AI forces intelligence rethink in IT

Artificial Intelligence (AI) adoption among Indian firms is more than 25 per cent, and the new tools can add more than $400 billion to the GDP over the next decade. As productivity gains shoot up, and layoffs become the norm, as we have mentioned several times in these columns, they create structural tensions, and a rethink about existing business models within the IT and software services sector. Tech that promises growth begins to weaken the earlier tech that built the former. For three decades, Indian IT scaled on a simple equation. More clients, more work per vendor, more people, more revenues, and more profits.
Billings are tied to efforts, teams are sized to timelines, and global clients outsourced works and processes to take advantage of Indian scale. AI breaks this equation. When the same work is done faster, with fewer employees, due to high productivity gains, there is a shakedown. Demand does not disappear. But billings based on manpower use shrink. In a business where billings matter, the killings begin, as the shillings begin to vanish. There are huge gaps in work flows that need fillings, but only if the affected firms are willing to do so.
This is the paradox of the $250-billion IT industry. Adoption remains high, productivity improves, and demand for tech is intact. Yet, the businesses are in trouble, which is reflected in their stock prices. The link between the delivered work and earned revenue begins to diverge, and growth is no longer guaranteed by scale. Big may not necessarily be beautiful, and although it is not a scary axiom, being big is not necessarily the only route to success. Big can turn ugly. In effect, smaller firms with smart AI can deliver better and faster.
Coding, testing, maintenance, and parts of consulting are now assisted by AI models trained on large datasets. Sequential tasks are parallel and simultaneous. Iteration cycles shrink. In some cases, projects that require weeks are completed in days. What changes is not the existence of work, but the quantum of human effort required. The implications create pricing pressures, as firms race to cut costs by doing the one thing they know best, i.e., mass sackings. Clients question why, with the advent of AI, the same work costs the same, especially if the timelines tighten. AI, in some senses, is deflationary for IT services.
Hence, the historical advantages collide with the future transition. IT was built essentially on labour arbitrage, whatever the firms may claim. It was the ability to deliver high-quality work at lower costs, either at the client’s end (onshoring), or in Indian campuses (offshoring and outsourcing) that defined success. AI introduces another form of arbitrage. What matters now is access to intelligence, models, and data, and the ability to deploy them effectively. The transition is from labour to intelligence arbitrage. IT needs less people, but more skilled people that can multitask, and handle multiple roles. It is a new paradox.
In an intelligence-based system, the growth is not linear (more work, more people, more work) but non-linear. AI can scale across clients without proportional increases in headcount. Hence, the new ecosystem needs a massive redesign to create and capture new values. Some of it is underway. One direction is outcome-based pricing, where clients pay for results rather than hours. In theory, this allows firms to gain from faster deliveries. In practice, it introduces new risks. Revenue is tied to performances, accuracy, and measurable outcomes. Managers need to rejig their mindset, and reduce focus on HR, hirings, and attritions.
Another shift is towards platform-led models. Instead of building custom solutions for each client, firms invest in reusable AI layers that are deployed across industries. Hence, there may be no dedicated teams to handle specific clients. Code-generation frameworks, testing engines, and domain-specific models for specific sectors will rule the show. The objective is to move from project-based revenues to scalable, repeatable products. Indian IT did this in some ways by engineering processes and systems that were replicable and scalable. But this will not work. The focus needs to change from people to tech, as it was in the 1980s.
A third layer relates to AI infrastructure. As firms deploy AI systems, they require support in areas such as data pipelines, model governance, compliance, and security. This creates a new category of services that is less about building software, and more about the operations and maintenance of the AI-driven systems. Hence, hardware to handle huge amounts of data is as important as AI. More importantly, the new hardware will cost a lot, unlike the old one, where costs tumbled. Data centres and power systems hint at brick-n-mortar solutions.
Economic risks hover. If the productivity gains are passed on to the clients via lower prices, margins compress. Revenue growth decouples from contracts and outsourced work. The industry grows, but with a different and potentially weaker economic profile. There is the positioning risk. Indian IT operates on a global scale, where competitors move along different trajectories. In the US, large tech firms embed AI into products and platforms, and capture value through IPRs. In China, firms leverage scale, data, and ecosystem integration. India’s strength remains in services, execution, and talent at scale.
At the same time, India retains advantages that are not easily replicated. It has one of the largest talents of tech pools, a growing base of digital infrastructure, and an expanding dataset. The ecosystem is moving, even if unevenly. The challenge lies in translating this into defensible advantage. The unevenness is crucial as IT sits between demand and capability, and its evolution will depend on how effectively it can reposition itself within this equation. If the shift succeeds, Indian firms will emerge stronger, with higher margins, leaner structures, and more scalable revenue streams. If not, there is trouble in paradise.
Growth may come from deploying capabilities. This seems to be the only way to stall value erosion, and enhance the chances of future existence. Hence, AI may not necessarily be a destabiliser, but a hidden opportunity to change models, and survive profitably. More importantly, this is not a fight between tech, like radio versus TV versus Internet versus AI. It is a form of internal transition within tech, like software-hardware to Internet to AI. Thus, adaptation and adoption are distinct possibilities. Indeed, many IT firms are in the middle of such changes. The terms to profit from the new opportunity have changed. It is imperative for the industry to do so too.















