When a 25% Drop in 1 Day for a $100 B Cloud Security Powerhouse, or in 9 Trading Days for a Prestigious Dow 30 Leader, Isn’t a Discount.
What the sudden collapse at IBM and CrowdStrike teaches us about AI disruption and the difference between price and value
On February 20th, at 1 PM ET, Claude announced “Claude Code Security”—an automated AI tool that scans codebases for vulnerabilities. A day later, CrowdStrike stock CRWD had erased $20 billion in market capitalization. Three days after that, IBM touched its steepest nine-session drawdown in decades.
Neither company is in the Cara 100 watchlist of investable companies, so my remarks here are for strictly educational purposes. This is about price versus value.
Let’s first examine the trigger for these stock price collapses.
Anthropic demonstrated Claude Code’s ability to understand and modernize COBOL—the 65-year-old language that still underpins global banking, government benefits, and transaction processing. The same announcement that punished a cybersecurity pure play also reached into the mainframe era and pulled a blue-chip incumbent into the vortex.
But, if you had bought either stock solely because it was down 25% in days, you would have missed the point entirely.
The Most Dangerous Phrase in Investing
“It’s down so much—it must be cheap.”
This is instinct talking, not discipline. And in an era of AI-driven narrative repricing, instinct is a quick way to impair capital.
The attached framework is explicit: a stock becomes attractive only when price < revised intrinsic value, not when price < prior price. After a structural shock—and AI capabilities are precisely that—the relevant comparison is against updated assumptions, not historical valuation anchors.
Those assumptions include:
Sustainable revenue growth
Long-term operating margins
Competitive advantage duration
Capital intensity
Risk discount rates
Without revising these inputs, “cheapness” can be illusory. A 25% stock drop may represent fair repricing, partial repricing, or overreaction—but you cannot know which without re-underwriting the thesis.
What Just Happened to IBM?
Over the past month, IBM’s share price has plummeted, from the low 300s to the low-mid 220s. On current numbers, IBM still looks like a solid, cash-generative incumbent. But here’s the warning: equity value lives in the out-years, where assumptions about moat duration, pricing power, and capital intensity reside.
Claude Code attacks the very foundation of IBM’s mainframe and consulting economics.
For decades, IBM’s ecosystem has benefited from deep enterprise entrenchment, high switching costs, and the scarcity of COBOL expertise. Modernizing large COBOL estates has historically required armies of consultants spending years on discovery—mapping dependencies, reverse-engineering business logic, documenting workflows.
Claude Code compresses that timeline from “years to quarters.” It reads entire COBOL codebases, maps program entry points and data flows, documents forgotten workflows, and surfaces risks that would take human analysts months to uncover.
Notably, none of these outcomes must be imminent for valuation to adjust. Markets price optionality and risk asymmetry, not just baseline forecasts. When a firm introduces an AI capability claiming to automate a previously labor-intensive process, investors reassess the distribution of future outcomes.
Before the announcement: base case dominates, disruption scenarios assigned low probability.
After the announcement: tail risks receive higher weighting, expected duration of moats may shorten.
Even modest adjustments—reducing long-term margin expectations by 100-200 basis points, pulling forward competitive pressure by 2-3 years—can justify substantial changes in intrinsic value. This explains why equity prices may move sharply despite no change in current earnings.
Markets are pricing “What if this technology works better than expected?” rather than “Has the business deteriorated today?”
The Three Things a Price Decline Can Mean
A common analytical error is interpreting rapid price declines as evidence of instantaneous business impairment. In reality, three distinct processes may occur:
A. Valuation Recalibration
Multiples compress. Discount rates rise. Growth durability assumptions soften. No immediate revenue or earnings impact required.
B. Sentiment & Positioning Effects
Momentum strategies de-risk. Crowded trades unwind. Volatility targeting reduces exposure. Price action may amplify beyond fundamental revisions.
C. Fundamental Revisions (Lagging)
Only later do analysts potentially adjust revenue forecasts, margin trajectories, and capex requirements.
In many cases, fundamentals remain intact while valuation resets persist.
Confusing price action with fundamental deterioration can lead to flawed decisions: selling high-quality assets during temporary multiple compression, buying “cheap” stocks without reassessing structural assumptions, misjudging risk-reward asymmetry.
