OPINION | Legacy systems might survive AI, but old strategies will not

AI does not need to replace legacy code to replace legacy advantage.

It is reasonable to argue that the AI moment is being oversold, that large enterprises will not abandon decades of legacy infrastructure overnight, and that complex systems in banking, government, and industry cannot simply be rewritten because a new model has arrived. There is truth in this caution.

Technology transitions in the real world are rarely clean, and adoption is often slower than hype suggests. But reassurance can become its own blindness.

Disruption does not require legacy to vanish in a single stroke. Retail did not disappear overnight because people still liked stores, yet the economics of retail changed all the same.

The deeper disruption of artificial intelligence is not about whether old systems are scrapped tomorrow. It is about whether the assumptions that sustained entire industries remain valid at all. AI does not need to replace legacy code to replace legacy advantage. The mistake is to confuse the pace of replacement with the pace of irrelevance. Enterprises may not abandon their past quickly, but markets have never waited for institutions to feel ready.

AI is not simply another productivity tool layered on top of existing systems. It is a structural force reshaping how work is priced, how decisions are made, and where future economic leverage will reside. That is why strategy for India’s IT and IT-enabled services sector must be built on preparedness, on the willingness to plan for the worst rather than assume continuity, and on recognising that complacency is often the most expensive legacy system of all.

The disruption is economic

The most profound impact of AI will not be that enterprises discard legacy systems overnight. It will be that the economic model around those systems begins to change. For decades, India’s information technology sector thrived on a powerful equation. Talent was abundant, labour was relatively inexpensive, and global firms were willing to offshore work at scale. The engine of growth was cost labour arbitrage combined with disciplined execution. India became indispensable not because it owned frontier intellectual property, but because it delivered reliability, scale, and efficiency to the world.

AI threatens to disturb this equilibrium in ways that go beyond coding assistance or incremental automation. When intelligence becomes cheap and replicable, the premium on routine cognitive labour begins to fall. Work that once required armies of engineers can increasingly be performed by smaller teams amplified by models. The question is no longer only about efficiency gains. It is about whether the global buyer still needs the same volume of human effort, whether the pricing of services can hold, and whether advantage will migrate away from scale toward platforms, intellectual property, and deep research.

The danger is not that services will disappear tomorrow. The danger is that differentiation will shrink, pricing power will weaken, and competitive advantage will move elsewhere. India built a services superpower. The next era may reward those who build intelligence infrastructure.

AI is accelerating risk, not just productivity

A second misunderstanding lies in treating AI as merely an enhancer of output. The real issue is that AI changes the nature of organisational risk. Human errors are usually bounded. They occur in pockets, in departments, in localised decisions. AI errors replicate at scale. A flawed model, a biased dataset, or an untested deployment can shape thousands of decisions instantly, often with no clear line of accountability. The risk is systemic, and the opacity of these systems makes governance harder even as adoption accelerates.

This is why the organisations that endure will be those that govern AI best. Strategy will require experimentation, but also restraint. Innovation, but also institutional seriousness. Trust, regulation, accountability, and control will become competitive capabilities, not bureaucratic afterthoughts. If leaders assume that AI adoption will be slow because legacy systems are complex, they miss the point. AI does not need to modernise the core to reshape the edge. Customer interaction, fraud detection, credit scoring, legal review, software testing, and knowledge work are already being transformed around the periphery, and competitive pressure will eventually force the centre to follow.

But there is another uncomfortable truth that India must confront. The global AI revolution is not being driven only by technology. It is being driven by the political economy of Western capitalism, which has long been willing to experiment at scale, even when those experiments carry enormous social cost. The Western world can push aggressively into automation, absorb employment shocks, and still keep markets chugging along because it has deeper safety nets, stronger capital buffers, and a different demographic reality.

Populated economies like India cannot afford disruption as casually. We do not have the luxury of treating labour displacement as collateral damage of innovation. The social consequences of rapid employment compression in a country of India’s scale are immediate, political, and destabilising. That is why the AI challenge here is not simply about adoption speed, but about the asymmetry of what different societies can endure.

This is also why moral comfort will not solve the problem. Assuming that human goodness will temper automation, or lamenting that AI investments may not always succeed, misses the structural imbalance. Western AI startups can afford to fail spectacularly. They can shut shop, burn billions, and disappear, while the broader economy keeps moving because the depth of investment pools remains intact. Capital in those systems is designed for massive experimentation, and geopolitical influence often ensures that technological standards are set in their favour regardless of failure.

India does not yet have that depth of risk capital, nor that cushion of geopolitical leverage. The very ability of the West to absorb repeated AI failures is itself a competitive advantage, because it accelerates learning and normalises scale.

The geopolitical layer India cannot ignore

AI is not unfolding in a stable global order. We are entering a world where trade is increasingly weaponised, technology is increasingly sovereign, and regulation is increasingly shaped by geopolitical rivalry. Big Tech firms are not neutral suppliers. They are strategic actors with platform power, regulatory influence, and the ability to set standards that others must follow.

In such a world, the old assumptions of frictionless globalisation cannot be taken for granted. India’s cost advantage was built in an era when work moved freely across borders and services were largely insulated from strategic contestation. AI will not operate in that environment. Data rules, model controls, compute restrictions, and national security frameworks will shape the terrain.

The question is not only whether Indian firms can adapt technologically, but whether they can adapt strategically in a world where sovereignty itself is part of the technology stack.

What India must resist is the assumption that past success guarantees future relevance. Building technology services at scale over the last three decades was a remarkable achievement, but it was shaped by a specific global moment and economic conditions that no longer hold. The AI era will not reward confidence performed for optics. It will reward humility, serious capability-building, and an honest recognition of what we still lack, starting with a stronger STEM ecosystem, better quality education, and far deeper investment in research.

For years, much of corporate India, including the technology sector, treated research as optional and financial engineering as sufficient. Excess capital was returned through buybacks and dividends instead of being channelled into frontier innovation. That imbalance was survivable when the global model was stable. It is not survivable now.

(Srinath Sridharan is Author, Policy Researcher & Corporate Advisor, Twitter: @ssmumbai.)

Views are personal, and do not represent the stand of this publication.