The Mainframe’s Ghost: A Mildly Irritating Observation

by

The market, that capricious beast, reacted with a predictable spasm. Anthropic’s pronouncements regarding COBOL, that venerable relic of a bygone computing age, were enough to send shivers through IBM‘s share price. A perfectly reasonable overreaction, one might say, if one hadn’t spent decades observing the peculiar habits of financial panics. It is as if a particularly loud sneeze in the accounting department is immediately interpreted as the harbinger of economic ruin.

The premise is simple enough: Anthropic’s AI, Claude, promises to modernize COBOL code, threatening IBM’s mainframe business. A perfectly logical concern, if one were to ignore the inconvenient truth that IBM has been anticipating this particular ghost for decades. It’s rather like being startled by a shadow you yourself have been diligently cultivating. The tumble in IBM’s shares, exceeding twelve percent, was a performance worthy of a particularly clumsy magician.

But let us not succumb to hysteria. The market, you see, appears to have forgotten a few rather important details. Details that, if remembered, might induce a mild case of indigestion in those who so eagerly sold their IBM stock on Monday. Allow me to illuminate these forgotten corners of reality.

1. The Serpent Eats Its Tail: IBM’s Own Alchemy

Anthropic’s innovation, while admirable, is hardly a bolt from the blue. It is, in fact, a rather predictable development, and one that IBM has been diligently preparing for. For years, IBM has been offering its own COBOL modernization tools, a quiet alchemy transforming ancient code into something… less ancient. They even provide options for migrating these systems from the mainframe, that imposing fortress of computing, to more… pliable platforms like Linux or Windows. A rather pragmatic approach, wouldn’t you say? Engineering one’s own obsolescence, with a touch of elegant resignation.

2. The Devil’s in the Debugging: AI’s Imperfect Gifts

Artificial intelligence, that modern Prometheus, has indeed wrought wonders. But let us not mistake a clever trick for genuine mastery. Engineers now describe what they desire, and Claude, or one of its brethren, dutifully spits out code. Often, it works. But when it doesn’t… ah, when it doesn’t, the results are… instructive. CodeRabbit, a source of some amusement, reports that AI-generated code contains roughly sixty percent more errors than that crafted by human hands. A rather significant margin of error, wouldn’t you agree? And as the complexity increases, so too does the difficulty of untangling these digital knots. It is as if one were attempting to build a cathedral from quicksand.

The truth, my friends, is that AI’s auto-coding tools remain, shall we say, under development. Claude is unlikely to be an exception. Consider, if you will, the critical systems that still rely on COBOL: ATM transactions, retail purchases, air travel bookings, bank transfers, and even the bureaucratic machinations of government agencies. These institutions are not inclined to risk a critical failure on the altar of technological novelty. They require guarantees, not promises. And Claude, alas, cannot provide them.

3. The Mainframe’s Enduring Spirit: A Fortress of Logic

Anthropic’s ultimate, unvoiced argument is that COBOL modernization will allow institutions to abandon IBM’s mainframes in favor of the siren song of cloud computing. A tempting proposition, perhaps, but one that overlooks a rather fundamental truth. Cloud computing, you see, is essentially rented space on someone else’s servers. Convenient, certainly, but also… slower. It lacks the raw, unadulterated speed of a mainframe, that imposing monolith of logic.

IBM’s “Z” systems, for example, can process twenty-five billion encrypted transactions per day. A rather impressive feat, wouldn’t you say? And they offer more than just speed. They are complete, self-contained systems, with built-in security features – including quantum encryption – and an almost preternatural level of dependability. They are, in short, a fortress against chaos.

But here’s the truly ironic twist: IBM’s newest mainframes are also remarkably well-suited for artificial intelligence. Each system can handle a staggering 450 billion AI inferences per day. A rather substantial capacity, wouldn’t you agree? It is as if the mainframe, far from being rendered obsolete, is poised to become the very engine of the AI revolution. Kyndryl, a firm with a keen eye for such matters, reports that over half of mainframe users are increasing their usage, with returns on modernization costs exceeding 300 percent. They are, in short, doubling down on the future, using the past as their foundation.

Loading widget...

IBM: Not a Doomed Vessel

The point, my friends, is this: While it is likely that AI-powered coding will eventually become sophisticated enough to modernize COBOL programs and move them off the mainframe, that day has not yet arrived. And it may never arrive, if mainframes continue to prove their superiority in several critical areas.

Therefore, the recent setback in IBM’s share price presents a compelling opportunity for investors. The market, in its characteristic shortsightedness, jumped to the wrong conclusion. IBM’s mainframe business, and its ancillary operations, are likely to remain remarkably resilient for years to come.

Straits Research predicts that the worldwide mainframe market will grow at an average annualized pace of nearly 8 percent through 2033. Not explosive growth, perhaps, but respectable. And IBM is well-positioned to lead the charge. You can acquire a stake in this growth while the stock is trading at a reasonable valuation of less than 19 times this year’s projected earnings. A bargain, one might say, for a company that continues to defy the predictions of its doom.

Read More

2026-02-28 00:33