Why Multiples May Not Mean-Revert
Traditional value frameworks assume eventual normalization: growth stabilizes, margins recover, multiples expand. But under structural change, normalization may not occur.
We have seen this before:
Legacy media vs. streaming platforms
On-premise software vs. SaaS
Physical retail vs. e-commerce
In each case, lower multiples reflected permanent economics resets, not temporary mispricing. Markets repriced sectors based on anticipated rather than realized disruption. Valuation compression preceded earnings deterioration.
The IBM-Claude episode is the same pattern with a new technology protagonist.
Reframing the Question
Instead of asking “Is the stock down enough?”, investors should ask:
What assumptions justified the prior valuation?
Which of those assumptions are now impaired?
Is the market overestimating or underestimating the change?
Value emerges from mispriced expectations, not price declines.
In IBM’s case, Claude Code raises legitimate questions about whether modernization work becomes less labor-intensive, less bespoke, and more contestable—which would pressure consulting margins and weaken the lock-in that supported premium pricing. A lower P/E or EV/EBITDA can be entirely rational if the market is simply adjusting to a business with more commoditized services and a shorter period of excess returns.
The Broader Pattern
The IBM episode is not isolated. It is an early stress test of how AI-driven modernization will reshape both the economics of legacy systems and the valuation frameworks investors use to price them.
AI announcements that credibly automate complexity—code analysis, security workflows, legacy modernization—shift models away from stable base cases toward fatter tails. The assumed half-life of competitive advantages shrinks.
Companies with premium valuation multiples, perceived technological leadership, and margin-rich business models are particularly sensitive. Small downward revisions to sustainable growth rates, competitive barriers, or return-on-capital durability can drive disproportionate equity declines.
Narrative risk becomes multiple risk. Multiple risk becomes market cap risk.
A Disciplined Playbook
For institutional and serious independent investors, the lesson is not “avoid incumbents hit by AI headlines,” but “treat every such drawdown as a forced re-underwriting, not an automatic bargain.”
The checklist:
Thesis Integrity – Is the original investment case intact? Have structural drivers changed?
Earnings Power Durability – Are margins at risk? Is pricing power weakening?
Competitive Position – Does AI weaken the moat? Does it commoditize key services?
Valuation vs. Revised Base Case – Is price below updated intrinsic value? Or merely below outdated expectations?
Time Horizon Alignment – Can the portfolio absorb prolonged multiple compression? Is patience capital available?
In IBM’s case, that means explicitly modeling scenarios where modernization becomes less bespoke, consulting margins compress, and mainframe upgrade cycles stretch—then asking whether today’s 220s price embeds enough pessimism to create favorable asymmetry.
Where Value Will Emerge
The COBOL story points to a broader playbook. Legacy software giants that aggressively deploy their own AI agents to cannibalize legacy economics—IBM with watsonx Code Assistant for Z, Microsoft with AI agents for COBOL migration—can potentially trade “high-margin services today” for “higher-quality, platform-anchored cash flows tomorrow.” Those that move slowly invite competitors to set the new price-performance frontier for their own customer base.
For investors, the opportunity will lie less in buying every drawdown and more in distinguishing tactical overreaction from genuine structural impairment: which incumbents are using AI to accelerate their customers’ transition and which are being dragged into it.
At the portfolio level, this argues for two complementary exposures: carefully underwritten positions in legacy firms with credible modernization roadmaps, and targeted allocations to AI-native platforms whose economics scale with the volume and complexity of code they are asked to analyze.
Final Observation
In periods of technological transition:
Cheap stocks can become cheaper.
Expensive stocks can become uninvestable.
Value is created by misjudgment of future cash flows, not by volatility alone.
Prices adjust to revised expectations. Fundamentals adjust to realized outcomes. The gap between the two is where volatility—and opportunity—resides.
The IBM-Claude inflection is a reminder that in an AI world, value will increasingly accrue to those who make complexity cheap, not to those who merely keep it running.
A sudden price decline in a structurally challenged business is not a value opportunity by default. It is a demand to completely re-underwrite the valuation, starting from revised economics rather than nostalgic anchors.
That is the difference between price and value. And in this market, it has never mattered more.