automate regression testing. This is a 20-30% efficiency gain.
Maintenance is where the AI impact is most profound. A significant portion of IT service work is maintaining and updating existing systems. AI can automatically analyze log files, identify anomalies, and suggest fixes. This could be 30-40% more efficient.
When you weigh these by their share of project costs, the blended average comes out to around 20-28%—which is precisely the 25% figure quoted by Hexaware's CEO. So the number is not fantasy; it is a reasonable estimate of the average efficiency gain.
But here's the problem. The number is true and irrelevant at the same time.
The critical assumption in the 25% is that the project remains the same. The same functionality is delivered. The same quality standards are met. And the only difference is that fewer human hours are required. This is a conservative, linear view of AI's impact. It assumes that AI is a tool that assists humans, rather than a technology that fundamentally changes the nature of the software creation.
I have seen in my own work how the AI changes the creative process. When I need to analyze a complex smart contract, I don't just use a tool to write code faster. I use it to explore the design space, to generate hypotheses, to test scenarios. This is not just a cost reduction; it's a capability expansion. The AI is enabling me to do things I could not do before, not just to do the same things faster.
The same is true for IT services. The ultimate impact of AI is not to reduce the cost of the same software development, but to enable the creation of software that was previously impossible or uneconomical. The client may not just want the same system for 25% less; they want a fundamentally different system—one that is more automated, more intelligent, and more integrated.
This is where the 25% narrative becomes dangerous. By focusing on cost reduction, the IT services industry is playing a game of defense, trying to protect the existing model of the existing model of the existing model of the existing model of the existing model. The real opportunity is not in cutting the cost of the old model, but in creating the new one.
But that requires a different type of conversation. It requires the IT services firm to say to its clients: "We will not just build the system you asked for, cheaper. We will show you how to redesign your business processes, to exploit the full potential of AI. The system you think you need will be obsolete in three years. Let's build the future instead."
That conversation, of course, is much harder. It requires a deep understanding of the client's business. It requires a willingness to challenge the client's assumptions. It requires a level of strategic thinking that the traditional IT services model does not reward.
Part IV: The Dilemma of the Medium Player
Hexaware is not TCS. It is not Accenture. It is a mid-sized player with approximately 30,000 employees and annual revenues in the range of a billion dollars. It is caught in a precarious position between the giants that have the resources to invest heavily in AI capabilities and the AI-native companies that have the potential to bypass traditional IT services entirely.
Let me position Hexaware in the competitive landscape. The first tier includes Accenture, TCS, and Infosys. These companies have announced AI investments in the billions. Accenture has committed $3 billion to AI. Infosys has launched the Topaz platform. Wipro has invested $1 billion in AI.
The second tier includes companies like Hexaware, LTIMindtree, Persistent Systems, and Mphasis. These companies have annual revenues between $1 billion and $5 billion. They have strong customer relationships and domain expertise, but they lack the financial resources to make massive AI bets. They are in a "fast follower" mode, waiting to see what the big players do, and then trying to replicate it on a smaller scale.
The third tier is the AI-native companies. These are startups like Devin, Cursor, and various other AI coding tools. They are not IT services companies in the traditional sense, but they threaten to be. If an AI can generate an entire application from a natural language description, why does the client need a team of developers at all?
This is the competitive squeeze that Hexaware faces. The first tier is pulling away in AI capabilities. The third tier is eating away at the bottom end of the market—the simple, repetitive development work that is the entry point for many IT services firms.
The Hexaware CEO's statement is an attempt to manage the narrative. By publicly declaring that AI will reduce costs, Hexaware is signaling to its clients: "We are not in denial. We will help you benefit from AI." It is also signaling to its investors: "We understand the new world order."
But there is a fundamental problem with this strategy. The 25% cost reduction is a universal benefit. Every IT services company will be able to offer it, because the AI tools are available to everyone. The AI is not a proprietary advantage; it is a commodity input. If all IT services companies can reduce costs by 25%, then the market will simply price this into the baseline. The clients will expect 25% lower prices from everyone. The cost reduction does not create a competitive advantage; it simply resets the competitive baseline.
