Why the IT Services Meltdown was Inevitable

The market is in two minds: one, that AI disruption will arrive gradually, a slow erosion of legacy business models, giving incumbents time to pivot. Or two, that AI is going to signal the end of the industry. In early February, we saw what investors thought... Nearly $280 billion in market capitalization wiped from IT services and software stocks in what analysts have already dubbed the “SaaSpocalypse.” Accenture down 22%. Cognizant down 21%. Capgemini down 27%. IQVIA, the hardest hit, down almost 31%. No geography was spared. No business model was safe. The brutality shocked even the most jaded analysts.

The trigger? Anthropic released eleven open-source plugins for its Claude Cowork agent, targeting precisely the workflows that IT services firms have monetized for decades: legal review, sales operations, marketing analytics, data processing, etc. Very soon this work will be done in seconds as opposed to hours/days. The work of entire outsourcing delivery centres could well be made redundant. Needless to say, investors did not like this answer and acted accordingly.

But here is the uncomfortable truth most industry commentary is avoiding: this was not a black swan event. This was the market finally catching up to a reality that has been building for at least eighteen months.

The Real Problem Is Not AI. It Is the Business Model.

For years, the global IT services industry has operated on a fundamental bargain: enterprises pay for human effort, measured in hours, headcount, and FTEs. The entire commercial architecture, from pricing to delivery to margin expansion, depends on people doing work that clients either cannot or will not do themselves.

Agentic AI does not just threaten to do that work faster. It threatens to make the unit of measurement obsolete. When an AI agent can execute a workflow end-to-end, ingesting data, applying business rules, generating outputs, and even iterating based on feedback, the concept of billing by the hour becomes as outdated as billing by the telegram.

This is what the market recognized this month. Not that AI models had suddenly become smarter (they had been getting smarter for years), but that the delivery mechanism had shifted. Plugins, agents, and orchestration frameworks turned theoretical AI capability into practical displacement. The gap between “impressive demo” and “production-grade replacement” closed overnight.

India’s IT Sector: Ground Zero

Nowhere was the impact more visceral than in India. The Nifty IT index suffered its steepest weekly decline since March 2020, a 9.4% drop in five days that erased approximately $24 billion in a single session. TCS fell below $110 billion in market capitalization for the first time since 2020. Infosys recorded its worst single-day decline in over two years. Wipro, LTIMindtree, and HCL Tech followed in lockstep.

The vulnerability is structural. Indian IT services firms generate the majority of their revenue from precisely the kind of work agentic AI targets: application maintenance, QA testing, business process management, data migration, and rules-based analytics. These are high-volume, lower-complexity engagements where the value proposition has always been labor arbitrage.

AI does not just match that cost advantage. It eliminates the need for it entirely.

Nasscom, to its credit, pushed back on the panic, arguing that enterprise-scale AI adoption requires deep domain expertise, governance frameworks, and human oversight that cannot be automated overnight. They are not wrong. But the market is not pricing in what happens next quarter. It is pricing in what happens over the next four years and Jefferies estimates that 9% to 12% of traditional IT services revenue could be at risk in that window.

What the Survivors Will Look Like

The firms that emerge strongest from this correction will not be the ones that resist the shift. They will be the ones that rewire their business models around it. This means three things:

First, moving from effort-based pricing to outcome-based pricing, and doing it credibly, not as a marketing exercise. If the value you deliver can now be replicated by an agent, then your pricing must reflect something the agent cannot provide: accountability, integration, and domain judgment at scale.

Second, becoming AI-native in delivery. Not “AI-augmented” in the way every services firm has been claiming for the past three years, but genuinely restructuring delivery teams so that AI agents handle execution while human professionals handle exception management, client advisory, and strategic escalation. The ratio of humans to agents will define margin structure for the next decade.

Third, investing in the data layer. As IBM demonstrated with its $11 billion acquisition of Confluent, AI does not fail because of weak models. It fails because the data foundation is brittle. Services firms that can help enterprises build real-time, governed, trusted data infrastructure will find themselves at the centre of the next wave of spending, not on the wrong side of it.

The Bottom Line

February 2026 was not the end of IT services. But it was the end of pretending that the old model has a long runway left. The $280 billion correction was the market’s way of saying what the industry has been reluctant to admit: effort-based services built on labor arbitrage are a depreciating asset in an AI-first economy.

The firms that understand this, and act on it with urgency, not incrementalism, will define the next era of the industry. The rest will become case studies in what happens when you mistake a cost advantage for a competitive moat.

The clock is ticking. The question is not whether to transform. It is whether you still have time.