The Substack Citrini is one of the more well-known independent research publications. The firm was founded by current CEO James van Geelen, who is known for prescient calls in recent years regarding GLP-1 drugs and artificial intelligence (AI).
Recently, Citrini, a paid publication, published a long piece on a scenario in which AI performs better than expected and subsequently tanks the market. The piece is co-published with Alap Shah, managing partner at Lotus Technology Management, and is written as if it is 2028 and the disaster has already happened.
The original article post on X had 28 million views as of March 2. While Citrini deemed the article a scenario rather than a prediction, it drew significant attention as a well-thought-out, underexplored idea that even seemed to contribute to the market's rout on Feb. 23.Here's what investors need to know.
If AI works, it could be a bloodbath for the white-collar job market
The crux of Citrini and Shah's argument is that AI creates a "negative feedback loop" that is difficult to escape. AI capabilities and agentic AI agents will make it much easier to develop apps and software products that used to take years and require many software engineers, coders, and support roles. This will lead to mass layoffs and a surge in the unemployment rate from 4.3% now to over 10% by 2028.
Furthermore, agentic commerce agents will be integrated into everyone's devices, making the consumer efficient to a fault. So many businesses actually profit from human inefficiency, whether it's subscriptions people forget to cancel or goods and services that end up costing more because no single person can conduct an entire internet search.
Image source: Getty Images.
Agentic commerce solves for these inefficiencies while removing barriers to entry. For instance, according to the post, AI agents will be able to reexamine and switch people over to the most affordable insurance options annually, eliminating a significant number of an insurance firm's customers who passively renew their coverage without thinking twice.
AI agents will also possess a wealth of real estate knowledge, eliminating the need for real estate agents who collect a 5% to 6% commission. Agents will even look for more efficient, cheaper payment options, like stablecoins, that remove fees but hurt the payment giants with once-impenetrable moats, such as Visa and Mastercard.
The problem is that all of these fees and human inefficiencies generate revenue that pays for human labor. But with AI automating many white-collar research tasks and improving pricing efficiency, the technology will not create new jobs like past technological disruptions have. The authors wrote:
The company that sold workflow automation was being disrupted by better workflow automation, and its response was to cut head count and use the savings to fund the very technology disrupting it.
Essentially, to protect margins, companies will further lean into AI and lay off workers, which is one of the highest costs in most businesses. And this could lead companies to lay off high-paid, white-collar workers who account for a disproportionate share of consumer spending in the U.S.
Roughly 70% of U.S. gross domestic product (GDP) is powered by consumer spending, so if unemployment surges, that could dry up spending and lead to a prolonged recession, as well as significant defaults among all consumer credit categories -- even high-income borrowers with jumbo mortgages
Citrini believes the S&P 500 (^GSPC 1.14%) will decline by 38% from late 2026 to mid-2028. However, the carnage won't be as obvious in the economic data, which will be distorted by higher productivity and what economic pundits will eventually refer to as "Ghost GDP."
The future is difficult to predict, and this is more of an exercise
Again, I want to reiterate that Citrini referred to this post as "a scenario, not a prediction like most of our work." But the article presents many concepts that are plausible and should be considered as AI rapidly integrates throughout society.
I do think the productivity angle is very real and likely responsible for some of the quirks we are seeing in the labor market right now, where nobody seems to be hiring or firing workers, and unemployment is still relatively low. AI models can already create complex spreadsheets in just seconds -- think how many tasks and jobs can potentially be either streamlined or replaced with this feature alone.
There's also the question of whether the world really has the resources and power capacity to support all of this compute usage. Furthermore, since the economy is largely driven by consumer spending, a significant decline there would likely, at some point, hurt the top and bottom lines of many companies and potentially send the economy into a deflationary spiral.
The future is extraordinarily difficult to predict, and even if one predicts the right outcome, the journey and results will likely be very different from what most expect. Think about the last five years. Could anyone really have predicted nearly any of the events that happened?
Ultimately, the threat of AI should be taken seriously, and I think Citrini's piece does a nice job of jolting many people awake to the many challenges this new, disruptive technology presents.