The real competitive advantage is not in the cost reduction, but in what the IT services company does with the freed-up resources. It can reinvest them in higher-value services: AI strategy consulting, data science, machine learning model development, and systems integration for AI. It can move up the value chain, from being a "body shop" to being a "solutions architect."
This is the classic "innovation dilemma" described by Clayton Christensen. The very efficiency that makes the current business model more profitable in the short term undermines its long-term viability. The IT services company must, in effect, make its own business model obsolete before the competition does.
Part V: The Unseen Cost of Human
The 25% cost reduction has a dark side that the CEO did not mention: the human cost.
The IT services industry is one of the largest employment sectors in India, providing jobs for millions of engineers. The lower-cost human arbitrage model has been the foundation of the Indian economy's growth for the past 30 years. The growth of the Indian middle class has been driven, to a significant extent, by the success of the IT services industry.
If AI reduces the cost of IT projects by 25%, it also reduces the number of human-hours required. This is a direct threat to the employment of junior and mid-level developers.
The picture is particularly acute for entry-level positions. AI tools are most effective in automating routine tasks that are typically assigned to junior developers. The code generation, test automation, bug fixing—all of these are entry-level jobs. If these jobs are automated, the entry-level position becomes the ladder into the industry, and it becomes much harder for new graduates to gain experience and advance.
The impact is not just on individual workers, but on the entire industry ecosystem. The IT services companies are the training ground for engineers. If the training ground shrinks, the long-term talent supply will be damaged.
This is the hidden cost of the 25% efficiency gain. The industry is not just cutting costs; it is cutting the roots of its own future.
The IT services industry is not an industry of direct employment. It is the backbone of the Indian economy. The annual revenue of the Indian IT services industry is estimated to be around $250 billion. It supports a much larger ecosystem of suppliers, services, and support staff. If the industry shrinks, the effects will be felt throughout the entire economy.
The CEO of Hexaware is not just making a statement about his company's efficiency. He is making a statement about the future of a nation's economic model. He is saying that the model of selling human time at a markup is coming to an end. And the consequences of that are far greater than a 25% cost reduction.
Part VI: The New Logic of Value
So, if the cost reduction is not the real story, what is? What is the actual impact of AI on the IT services industry?
I believe the answer is the shift from a "time-based" pricing model to an "outcome-based" pricing model. The IT services industry has always been selling "input" - hours of developer time. The AI makes the input less valuable. The industry needs to start selling "output" - the business outcome that the software enables.
This is a fundamental transformation. It requires IT services companies to understand the client's business at a much deeper level. It requires them to take on a much higher level of risk. The client is no longer buying an uncertain process; they are buying a guaranteed outcome.
For example, an IT services company might say, "We will build you a system that will reduce your customer churn by 10%. We will be paid based on that outcome." This is a much more valuable proposition than "We will build you a system for $1 million."
The outcome-based model aligns the incentives of the IT services provider and the client. If the system doesn't work, the provider doesn't get paid. This is a much more honest and effective business model.
The outcome-based model is also much more defensible in the age of AI. If the AI can reduce the cost of the input, the outcome-based model captures the value of that efficiency. The provider can deliver the outcome at a lower cost and keep the difference as profit.
This is the strategic opportunity for the IT services industry. The 25% cost reduction is the price of admission. The real value is in the ability to deliver outcomes that clients cannot achieve on their own, even with AI tools.
The IT services industry needs to move from "we will do what you ask" to "we will tell you what to do, and we will do it." This requires a level of confidence and expertise that the industry has not yet been cultivated. But it is the only path to survival.
Part VII: The Competitive War
The 25% cost reduction will not create a level playing field. It will actually increase the competitive advantage of the largest players.
The large IT services companies like Accenture, TCS, and Infosys are investing billions of dollars in AI. They are building their own AI platforms, training their own models, and developing their own AI-based methodologies. They have the scale to do this. They have the data. They have the talent.
The medium-sized companies like Hexaware do not have the same resources. They are forced to rely on third-party AI tools, which means they do not have a unique advantage. They will be able to offer the 25% cost reduction, but so can their competitors.
The AI-native companies are a different kind of threat. They are not just cutting the cost of the existing work; they are changing the nature of the work. They are developing systems that can build entire applications without human intervention. If this becomes a reality, the entire IT services industry will be threatened.
