Citrini Research AI scenario triggers massive selloff in software and payments stocks

How AI agents could destroy the economy

On Sunday, an analyst group called Citrini Research published a remarkable piece illustrating how agentic AI could bring on mass economic destruction over the next two years. The scenario imagines a report from two years in the future, in which unemployment has doubled, and the total value of the stock market has fallen by more than a third. As the report puts it: AI capabilities improved, companies needed fewer workers, white collar layoffs increased, displaced workers spent less, margin pressure pushed firms to invest more in AI, AI capabilities improved... It was a negative feedback loop with no natural brake...The system turned out to be one long daisy chain of correlated bets on white-collar productivity growth. It's a new kind of bear case, focused not on Skynet-style misalignment but on the gradual unspooling of the economy itself. In particular, the Citrini scenario looks at the implications of integrating AI agents into the economy at large, and what it would mean when outside contractors get replaced by cheaper in-house AI. It's similar to the Death of SaaS scenario, but Citrini goes further, implicating any business model that involves optimizing transactions between companies. As you might expect, the report is causing quite a stir online. Not everyone is buying it -- even Citrini describes it as more of a scenario than a prediction -- but it's not so easy to name the specific point where you think the scenario goes wrong. Personally, I'm not sure companies are ready to hand off purchasing decisions to AI agents, no matter how smart they are. But in Citrini's scenario, most of the impacted decisions have already been handed off to third-party contractors, so it's not quite as implausible as it seems.

Software, Payments Shares Tumble After Citrini Post on AI Risks

Delivery, payments, and software stocks slid sharply Monday after Citrini Research published a report laying out the potential risks that artificial intelligence could pose to various segments of the global economy. DoorDash Inc., American Express Co. and Blackstone Inc all slumped more than 7%. Shares of other companies name-checked in the article, including Uber Technologies Inc., Mastercard Inc., Visa Inc, Capital One Financial Corp., Apollo Global Management Inc. and KKR & Co. Inc. were all lower by at least 3%. "The sole intent of this piece is modeling a scenario that's been relatively underexplored," a preface to the article, which was published Sunday, said. "Hopefully, reading this leaves you more prepared for potential left tail risks as AI makes the economy increasingly weird." Monday's selloff is the latest in a string of AI-fueled routs that have rippled through US stocks for more than a month. Sectors from software, to wealth management and logistics have all been swept up in recent weeks as investors nervous about the potential disruptions from new AI tools have slipped into a "shoot first, ask questions later" mode. While software companies have been among the hardest hit, insurance brokers, private credit firms, cybersecurity and even real estate services stocks have all been caught up in the so called "AI scare trade." Yet, analysts, strategists and investors have also warned that many of these reactions are exaggerated and are likely overestimating any AI-related risks at this point. "It is a remarkable reaction," said Michael O'Rourke, chief market strategist at Jonestrading. "I have seen this market exhibit incredible resilience in the face of actual negative news. Now a literal work of fiction sends it into a tailspin."

Software companies face higher borrowing costs, tougher scrutiny as AI threatens businesses

