On 23 February 2026, the global technology markets witnessed not merely a selloff, but a moment of strategic inflection. Shares of IBM fell approximately 13 percent in a single trading session, marking their steepest one-day decline in more than two decades, after Anthropic announced that its Claude system could materially streamline and accelerate COBOL code modernization. Nearly 30 billion dollars in market capitalization was erased in hours. Markets rarely react with such force to a single product statement unless it touches a structural nerve. This announcement did precisely that. It challenged the economics of legacy expertise that has underpinned enterprise IT services for over half a century.

COBOL remains one of the most consequential programming languages ever written. Industry estimates frequently cited by banking and technology associations indicate that more than 200 billion lines of COBOL code remain active globally, processing trillions of dollars in daily financial transactions across banking, insurance, pensions, tax administration and government services. In the United States alone, it is often estimated that a majority of financial transactions still touch COBOL-based systems at some stage. This vast installed base created a durable economic moat. Scarcity of skilled engineers, high switching costs, regulatory entrenchment and institutional inertia combined to make modernization a slow, expensive and highly billable endeavor. When an advanced AI model credibly claims it can ingest, interpret and refactor these codebases in weeks rather than years, investors do not wait for incremental validation. They discount the future.

The Technological Breakthrough and Its Real Boundaries

The technological claim underlying the market reaction is neither speculative hype nor science fiction. Over the past three years, large language models have demonstrated significant advances in code comprehension, generation and translation. Empirical research examining AI-assisted software engineering has shown measurable productivity gains in code documentation, dependency mapping and refactoring support. In controlled experiments, LLM-augmented workflows have improved developer efficiency in certain tasks by 30 to 60 percent, particularly in legacy code comprehension, which historically consumes a large portion of modernization budgets.

Anthropic’s assertion that Claude can analyze COBOL, map business logic dependencies, identify risk clusters and generate modernization pathways strikes at the most labor-intensive phase of legacy transformation. The discovery and documentation phase often accounts for a substantial share of total project cost because enterprise COBOL systems embed decades of regulatory rules, exception handling logic and undocumented workflow assumptions. Accelerating this phase compresses timelines and alters cost structures.

Yet intellectual rigor demands acknowledging limits. Translation is not transformation. Syntactic conversion from COBOL to Java or another modern language does not automatically guarantee semantic fidelity, regulatory compliance or operational resilience. Enterprise systems interact with downstream databases, batch processes, audit logs and external regulatory reporting frameworks. Academic research has consistently cautioned that LLM outputs must be validated against domain constraints, integration contracts and compliance standards before production deployment. The technology is powerful, but it remains an augmentation layer rather than an autonomous replacement for systemic accountability.

Why the Contagion Spread Across Global IT Services

The shockwaves did not stop with IBM. Investors quickly extrapolated implications for global IT services firms such as Tata Consultancy Services, Infosys, Wipro, HCL Technologies and Cognizant. While precise exposure varies by firm and vertical, application maintenance and modernization services constitute meaningful revenue streams, particularly in financial services, insurance and government sectors where COBOL footprints remain extensive. Even if only a portion of revenue is directly tied to legacy maintenance, these segments often enjoy stable margins because they rely on scarce skills and long-duration contracts.

If AI tools reduce manual code analysis hours by half or more, clients will demand repricing. The traditional outsourcing model, built on labor intensity and billed hours, faces compression. The market reaction reflects this logic. It is not that revenues will disappear overnight. It is that pricing power and margin structures may be structurally altered. In capital markets, expectations move faster than balance sheets.

The Economics of Scarcity in an Age of Scalable Intelligence

For decades, COBOL expertise commanded premium compensation because it was rare and mission-critical. Universities largely ceased teaching it. Veteran engineers aged out of the workforce. Enterprises hesitated to rewrite stable systems that processed billions in daily transactions. This created a scarcity premium. AI disrupts precisely that premium. When a model can parse millions of lines of code in minutes, explain dependencies, and generate structured documentation, the cognitive barrier to entry collapses.

