As India’s AI Summit 2026 unfolds, celebrating the country’s emergence as an AI powerhouse, a parallel story is flashing red on trading screens.

 

Indian IT majors TCS, Infosys, Wipro, and Tech Mahindra have collectively shed over $80 billion in market value. The Nifty IT index has suffered its sharpest fall since the pandemic, wiping out nearly ₹2.5 lakh crore in a matter of days.

 

 

The trigger was Anthropic launching “Claude Cowork”, an agentic AI system capable of planning, executing, monitoring, and autonomously completing multi-step workflows. AI was no longer a co-pilot. It was becoming autopilot.

 

Unlike earlier generative AI tools that drafted emails or wrote code snippets, agentic AI executes backend coding, contract review, and sales workflows independently. Combined with the “Patil Effect”, a cost breakthrough that made these systems cheaper than offshore labor, the economics changed overnight.

 

Indian IT’s traditional model is linear: Revenue ≈ Billing Rate × Headcount × Utilization.

 

AI platforms operate on exponential economics: Revenue ≈ Model Capability × User Base ÷ Marginal Cost.

 

If 40–50% of entry-level coding, testing, and maintenance work can be automated, the billable-hour model looks structurally exposed. Markets extrapolated margin compression, headcount redundancy, and contract renegotiation risk.

 

But history cautions against linear extrapolation.

 

Markets systematically overprice first-order destruction and underprice second-order adaptation. We have seen this before. In 2013, it was believed that the cloud would kill Indian IT. Instead, firms became cloud integrators. In the Dot-com era, everyone thought retail would die, but it evolved into an omnichannel model.

 

Going back further in history, Jevons Paradox (1865): More efficient coal engines led to higher coal consumption, not lower. When photography became ubiquitous through smartphones, professional photography did not disappear. Value shifted from technical execution to creative direction. AI is doing the same to knowledge work.

 

Coding hours may commoditise. Orchestration, governance, compliance, and domain integration become scarce.

 

Moreover, Enterprise adoption follows S-curves. Cloud took 15 years. Mobile-first took 10+.
 

Agile transformation took over a decade. AI capability is exponential. Enterprise adoption is sigmoid.

 

 

Markets may be pricing 2028 disruption into 2026 earnings.

 

Globally, the tech puzzle appears to be more of an issue of Capex vs. Monetization. US hyperscalers are projected to spend ~$700 billion in AI capex in 2026 alone. The “Magnificent Seven” are investing heavily in GPUs and data centers, but visibility into monetization varies.

 

 

Microsoft has ~$13B AI revenue run rate. Alphabet integrates AI into search and cloud. Meta spends aggressively with limited direct AI revenue. The market is rewarding clarity on revenue, not capex announcements. This resembles the 2000 fiber bubble. Infrastructure was built years before applications caught up. Builders suffered first; platform beneficiaries captured value later.

 

Like always, no one can predict the future with certainty, so let us assign probabilities to the possible future paths.

 

1) Slower capability plateau (AI Winter 2.0): 15%

 

• AI progress may slow after early productivity gains. Current breakthroughs could represent low-hanging fruit in pattern recognition rather than sustained exponential growth in intelligence. If model improvement is constrained by compute limits, data constraints, or algorithmic ceilings, disruption stretches over a longer horizon.

 

• Implication: Indian IT and SaaS models remain viable longer.         Current panic proves excessive.

 

2) Regulatory friction: 20%

 

• Governments may slow adoption through data protection rules, liability frameworks, employment safeguards, copyright enforcement, or national security restrictions. In markets like India and Europe, political sensitivity around job displacement could delay aggressive AI automation.

 

• Implication: Transition extends by 3–7 years. Incumbents get         adaptation time. Earnings collapse scenarios weaken.

 

3) Enterprise resistance and complexity: 30%

 

• Agentic AI may work in demos but struggles in real-world enterprise environments, where legacy systems, compliance constraints, cybersecurity risks, hallucination errors, and change-management challenges slow integration.

​​​​​​​

• Implication: AI becomes an augmentation before full automation. Integration demand rises. Service firms pivot rather than disappear.

 

4) New bottlenecks emerging: 35%

 

• As AI automates routine cognition, new constraints appear:
  • Data quality
  • Governance and compliance
  • Security architecture
  • Outcome measurement
  • Human oversight

​​​​​​​

• Value migrates rather than vanishes. Entirely new service layers emerge.

 

• Implication: Revenue pools shift up the stack. Survivors expand margins under new pricing models.

 

Considering all scenarios, markets are currently pricing a rapid, uniform collapse of labor-heavy models. However, the highest-probability outcomes involve slower, uneven, and adaptive disruption, not immediate extinction.

 

Current valuations imply near-total, immediate collapse. That statistical mismatch is where alpha may lie. When cognition becomes cheap, coordination becomes expensive. The AI stack has three layers:

 

1. Foundation models & compute (capital-intensive, winner-take-most)
2. Integration & orchestration (enterprise complexity layer)
3. Vertical/domain-specific applications

 

Layer 2 is where incumbents can survive. Indian IT’s “body shop” model is fading. The future is system integration, AI governance, and managed orchestration. Agentic AI hallucinates, and it breaks compliance; it needs supervision. That supervision is billable, but under outcome-based pricing. Under hourly billing, productivity reduces revenue. Under outcome pricing, productivity expands margins. Hence, the pivot in the pricing model is more important than AI adoption itself.

 

My dear readers, the question is not whether AI changes everything. It will. The question is: who captures the new profit pool? And that, for patient capital, is an opportunity—not a catastrophe. Ask:

 

• Is management reallocating capital toward AI in a meaningful way?
• Is pricing shifting toward outcomes?
• Does the firm operate in complex domains?
• Does it have orchestration capability?
• Can it fund a 2–3 year transition without stress?

​​​​​​​

Companies that score high on at least three of these five parameters are likely to belong in portfolios. Those scoring one or fewer are candidates for structural decline — regardless of current valuation comfort. Reduce exposure to pure, linear, headcount-dependent firms lacking an AI strategy.

 

Markets are correct that disruption is real. They are incorrect that it is uniform, immediate, and zero-sum.

 

 

Best Regards

 
Dr. Tarunika Jain Agrawal Ph.D. 

Alpha Research

 

Disclaimer:  The perspectives provided reflect our expectations of the economy, market, and investments, which may or may not materialize. Investors are encouraged to conduct their own due diligence before investing.