Feb 23 (Reuters) - Software companies are delaying debt deals as higher borrowing costs and tougher scrutiny from lenders weigh on the sector, at a time when mounting pressure from artificial intelligence threatens their business models, industry sources said. Software firms both in the U.S. and elsewhere have already paused or postponed fundraising efforts as lenders and investors expect AI to upend the industry. These concerns have been underscored in loan markets, where spreads for risky companies have started to price in more defaults. AI jitters also affected private capital manager Blue Owl, whose shares slid after its latest move to sell $1.4 billion in assets to return money to investors. "We expect AI disruption risk to be increasingly reflected over 2026 to early 2027, particularly for lower‑quality credit sectors with elevated refinancing needs -- and more so in the U.S. than in Europe," said Matthew Mish, UBS' head of credit strategy. Leveraged loans, especially for U.S. tech companies, have begun to price modestly higher defaults. UBS expects defaults to rise 3% to 5% in a scenario of quicker market disruptions, compared with market expectations for an increase of 1% to 2%. "The disruption is going to play out over two years," Mish said. "We ultimately think that the market will price in a majority, but not all of the defaults that we're forecasting." Even those companies whose debt is deemed higher quality and less vulnerable to the impact of AI have held off on tapping markets until trading levels recover, one banker said. The market will closely watch investor reception to Qualtrics, a well-established software maker whose lenders will be in the market next month to raise a $5.3 billion acquisition financing package for its purchase of rival Press Ganey Forsta, a source familiar with the matter said. Qualtrics declined to comment. Press Ganey did not immediately respond to a Reuters request for comment. LEVERAGED LOANS The potential disruption from AI is having a bigger impact on more leveraged loan deals than high-yield bond deals, according to two bankers who declined to be identified discussing transactions. Technology industry borrowers, of which 60% are in software, account for the largest portion of leveraged loans, according to Brendan Hoelmer, head of U.S. default research at Fitch Ratings. Tech loans represent 17% of outstanding loans in the leveraged market, valued at $260 billion. Meanwhile, tech borrowers make up just 6% of outstanding high-yield bonds totaling $60 billion, Hoelmer noted. Of those, 70% are to software borrowers. A majority of the software sector's exposure is tied to lower credit ratings - with 50% of the loans holding a "B- or lower" credit rating - loans which typically denote a higher risk of default, Morgan Stanley estimates. Private credit software and services exposure is about 20%, BNP Paribas analysts estimate. U.S. stocks have also been roiled by AI, starting with investors dumping shares of software companies, then companies in sectors vulnerable to automation. The software index (.SPLRCIS), opens new tab is down 20% so far this year. Only 0.5% of outstanding software sector loans are due this year, while 6% are due in 2027, Fitch's Hoelmer said. On the high-yield side, only 0.7% of software debt is due this year and 8% in 2027, he added. Still, companies in the sector that have tried to tap U.S. debt markets have faced significantly higher borrowing costs from banks to underwrite the debt. Banks marketing the loans also are facing more skepticism from potential investors, according to the two bankers. Banks likely will ask for higher yields on new debt and deeper discounts on earlier debt, said the first banker, who declined to be named discussing specific deals. Companies will come off the sidelines when prices improve, the first banker said. Future deals are also likely to include stricter covenants, or legal protections for investors, to get done, the second banker noted. These include maintenance covenants, which force borrowers to keep their debt-to-earnings ratios below specific levels, the banker added. Several planned deals in the tech sector have been pulled or delayed since late January. European digital service provider Team.blue postponed an extension of its 1.353 billion euro ($1.60 billion) term loan from September 2029 and a repricing of its $771 million term loan, according to the first banker. Team.blue declined to comment. There are currently no leveraged loan deals for software companies, as companies and banks wait for trading levels on existing debt in the sector to recover from their losses since late January, when AI disruption fears rose. Meanwhile lower-rated companies with upcoming maturities "are likely to face greater refinancing and default risk in 2026," according to a Moody's Ratings report published in January. "I don't really see software and business services as being hot sectors for issuance over the next year," said Jeremy Burton, portfolio manager on the leveraged finance team of asset manager PineBridge Investments. "The technology is changing so quickly that you've really got to be confident." ($1 = 0.8482 euros) Reporting by Matt Tracy and Saeed Azhar, editing by Lananh Nguyen and Diane Craft Our Standards: The Thomson Reuters Trust Principles., opens new tab * Suggested Topics: * Finance Matt Tracy Thomson Reuters Reports on U.S. credit market activity, including corporate debt and credit ratings, U.S. Treasuries, commercial mortgages, and the ongoing public vs. private financing dynamic. He often reports on other topics and events in cooperation with team members across Reuters. Matt previously covered regulatory reviews and investigations into mergers and acquisitions, specifically anti-monopoly, national security, FCC and state-level investigations of some of the biggest deals since 2016. He has broken news on the government investigations into AT&T's merger with Time Warner, T-Mobile's purchase of Sprint, Bayer's merger with Monsanto, and many other multi-billion dollar combinations. Saeed Azhar Thomson Reuters Saeed Azhar is a Reuters financial journalist and part of the U.S. banking team, which covers Wall Street's biggest banks. He focuses on Goldman Sachs and Bank of America, and also writes about regional banks. Before moving to New York in July 2022, he led the finance team in the Middle East from Dubai, and also worked in Singapore, covering Southeast Asia finance.