In this environment, the medium-sized IT services companies are in the most dangerous position. They are not large enough to make the massive AI investments required to differentiate, and they are not small enough to be the AI-native companies that can be revolutionary. They are stuck in the middle.
The CEO's statement is a move to defend the middle. It is a message to clients: "We are still relevant. We can give you the AI benefits." It is a message to investors: "We are still relevant. We are on top of the AI trend."
But the statement is not enough. The real test is whether Hexaware can demonstrate AI capabilities in actual client projects. Whether it can show a genuine difference in the value it delivers.
Part VIII: The Unanswered Questions
The statement from Hexaware's CEO raises many questions that are not answered. These are not the questions of a journalist; they are the questions of a systems analyst who has seen too many projects fail.
First, what is the quality of the AI-generated work? The 25% cost reduction assumes that the AI can do the work of a human developer, but at a lower cost. However, the AI-generated code is not always of high quality. It can be insecure, inefficient, and difficult to maintain. The cost of fixing the AI-generated code can sometimes exceed the savings.
Second, what is the impact on the architecture? The AI tools are best at generating the individual components of a system, but they are not good at designing the overall architecture. A poorly designed architecture can lead to significant problems in the future. The cost of the AI is not the cost of the components; the cost is the design of the overall system.
Third, the client is being trained. The AI tools are available to the clients as well. The client is not going to be passive. The client will use the AI tools to do some of the work themselves. They will become more informed and more demanding. The IT services company will have to provide a higher level of value to justify its existence.
Fourth, what is the AI actually learning from? The AI tools are trained on the existing codebase, which includes a lot of legacy code. The AI could be learning the legacy system. It could be perpetuating the bad code of the past. The AI is not a magic solution; it is a tool that reflects the quality of the data on which it is trained.
Part IX: The Human Cost
I cannot end this analysis without addressing the human cost of the IT services transformation. The 25% cost reduction is not just a number; it is a loss of livelihoods. It is a loss of opportunity for the next generation of engineers.
The IT services industry is the backbone of the Indian economy. It has provided the middle class with opportunities. It has been the ladder for social mobility.
When the CEO says AI will cut costs by 25%, he is saying that the industry will need 25% fewer people to do the same work. This is not just a matter of profit margins; it is a matter of lives.
The industry needs to think about a "just transition" - a way to move to the AI era without causing massive unemployment and social upheaval. This requires investment in retraining, in education, in social safety nets. The IT services companies cannot just shed their employees like a cost of a software project.
The IT services industry has a responsibility to manage this transition responsibly. It has a responsibility to help its employees learn new skills. It has a responsibility to invest in the future of its workers, not just in the future of its profits.
Conclusion: The Architecture of Trust
The Hexaware CEO's statement about the 25% AI cost reduction is a small thing. It is a single data point in a large and complex system.
But it is a signal. It is a signal that the IT services industry is entering a period of fundamental change. The change is not about a 25% cost reduction. It is about the end of the time-based pricing model. It is about the end of the human labor arbitrage model. It is about the end of the industry as we know it.
The future of IT services is not in the ability to do the same work at a lower cost. It is in the ability to do new work. It is in the ability to create new value for clients. It is in the ability to become a strategic partner rather than a vendor of hours.
The future of IT services is in the ability to build trust. Not trust in the old sense of "we will do the work you ask, and we will be there to do it." But trust in the new sense of "we understand the technology, we understand the business, and we will be a reliable guide through the complexity of change."
The architecture of trust in a trustless system is built on the same principles. It is not built on promises, but on the ability to deliver. It is not built on the big statements, but on the small details. It is not built on the 25% cost reduction, but on the 100% reliability of the system.
The IT services industry is facing the same challenge as the blockchain industry. It is the challenge of trust. It is the challenge of value. It is the challenge of change.
The companies that will survive are not the ones that quote the most impressive numbers. They are the ones that deliver the most reliable outcomes. They are the ones that can navigate the complex, unpredictable, and chaotic environment.
The industry is at a turning point. The 25% cost reduction is not the destination. It is the beginning of the journey.