The AI scare trade continues to rattle investors following another sharp sell-off in software stocks this week. Key companies like Workday (WDAY), CrowdStrike (CRWD) and Datadog (DDOG) have led the retreat. This follows the publication of a ‘doomsday’ AI report.
The old ‘Big Blue’, or IBM (IBM) posted its steepest daily drop in more than 25 years (-13%) after Anthropic’s Claude AI code tool could be used to modernise COBOL, a programming language run on IBM systems. COBOL powers thousands of legacy IT systems around the world. These are in banking, insurance, cybersecurity and governments.
Global Intelligence Crisis
The ‘2028 Global Intelligence Crisis’ report by Citrini Research painted a bleak, hypothetical future. In this scenario, rapid AI adoption triggers mass layoffs, credit stress and a deep equity drawdown by mid-2028.
‘What follows is a scenario, not a prediction’ [Citrini’s bold emphasis]. ‘This isn’t bear porn or AI doomer fan-fiction. The sole intent of this piece is modelling a scenario that’s been relatively underexplored’, reads the introduction.
The post paints a picture of a future in which AI disrupts lots of different kinds of white-collar work and service-industry business models. This would happen in industries like software, finance, business and legal services, and more. Moreover, this disruption causes an economic crisis.
Citrini’s scenario, explicitly framed as an exercise rather than a forecast, imagined the S&P 500 falling 38%. Unemployment would surge above 10% and private credit markets would unravel. This would occur as AI-driven productivity shocks outpace the economy’s ability to adjust.
It came a couple of weeks after Matt Shumer, an AI entrepreneur, published his ‘Something big is happening’ essay. This essay also looked at the huge disruptive influence AI is already having on financial markets, businesses and societies.
Scary bedtime story
We’ve been here before. In 1984 Cyberdyne Systems’ AI-powered Skynet was predicted to enslave humans and create a new world dominated by machines. That may have been fiction – the premise of the original Terminator movie, but people have been having nightmares about rapid change for centuries. In 1712, Thomas Newcomen’s first steam engine (a steam-powered water pump) created a similar wave of fear and uncertainty about humanity’s place in the order of things.
People have a natural inclinantion to dwell on the dark side. A few years ago, my team was asked by a national tabloid to model an investment portfolio strategy based on the aftermath of a meteorite the size of the one that wiped out the dinosaurs hitting earth. Cue plenty of eyerolling, bemusement and side-splitting laughter… it’s hard to take seriously portfolio positioning in the face of human oblivion.
Recent market action makes clear that capital is being repriced not around near-term earnings but structural implications of which we have limited understanding.
In the meantime, analysts from many of the largest financial institutions have been trying to bring some calm, thoughtful analysis to this important topic. For example, HSBC published a note this week that states its belief that software will remain the core mechanism. This will be how the world’s largest companies adopt AI, not at the enterprise software industry’s expense despite rapid advances in foundation models.
Analyst Stephen Bersey wrote that ‘software will be the primary mechanism for the diffusion of AI across the world’s largest enterprises’. He added that 2026 marks ‘the kick-off for monetisation within software.’
HSBC believes foundation models are ‘inherently flawed’ and unsuitable for ‘lift-and-replacement’ of major enterprise platforms. While they may work for narrow uses such as image creation or small apps, the bank said this is ‘not realistic for the majority of high-fidelity enterprise class platforms.’
Artificial but not intelligent
Yardeni Research goes further. It a response to Citrini’s report, it pushed back against the growing hype around AI, arguing that despite rapid advances in large language models (LLMs), the technology still lacks true understanding.
‘We still believe that AI is artificial but not intelligent’, Yardeni wrote in a Tuesday note. He added that while the output of LLMs may sound sophisticated, ‘these models don’t have a clue about what words actually mean.’
The firm said that equity markets have at times appeared to price in a more ominous scenario in which AI becomes a ‘Frankenstein monster’, though it does not share that extreme view. Instead, Yardeni believes that AI is more likely to boost productivity than eliminate large swaths of jobs.
‘We continue to believe that AI is augmenting workers’ productivity rather than making them extinct’, it wrote.
Analysts at Wolfe Research have also waded in, with chief economist Stephanie Roth urging investors to consider a more balanced path. Roth acknowledged Citrini’s thought experiment highlights legitimate concerns.
‘The argument drew attention because it highlighted a real risk: what happens if AI adoption moves faster than the economy’s ability to adjust?’ she wrote. Still, she emphasised that the same forces underpinning the bearish case could produce a very different macro outcome.
In Wolfe’s base narrative, the early phase of AI adoption between 2025 and 2026 does look unsettling. Hiring freezes emerge in white-collar industries. Productivity jumps and profit margins temporarily widen, creating the impression that gains are flowing disproportionately to capital. Tech firms trim hiring targets, consulting groups automate junior tasks and financial companies lean more heavily on AI-driven research tools.
Short-lived fears
As adoption broadens through 2026 and into 2027, competitive forces begin to erode early margin windfalls, Roth predicts. What initially appears as excess profitability increasingly shows up in lower prices, faster services and new product offerings. Premium AI capabilities quickly become standard features across enterprise software and customer workflows.
Under this path, the macro picture remains far more benign than feared. Wolfe’s scenario sees unemployment rising only modestly to about 4.5%, while inflation cools to roughly 1.8% year-on-year by May 2028. Rather than creating a demand shock, the productivity surge behaves more like a positive supply shock.
The labour market also adjusts more gradually than the bearish narrative suggests.
Returning to Matt Shumer, his essay states a few things he knows:
‘I know this [AI] isn’t a fad. The technology works, it improves predictably, and the richest institutions in history are committing trillions to it.’
‘I know the next two to five years are going to be disorienting in ways most people aren’t prepared for. This is already happening in my world. It’s coming to yours.’
And, perhaps most importantly, Shumer says the people who will come out of this best are the ones who start engaging now — not with fear, but with curiosity and a sense of urgency.
It is worth remembering that Citrini Research makes money by taking short positions on stocks and markets before publishing negative reports. Sometimes, short attacks provide an invaluable service to investors (Wirecard, Enron, Adani, Carillion), flagging financial malpractice. Other times they do not.
This is part of the financial markets rough and tumble. Maybe Citrini has given investors the excuse they needed for a healthy correction in the face of the frothier valuation multiples in some parts of the market.
We may all know a little more after Nvidia’s (NVDA) highly anticipated Q4 and full year 2026 earnings after the bell this evening (25 Feb).
You might also like:
