Software’s Spectral Shift

The year, as these things are measured, has begun with a certain digital malaise. A tremor, if you will, through the software stratum. The latest emanations from Anthropic—Opus 4.6, a name redolent of both precision and a faintly unsettling numerology—have prompted a rather theatrical re-evaluation of value. Investors, those exquisitely sensitive seismographs of capital, now ponder a question both pedestrian and profound: will these emergent intelligences replace software, or merely diminish the appetite for its licenses? A distinction, naturally, with a considerable difference to the balance sheet.

It is curious, then, to observe a divergence amongst the oracles of Wall Street. Tom Lee, a man not unfamiliar with bullish pronouncements, now speaks of impending job losses, a rather grim prognosis. Dan Ives, conversely, sees a “golden buying opportunity,” a phrase that, after decades of market cycles, feels increasingly like a well-worn incantation. Which of these gentlemen, one wonders, possesses a clearer view through the algorithmic fog? The answer, as is so often the case, is likely less a matter of right or wrong, and more a question of temporal perspective.

One recalls a similar episode a mere twelvemonth ago, when the DeepSeek model—an efficient, almost impertinent newcomer—sent a shiver through the AI complex. A panic, briefly, ensued. Yet, as often happens with these fleeting fits of market hysteria, it proved a rather splendid moment for those with the patience—and the capital—to accumulate. Jevon’s Paradox, that mischievous principle suggesting increased efficiency begets increased consumption, played its familiar role. Will software follow suit? The iShares Expanded Tech-Software Sector ETF (IGV +1.69%) currently suggests not, having surrendered a rather substantial 21.7% of its value. A decline, one might say, with a distinctly melancholic air.

The current unease stems, of course, from the accelerating proficiency of these artificial agents. Opus 4.6, in particular, introduces the notion of “swarms”—teams of agents dividing tasks with a chilling efficiency. The implications are rather obvious, are they not? If these agents can now write software, perform the very functions that once required human programmers, does the value of the underlying code itself diminish? A question that haunts the corridors of Silicon Valley, and, no doubt, the dreams of its CEOs.

The interesting thing—and one that speaks volumes about the current state of affairs—is that even the most ardent bulls on Wall Street are divided. Lee’s gloomy pronouncements, as we’ve noted, are a departure from his usual optimism. Ives, on the other hand, clings to the belief that AI will expand software usage, offsetting any pricing pressure. A rather neat symmetry, wouldn’t you agree? Perhaps the truth lies somewhere in the nuanced space between these two perspectives—a space that, alas, is rarely acknowledged in the breathless reporting of financial news.

What, then, are the executives saying? The usual platitudes, naturally. They assure us that today’s fears are overblown, that AI will complement software, not replace it. One suspects, however, that this is less a statement of genuine belief and more a carefully calibrated attempt to reassure investors. Jensen Huang of Nvidia, a man not given to understatement, suggests that AI will merely use existing software tools. A pragmatic observation, to be sure, but one that fails to address the underlying issue of pricing power.

ServiceNow’s Bill McDermott, ever the astute salesman, points out that his company offers both seat-based and usage-based pricing models. A clever hedge, to be sure, but one that merely postpones the inevitable reckoning. The question is not whether software companies can adapt their pricing models, but whether they can maintain their margins in a world of increasingly commoditized intelligence. Salesforce, Figma, and others are all rushing to partner with LLMs, hoping to capture a piece of the AI pie. But will they end up merely as conduits for these new technologies, surrendering a significant portion of their value to the LLM providers?

The coming months will undoubtedly be turbulent. Software valuations, already elevated by years of easy money and unbridled optimism, are likely to remain under pressure. The incumbents—Salesforce, ServiceNow, and the like—possess certain advantages: vast troves of data, established relationships with enterprise customers, and a degree of inertia that is difficult to overcome. But these advantages are not insurmountable. A nimble newcomer, with a disruptive technology and a compelling value proposition, could easily unseat the established players. The question, as always, is not whether disruption is inevitable, but when and how it will occur.

I suspect the incumbents will survive, but with diminished margins and a more competitive landscape. The certainty, however, is low. A newcomer offering similar capabilities at a fraction of the price could easily gain traction. Even if such a newcomer fails to unseat the incumbents, it could force them to lower their prices, squeezing their profits. And even if prices remain stable, increased usage could still offset the decline in per-unit revenue. These are complex questions, with no easy answers. As such, I would advise caution. A long-term buying opportunity may eventually present itself, but a significant rerating higher is unlikely in the near future.

2026-02-18 19:03

Read More