Inside the AI Doomsday Memo That Wiped Trillions Off Wall Street in 48 Hours

A blog post started making the rounds among analysts and hedge fund managers late on a Sunday night, when most of Wall Street was calm and traders were getting ready for the coming week. It didn’t appear to be conventional financial research at first. The essay felt more like a work of speculative fiction, like a message from the future speculating on what the world might be like in June 2028. However, markets responded as though it were a genuine warning within 48 hours.

According to the report, which was released by Citrini Research, a Substack finance monthly operated by analyst James van Geelen, artificial intelligence had developed so quickly that significant portions of the software economy were crumbling due to their own inefficiency. According to the memo, AI systems could write almost all of the code needed to create new items. Software development would become nearly frictionless in that scenario.

Products that used to require months of labor and hundreds of engineers may now be created in a matter of hours. There would be a significant decrease in the entry barrier for new businesses. And competition might suddenly take off if that barrier vanished. It sounded like science fiction. However, Wall Street’s response pointed to a more profound issue.

Several of the companies listed in the memo saw immediate pressure when markets opened the next morning. As investors processed the situation, shares of companies ranging from fintech to enterprise software, including DoorDash, Visa, Mastercard, ServiceNow, and Blackstone, decreased. It must have been weird to watch the trade screens that morning. Suddenly, a hypothetical thought experiment had turned into a financial reality.

The memo took care to present the story as a scenario. Van Geelen and his partner, software entrepreneur Alap Shah, never asserted that the events outlined will undoubtedly occur. Rather, they offered a fictitious macroeconomic research paper that was produced from the viewpoint of 2028. However, fiction has a peculiar way of revealing actual fears.

Artificial intelligence has long been hailed by investors as the next major driver of economic expansion. AI promises increased productivity, automation, and efficiency across all sectors. In an attempt to profit from this shift, businesses have invested billions in machine learning technology. However, the Citrini document presented an alternative viewpoint.

What happens if AI gets overly effective?

In the hypothetical 2028 scenario, AI technologies may virtually instantaneously create whole software platforms. The software sector experiences pricing wars as a result of this capability, which has a terrible economic impact.

Large corporate contracts are a major source of revenue for traditional enterprise software companies. These agreements can be worth millions of dollars a year and frequently endure for several years. However, customers suddenly gain leverage if AI tools enable businesses to develop comparable software internally. They may make threats to develop their own fixes.

Alternatively, businesses may look to new rivals who employ AI to introduce almost comparable items in a much shorter amount of time. One of the main factors influencing high software valuations, difference, may disappear, according to the document. After reading the article, investors appeared to rapidly envision the repercussions.

Software companies that formerly benefited from premium pricing may now be up against fierce competition. Margins would decrease. Growth forecasts may appear implausible at any time. And there might be huge repercussions if that occurs throughout the technology industry.

That week, when I strolled around Midtown Manhattan, where financial businesses are housed in glass skyscrapers above busy streets, I heard discussions about the memo in cafés and on elevators. The analysis was written off by some traders as overblown conjecture. Others didn’t seem so sure. After all, AI coding tools are already developing quickly.

It is no longer speculative to have programs that can produce useful software code from straightforward cues. They are frequently used by engineers to expedite development work. The document imagined a time when such tools would become significantly more powerful, therefore extending that tendency by a few years. The authors may have purposefully tended to be excessive.

Dramatic research notes intended to spark discussion have a long history in the financial markets. However, those provocations might occasionally highlight more fundamental structural issues. In this instance, the issue is whether technology firms based on intellectual property can continue to control prices in a world where software can be swiftly replicated by algorithms. The note lists businesses from a variety of industries.

DoorDash works in the logistics of food delivery. Global payment networks are dominated by Visa and Mastercard. Enterprise workflow software is offered by ServiceNow. Large financial asset pools are managed by Blackstone. Although their business strategies are different, they have one crucial factor in common: reliance on digital platforms that produce steady income streams.

Investors want to know if AI affects such platforms’ economics. However, markets occasionally overreact to extreme scenarios for a cause.

Technology revolutions can take longer to materialize than anticipated. Businesses still need security, dependability, regulatory compliance, and customer support—elements that cannot be instantaneously automated—even if AI is able to write complex code. In times of market worry, that subtlety tends to vanish.

There was a sense that investors were facing a new form of uncertainty when they observed the response during those two turbulent days. It’s not only a question of whether AI will transform industries, but also of how soon.