For a few years now, every market convulsion tied to AI has followed a familiar pattern: a single announcement, a wave of panic, a brutal repricing of entire sectors. The easy read of these events is noise — the market overreacting, as it tends to do, before settling back to something sensible.
February 2026 made me think harder about that framing.
On a Tuesday afternoon, a company called Algorhythm Holdings — previously a karaoke equipment business with roughly three million dollars in market cap, under two million in quarterly revenue, and a net loss on the books — issued a press release. The claim: they were scaling freight volume by 300 to 400 percent without adding headcount, using AI. Within hours, CH Robinson, one of the largest freight brokers in the world, dropped 24 percent intraday. The Dow Jones Transportation Average shed 17.4 billion dollars. Not because of what Algorhythm actually did. Because of what the market imagined it might mean.
The market is holding two ideas that cannot both be true
What struck me most is the paradox at the center of all this. The same market that has been skeptical of AI infrastructure returns — selling off chip makers and cloud providers on concerns about whether the spend will ever pay off — is simultaneously pricing in catastrophic disruption to incumbents across every sector. A world where AI does not deliver enough value to justify the infrastructure investment is not the same world where a press release from a former karaoke company should erase 17 billion dollars from the freight industry.
Both beliefs cannot be true. But the contradiction does not protect you from the damage. A CFO watching their company’s stock drop 24 percent in a single session does not have the luxury of pointing out the logical incoherence. The board meeting happens. The hiring freeze goes in. The roadmap gets rewritten around an “AI strategy” assembled over a weekend.
That is the reflexivity trap, and it is the part worth sitting with. The panic itself becomes the mechanism. Companies that respond to AI fear by cutting product teams and signing performative AI vendor deals are the ones who will face actual disruption in three years — not from karaoke companies, but from competitors who used the scare-trade moment to build genuine capability instead of investor optics.
The three-category distinction the market ignores
The analysis that emerged from this event introduced a genuinely useful framework: separating AI-exposed sectors into three buckets.
The first is where AI is genuinely displacing labor now. Software development is the clearest example — coding assistant tools have reached hundreds of millions in annual revenue, and per-seat pricing models are dying. The second bucket is where disruption is real but the timeline is measured in years, not quarters: wealth management, for instance, where an AI rate-comparison tool does not yet replace the trust relationship and behavioral coaching a financial advisor provides. The third is where the market has lost the plot entirely. CH Robinson’s hundred thousand shipper-carrier relationships, its proprietary freight lane data, its cross-border regulatory and contractual complexity — none of that is invalidated by a press release.
The market is pricing all three identically. That is the mispricing. And if you are a knowledge worker trying to figure out where you actually stand, this distinction is far more useful than watching the stock ticker.
What the scare trade revealed about careers
Here is the signal worth sitting with: every organization that watched its sector get hammered in February now has a board asking “what can AI actually do to our business?” And almost nobody in those organizations can answer that question with genuine specificity.
The valuable person is not the one who says “I have heard AI can do this.” It is the one who can say: “I tested this on our actual contract review workflow.” It handles roughly 70 percent of initial analysis accurately, but misses conditional clauses with cross-references about a third of the time — which means we need a human check at this specific stage. Deploy it this way and we cut review time by 40 percent and reduce outside counsel spend by roughly 200 thousand dollars per quarter.” That is not a vendor demo. That is credible, domain-specific knowledge built from real tests, with real numbers.
The distinction the market panic made visible to every executive in ten days is one we should have been discussing anyway: the difference between roles whose contribution is synthesis — aggregating, summarizing, producing reports that draw together other people’s work — and roles whose contribution is judgment, knowing which information matters and why the standard approach will not work for this client.
Synthesis is now competing directly with a faster, cheaper tool. Judgment is more valuable than ever.
The career move that actually differentiates
The 2024 version of this advice was “learn AI.” That is table stakes now — it gets you considered, not promoted. The version worth giving going into the second half of 2026 is narrower and more demanding: run real tests on real workflows in your specific domain, and build credible numbers from what you actually observe.
The scare trade will pass. The organizations that used the panic as cover to build genuine capability — and the people who can bridge the gap between what AI can actually do and what the business actually needs — are the ones who will look back on this period as an opening, not a scare.
Text summarized and optimized using Anthropic’s models and reviewed by a human.