IBM Shares Plunge 13.2% Amid AI Disruption Fears Triggered by Anthropic
On February 23, financial markets experienced a sudden and dramatic reaction, seemingly pricing in a future scenario reminiscent of Citrini Research's provocative report titled "THE 2028 GLOBAL INTELLIGENCE CRISIS." Shares of International Business Machines (IBM) plummeted by a staggering 13.2%, marking their steepest single-day decline since October 18, 2000. This sharp downturn occurred immediately after AI startup Anthropic hinted that its innovative Claude Code tool could potentially revolutionize the modernization of legacy COBOL systems, which are predominantly run on IBM mainframes.
Anthropic's AI Tool Sparks Market Panic
Reuters reported that COBOL, a decades-old programming language still extensively utilized in critical sectors such as banking, insurance, and government systems, might be transformed through AI-driven automation. In a statement, Anthropic elaborated: "Modernizing a COBOL system once required armies of consultants spending years mapping workflows. Tools like Claude Code can automate the exploration and analysis phases that consume most of the effort in COBOL modernization." The company further added, "With AI, teams can modernize their COBOL codebase in quarters instead of years."
The market's response was swift and far-reaching. Not only did IBM suffer, but other software and cybersecurity stocks, including prominent names like CrowdStrike and Datadog, also slid significantly. Investors appeared to be reassessing how artificial intelligence could disrupt established revenue models and threaten traditional business frameworks across the technology landscape.
The Viral Report That Spooked Investors
"The 2028 Global Intelligence Crisis" presents a gripping and alarming narrative. It envisions a world, merely a few years from now, where AI agents systematically hollow out white-collar employment, "Ghost GDP" dominates national economic accounts, unemployment surges past 10%, and the consumer economy deteriorates because, as Citrini describes, machines "spend on discretionary goods. (Hint: it’s zero.)"
It is no coincidence that such a report moved markets. As noted by economist Noah Smith in his analysis titled "The Citrini post is just a scary bedtime story," a cluster of software and finance stocks sold off immediately after the report went viral. Smith interprets this reaction as less about analysts uncovering a genuine blind spot and more about a wave of sentiment: traders read an evocative crisis narrative, saw their tickers referenced, and engaged in synchronized panic-selling.
Assessing the Impact on Jobs and Firms
On a microeconomic level, Citrini sketches a scenario where AI agents "take care of almost all white-collar work, like coding, research, transactions, and even making strategic decisions." While the exact industry-by-industry trajectory remains uncertain, current usage patterns offer some clarity. The Economist highlights that by late 2025, approximately 41% of American workers had used generative AI at work, yet only 13% of working-age adults used it daily, and merely about 5–6% of work hours involved generative AI. Most applications still focus on discrete tasks—such as drafting, summarizing, or coding assistance—rather than fully autonomous agents managing entire workflows.
When AI is deployed, task-level productivity gains are substantial. Studies on tools like ChatGPT indicate completion times for writing tasks can decrease by nearly 40%. Experiments at firms like Boston Consulting Group demonstrate productivity lifts of 12–25% on realistic professional tasks, and a broader academic review finds gains of 15–30% in real-world settings. Thus, there is a solid micro story that AI will gradually reshape many white-collar jobs and business models over time. However, this differs significantly from the claim that such transformation will occur nearly totally and very rapidly by 2028.
Macroeconomic Implications: From Disruption to Crisis?
The stronger assertion in Citrini's piece is macroeconomic: that rapid micro-level disruption could crash the entire economic system. The report imagines an unemployment rate exceeding 10%, a sharp decline in consumption, and a world where the "human-centric consumer economy, 70% of GDP at the time, withered."
Smith points out that Citrini never fully articulates a macroeconomic model. They do not elucidate the transmission mechanism from firm-level disruption to aggregate crisis; readers are simply expected to accept that enough business models will break, leading to GDP and employment collapse. Smith suggests two vaguely plausible channels through which a service-sector AI productivity boom could end poorly: a financial crisis triggered by mass business-model failure, or a demand shortfall if labor income collapses faster than prices adjust and policymakers respond.
Neither scenario is impossible, but both require strong, specific assumptions: that disruptions arrive faster than balance sheets and regulations can adapt, and that policymakers remain passive in the face of a visible technology shock. Historical precedent tends to contradict such assumptions.
Current macroeconomic data paints a very different picture from Citrini's narrative. As The Economist emphasizes, there is little evidence of a runaway AI productivity surge in actual numbers. The US economy grew by 2.2% in 2025 while employment remained essentially flat, translating to productivity growth of roughly 1.9%—just below the post-war average and far lower than during the internet boom years. Much of this growth stemmed from investment in data centers and related infrastructure. When researchers adjust for that investment, underlying productivity appears "close to zero."
This is not the footprint of an economy already in the grip of an AI super-cycle destined to explode by 2028. Instead, it reflects an economy still in the build-out phase, with modest realized productivity gains thus far.
Ghost GDP, Distribution, and Policy Responses
The most original concept in Citrini's essay is "Ghost GDP": output produced by AI that does not translate into human income, and thus into demand. The concern is intuitive: if AI enables firms to accomplish the same work with far fewer people, labor income shrinks, and because "machines spend zero," the system loses its primary engine of consumption.
Smith's macroeconomic response is that distribution matters more than aggregates, and policy is not powerless. GDP represents total output; the crucial question is who owns the machines and the profits. Owners are still humans, households, institutions, and governments. Their propensity to consume is lower than that of workers who lose jobs, but it is not zero. Dividends, capital gains, and tax receipts all fund spending.
Moreover, history shows that when job losses mount, governments typically react. In a world where unemployment genuinely approaches 10% due to visible technological change, it is difficult to imagine central banks and treasuries remaining idle. Policy tools—such as automatic stabilizers, direct transfers, wage subsidies, public employment, and even variants of basic income or job guarantees—can convert "Ghost GDP" into purchasing power if voters demand it.
This does not guarantee a smooth transition; politics can certainly misfire. However, treating a demand collapse as an unavoidable mechanical outcome of AI, rather than a contingent product of choices, overstates the inevitability of Citrini's crisis scenario.
Why the Crisis Narrative Still Holds Relevance
If "The 2028 Global Intelligence Crisis" is, as Smith describes, "just a scary bedtime story," it remains a useful one. It forces individuals to contemplate a world where AI genuinely disrupts white-collar work and exposes the fragility of certain software- and fee-based business models. It demonstrates how quickly financial markets can be spooked by narrative alone, and it pushes macroeconomists and policymakers to consider distributional consequences before they materialize.
Yet, when juxtaposed with current data and a more explicit macroeconomic framework, its strongest claims appear as outliers. AI will almost certainly eliminate some jobs and transform many others. It might break specific companies and sectors, and it could even contribute to a future downturn if shocks and policy mistakes align unfavorably.
Nevertheless, the most likely path for this decade is messier and less cinematic: gradual adoption, uneven productivity gains, choppy labor-market adjustments, and ongoing policy debates over how to share the benefits. This presents ample challenges without assuming that by 2028, we will inhabit a world where machines excel, the economy falters, and only "Ghost GDP" remains.
The story serves as a valuable warning worth reading, but based on current evidence, it does not represent our baseline future. AI is undeniably set to alter who wins, who loses, and how income is distributed. It is already rewriting some balance sheets. The evidence so far suggests something more arduous and less theatrical than a 2028 collapse: a long, uneven slog where productivity gains arrive late, business models are ground down rather than blown up, and politics ultimately determines whether "Ghost GDP" remains a metaphor or becomes a statistic.