However, scarcity does not vanish. It migrates. The new scarcity lies in system architecture, domain interpretation, regulatory assurance and AI governance. Understanding a pension calculation algorithm is less about syntax and more about public policy, actuarial assumptions and compliance requirements. An AI can summarize logic, but it cannot bear fiduciary responsibility. The premium therefore shifts from language fluency to architectural and governance fluency. Firms that recognize this shift early will reposition their talent base accordingly.

From Labor Arbitrage to Outcome Orchestration

The deeper disruption is economic rather than technical. Traditional IT services monetized effort. Projects were priced on headcount, hours and offshore leverage. AI compresses effort. As effort shrinks, outcome-based pricing becomes more rational. If a modernization assessment that previously required 10,000 consulting hours can now be executed in 3,000 AI-augmented hours, clients will demand cost savings. This pressures revenue per project but potentially increases throughput.

Strategically agile firms can convert this pressure into advantage. By embedding AI into proprietary delivery platforms, they can execute more engagements in less time, preserving aggregate revenue while reducing cost structures. More importantly, modernization rarely ends with code translation. Once legacy systems are understood, enterprises typically embark on cloud migration, API integration, real-time analytics deployment and AI-enabled decision systems. These higher-value transformation layers offer new revenue pools. The margin does not disappear. It relocates upward in the value chain.

Governance, Compliance and the Human Anchor

Enterprise modernization is inseparable from regulatory and operational risk. Banks operate under capital adequacy norms. Insurers face solvency regulations. Governments manage statutory audit obligations. A minor computational discrepancy in a tax system can trigger legal consequences. AI-generated refactoring must therefore undergo rigorous validation, testing and compliance certification.

This governance layer becomes a strategic differentiator. Firms that integrate AI with strong assurance frameworks, automated testing pipelines, explainability mechanisms and audit trails will command trust. In regulated industries, trust is monetizable. The competitive landscape will not reward raw automation alone. It will reward accountable automation.

Geopolitical and Workforce Implications

The implications extend beyond corporate earnings. Much of the world’s legacy IT work has flowed to cost-competitive geographies, particularly India. If AI reduces labor intensity, comparative advantage shifts from wage differentials to innovation capacity. Countries that cultivate AI research ecosystems, advanced digital skills and regulatory clarity will capture greater value in the new cycle.

For the workforce, the transition demands urgency. Purely syntax-based roles will decline in relative value. Roles in AI supervision, enterprise architecture, cybersecurity integration, regulatory technology and digital governance will expand. The strategic question for firms and policymakers is not whether AI will automate legacy tasks. It is whether they will invest early enough in reskilling to remain indispensable in higher-order functions.

Future Scenarios and Strategic Imperatives

In the near term, volatility will continue as enterprises pilot AI-assisted modernization tools and quantify reliability and cost savings. Within two to three years, AI augmentation is likely to become a standard component of modernization RFPs. Contracts will increasingly specify productivity benchmarks enabled by AI. In the longer horizon, legacy modernization may become commoditized, with differentiation shifting toward full-spectrum business reinvention, cybersecurity resilience and AI-native enterprise architectures.

The imperative for incumbent firms is clear. First, integrate AI deeply into delivery models rather than resisting it defensively. Second, migrate pricing frameworks from hours to outcomes. Third, invest heavily in governance capabilities that ensure explainability and regulatory compliance. Fourth, cultivate intellectual capital in architecture, systems thinking and sector-specific knowledge.

The Rewriting of Economic Gravity

The market reaction to IBM’s decline was not a verdict on a single company. It was a recognition that intelligence itself has become scalable. When machines can interpret the digital arteries of global finance, the scarcity of knowledge ceases to be a durable moat. The new moat is orchestration. It lies in the ability to combine AI capability, human judgment, regulatory assurance and strategic foresight into coherent enterprise transformation.

The question is not whether AI will eat IT companies. The question is whether IT companies will metabolize AI faster than the market discounts their relevance. History suggests that technology waves rarely annihilate entire sectors. They reorder them. Those who adapt ascend to higher value planes. Those who cling to yesterday’s complexity as protection discover that code can now rewrite not only itself, but the economics of an entire industry.

(Major General Dr. Dilawar Singh, IAV, is a distinguished strategist having held senior positions in technology, defence, and corporate governance. He serves on global boards and advises on leadership, emerging technologies, and strategic affairs, with a focus on aligning India’s interests in the evolving global technological order.)