COBOL’s AI moment: Anthropic rattles IBM as India spots a $1.6 trillion services opening

A strange thing happened in enterprise tech over last week. A 67 year old programming language briefly took centre stage in the global AI economy, and a 115 year old company took a market hit that looked like panic.

When Anthropic claimed that Claude Code can help modernise COBOL in “quarters instead of years”, IBM shares fell 13.2% in a single session, its steepest daily drop since 2000.

On the surface, this looks like yet another overreaction to the ongoing AI hype carnage. But, if one digs deeper,  it unearths something far more consequential of a language that’s considered inconsequential. Let’s accept it,  AI has started shaking the economics of the most entrenched enterprise work, the kind that built and fed the global IT services industry for decades.

And India, sitting at the heart of that industry, has to read the moment carefully.

Why IBM got rattled

IBM’s mainframe world is not just a technology stack. It is a trust stack. Banks, insurers and governments run critical transactions on systems that cannot afford mistakes and cannot afford downtime. COBOL modernisation, historically, has been a slow, consultant-heavy business because so much of the work is not “writing code” but understanding business rules, dependencies, testing and cutover risk.

Anthropic’s statement did not claim to replace all that. But it attacked the most labour-intensive layer of the stack: exploration, analysis, documentation, refactoring et al- essentially it’s a legacy modernization prep up . It implied that the part that used to need army of specialists could be compressed by AI tooling. Reuters captured the investor fear clearly: the claim raised questions about IBM’s relevance and revenue potential in system modernisation services.

That is what the market was really pricing. Not the death of mainframes overnight. The erosion of a moat built on scarcity, complexity, and long modernisation timelines.

Legacy lock-in, in a subtle form

This is where the “lock-in” debate becomes nuanced.

IBM will argue, credibly, that enterprises do not want a dramatic rewrite. They want safer modernisation, step by step, with auditability, compliance and continuity. That is a rational CIO position.

But there is a second truth that can co-exist: modernising “in place” extends the life of the incumbent ecosystem. If the modernisation toolchain, workflows, governance and delivery model remain anchored to the same vendor and its partners, the customer may get modernisation without necessarily gaining freedom. The lock-in does not look like a trap. It looks like comfort. And comfort nevertheless sells in regulated industries.

The counter-narrative: tech debt as expanded TAM

Now cut to some of the conversation that happened in the recent Nasscom’s NTL 2026 sessions, and you can see why the Indian services industry is eager to own this story.

Let’s look at what Srikanth Velamakanni, Vice Chairperson, Nasscom stated in a discussion.

Velamakanni’s said: “With over 800 billion lines of COBOL code in play, even $2 a line translates into a $1.6 trillion tech services opportunity. That is tech debt turning into an expanded TAM.”

What can one read out of this? It says: do not see legacy as baggage. See it as inventory. AI makes it finally fundable.

That is the bet. AI cuts cost and risk enough that boards will approve programmes they postponed for years. If the funnel widens, the overall opportunity can grow even if effort per project falls.

But the comfort messaging from AI CEOs needs scrutiny

Here’s where the industry conversation becomes slippery. AI-native leaders have increasingly warned that white-collar work will be automated, often including parts of coding and software delivery. Yet almost every such warning comes wrapped in the reassurance: “It will also create jobs.”

Is that a fact, or a diplomatic afterthought? It can be both. Over the long evolution of technology, new jobs do emerge. But the harder question is timing and distribution. Jobs can be created in one set of skills, locations and wage bands, while being destroyed in another. For India’s IT services machine, that transition risk is not theoretical anymore. It is already visible as hiring slows and delivery models change.

India’s tech outlook: revenue holds, headcount decouples

Nasscom’s latest numbers still show growth. The industry body expects India’s tech sector to grow 6.1% to $315 billion in FY26, and AI is compressing traditional work lead times but expanding other areas.

That is the real structural shift. The Indian services industry was built on a simple engine: more revenue, more people. AI is breaking that linkage.

Job cuts and the new services equation

This is where the story becomes uncomfortable, especially for the middle layer.

In recent months, Indian news coverage has tracked more layoffs and planned reductions, with some firms saying it as performance pruning, pyramid rebalancing, or demand cyclicality. Yet the broader narrative is hard to miss: AI-assisted delivery is pushing productivity up, reducing the need for large teams for certain kinds of work, and slowing hiring momentum.

The COBOL moment makes this sharper. Because COBOL modernisation has traditionally been a labour sponge. If AI compresses the sponge, the industry has to replace volume with higher-value assurance, domain expertise, governance, and outcome-led programmes.

That is not a small transition. It reshapes billing, staffing pyramids, training pipelines, and the fresher machine.

So what is the way forward

Traditional providers do not have to lose this game. But they cannot win it with yesterday’s playbook.

Take 1: COBOL modernisation is not just code conversion. The premium is in equivalence proof, testing depth, audit trails, and cutover safety. That is where large providers can still charge for outcomes.

Take 2: AI can generate code faster. It cannot automatically guarantee that a pension system or payments backbone will behave correctly under every edge case. The market will pay for confidence.

Take 3: This is not about choosing between “AI-native” and “enterprise legacy.” The winners will be those who can bring AI-native tooling into legacy estates with governance, security and repeatability.

Take 4: India’s services sector should stop selling only the comfort line that jobs will be created. The more credible story is: some roles will shrink, many will change, and the industry must invest at scale in reskilling, especially for the middle layer that powered the services pyramid.

The deeper takeaway

COBOL is not back because the world suddenly fell in love with old code. COBOL is back because AI has made legacy transformation economically tempting and strategically contested.

IBM’s wobble was a signal that even iconic incumbency has a shorter half-life now. Nasscom’s $1.6 trillion framing is a signal that India’s services industry wants to turn that shock into opportunity. Both can be true.

The only question is whether the industry tells itself a comforting story, or does the hard work of evolving its model before the market forces it to.