FILE PHOTO: Collectible figurines with computer systems and smartphones are seen in entrance of the phrases “Synthetic Intelligence AI” on this illustration taken, February 19, 2024. REUTERS/Dado Ruvic/Illustration/File Photograph
| Photograph Credit score:
Dado Ruvic
Shares of IBM tumbled over 13 per cent on Monday, wiping out greater than $30 billion in market capitalisation and dragging Indian IT shares decrease, after a weblog put up by Anthropic stoked investor considerations that AI-led instruments like Claude Code might speed up COBOL modernisation.
The decline weighed on Indian IT shares, with the Nifty IT index dropping practically 5 per cent on Tuesday. LTIMindtree fell 6.43 per cent, Tech Mahindra declined 6.17 per cent, HCLTech slipped 5.83 per cent, whereas Infosys and TCS had been down 3.56 per cent every.
Anthropic impact
Analysts attributed the inventory sell-off to a weblog put up by Anthropic, which highlighted the fragility of legacy COBOL programs. Frequent Enterprise-Oriented Language (COBOL) is a high-level, English-like, compiled programming language developed for enterprise knowledge processing wants.
Anthropic famous that lots of of billions of traces of COBOL nonetheless energy vital infrastructure even because the pool of engineers who perceive the language is shrinking. A lot of this code was constructed many years in the past and is poorly documented.
Historically, COBOL modernisation required massive consulting groups, multi-year timelines, and excessive prices. Anthropic argued that AI modifications that equation; instruments like Claude Code can compress modernisation cycles from years to quarters.
Responding to a mail despatched by businessline, IBM said, “IBM has been investing in code modernisation for years – each via skilling initiatives and thru our personal GenAI capabilities. Over two years in the past, we launched watsonx Code Assistant for Z (IBM’s mainframe) as a result of we perceive the advantage of AI in modernising code. New AI instruments emerge each week, together with our personal. What they don’t change is the basic engineering problem of working mission-critical workloads at scale. Translating COBOL is the straightforward half. The actual work is knowledge structure redesign, runtime substitute, transaction processing integrity, and hardware-accelerated efficiency constructed over many years of tight software program and {hardware} coupling. That’s the drawback IBM has spent many years studying to resolve, and AI is probably the most highly effective software we’ve got ever needed to do it.”
Pareekh Jain, founder and CEO of EIIR Pattern, mentioned the sell-off could also be an overreaction, however the underlying threat is actual, noting that the prospect of labor being accomplished in as little as two weeks threatens conventional effort-based billing fashions.
AI-native companies like Anthropic are shifting quicker as a result of they haven’t any legacy income to guard. In IBM’s case, COBOL and mainframes have been core for many years, and its differentiation might erode if instruments like Claude Code allow different service suppliers, together with mid-tier companies, to speed up modernization. Whereas service suppliers will nonetheless be wanted, if AI handles 80 per ent of the work, their position shrinks to the remaining 10-20 per cent, making it extra commoditised.
“With Claude Code, there are two views — the agent could not work successfully as a result of there isn’t sufficient publicly accessible coaching knowledge, and hypothesis round whether or not proprietary IBM knowledge could have been accessed to coach such fashions,” he mentioned.
Whereas brokers and repair suppliers will coexist, the broader concern is cannibalisation, Jain added. Legacy spend could decline quicker than new AI-led spend emerges. Whereas AI is creating new alternatives, the tempo of latest investments relies upon not simply on expertise but in addition on macro situations. “If the pace of decline is quicker than the pace of latest spending, buyers fear,” he mentioned.
Automation
Aravind Putrevu, VP of Progress at Coderabbit, defined that Anthropic is explicitly saying Claude Code automates “exploration and evaluation” and might get one to a modernizable plan quicker, even in quarters as a substitute of years.” That is plausible as a result of most COBOL applications usually are not exhausting due to syntax, however as a result of the enterprise logic is fossilised and the dependencies are tribal. What’s much less confirmed is end-to-end modernization “at scale” with out people proudly owning take a look at technique, knowledge migration, and production-grade validation. Whereas AI makes the unknown quicker, it doesn’t magically make regulated manufacturing programs protected.
Nonetheless, IBM’s place in mainframes is extra defensible than the market response suggests, he mentioned. “IBM’s moat is platform-level: tight hardware-software coupling, transaction throughput, IO subsystem optimization, accelerators, and many years of efficiency tuning. IBM is already framing this as ‘translation will not be modernization,’ arguing you could’t replicate the system-level properties by simply shifting code. Additionally, IBM’s present numbers present the mainframe cycle will not be lifeless weight proper now: Infrastructure was up, and IBM Z was sharply up year-over-year in This fall. COBOL is the headline, however reliability, safety posture, and operational predictability are the true causes these programs keep.”
In the meantime, Naga Santhosh Josyula, Co-founder, Tablesprint, argued that whereas AI instruments are more and more automating work that used to take massive groups of consultants months or years, consulting gained’t disappear; slightly, the character of billable work shifts towards governance, technique, and high quality assurance, with fewer hours spent on repetitive technical duties.
Revealed on February 24, 2026