An AI Thought Experiment on Substack Is Sending The Stock Market Spiraling

On Sunday, analysis firm Citrini Research published a thought experiment on Substack that reads like a piece of fiction. Titled "The 2028 Global Intelligence Crisis," it's a scenario set in June of that year in which unemployment has shot up to 10.2% in the U.S. due to mass AI-driven white-collar unemployment, which begins with an "initial wave of layoffs" in early 2026. GDP growth is still great, and productivity is booming in this hypothetical scenario, with AI agents creating a "ghost GDP," aka "output that shows up in the national accounts but never circulates through the real economy." In this dystopia, consumer spending has hugely declined as a negative feedback loop has been formed. "AI capabilities improved, companies needed fewer workers, white collar layoffs increased, displaced workers spent less, margin pressure pushed firms to invest more in AI, AI capabilities improved," the essay forecasts. AI disrupts software, which leads to software-backed loan defaults. The pace of AI disruption is not contained to software in this scenario: AI agents, vibe coding, and autonomous driving come for the throne of delivery apps as dozens of vibe-coded DoorDash alternatives spring up overnight. Then, agentic commerce, coupled with stablecoins, gets rid of transaction fees and upends the business models of payment processors like Mastercard and card-focused banks like American Express. "What follows is a scenario, not a prediction," the authors wrote. "Hopefully, reading this leaves you more prepared for potential left tail risks as AI makes the economy increasingly weird." The result was mayhem on the market. Software stocks, like ServiceNow, which were already on an AI-driven downward journey, slid even more. This time, accompanying them were shares of delivery giants DoorDash and Uber, as well as payments companies like American Express, Blackstone, KKR, Mastercard, Visa, and Capital One. Some investors, at least, seem convinced, likely because there was already AI-driven negative sentiment in the market to begin with. Software stocks have been battered in February in what has been called the SaaSpocalypse. Scared investors are largely citing the release of new AI capabilities and agents in Anthropic's AI work companion, Claude Cowork, as the trigger for the sell-off. Wall Street seems certain that AI is on the road to becoming so good that people won't have to rely on the numerous specialized services provided by software companies anymore, and instead just vibe code whatever they need or ask ChatGPT or Claude to perform the task. "The seemingly wide moats of these companies feel a lot more narrow today as competition from AI-created products intensifies," Ocean Park Asset Management chief investment officer James St. Aubin told Reuters earlier this month. "Perhaps this is an overreaction, but the threat is real and valuations must account for that. My biggest fear is that this is a canary in the coal mine for the labor market." But flashy online projections of complete AI takeover are not magic crystal balls into the future. Similarly, early last year, former OpenAI employee Daniel Kokotajlo published a comprehensive prediction of a swift timeline to humanity-ending superintelligence, in a report that made waves across the internet and was even referenced by Vice President JD Vance. Just last month, Kokotajlo had to walk back his claims, saying that superintelligence development was actually happening slower than he anticipated. It's important to keep in mind that both the concept of and road to artificial general intelligence, aka AI that can surpass human intelligence and ability, is already a contentious topic among experts. Tech leaders like xAI's Elon Musk, Anthropic's Dario Amodei, and OpenAI's Sam Altman claim that AGI will be achieved sometime in the next 2-4 years (all three companies reportedly have flashy and highly consequential IPO plans for this year), but a November survey of industry leaders found that the average expert disagreed with that timeline. Even if the superintelligence timeline does become reality, experts don't see eye to eye on just how much it can disrupt industries. Experts claim that the software sell-off is "illogical" and "overdone," and improving AI ability would actually lead to demand for more software and the ability to create better applications at lower prices. There is also the irony that, as investors are selling off shares of other companies rapidly because they think AI is about to get so big so fast, they are simultaneously souring on astronomical AI spending commitments made by tech giants, and voicing concerns over an AI bubble. “It seems like markets found a reason to be worried about too little AI and too much AI at the same time,†Morningstar equity market strategist Lochlan Halloway told The Guardian last week.

Citrini's AI Doom Report Leads to Tech Stock Selloff

A new report by Citrini Research has been partially blamed for a software and payments stock sell-off on Monday, where it outlined extreme scenarios in which AI could severely disrupt the economy, from wiping out a sizable share of the workforce and slashing consumer spending to threatening the $13 trillion US mortgage market. Citrini was little-known up until Monday, when its "Global Intelligence Crisis" report amassed over 22 million views on X alone and discussed how AI agents could drive corporate profits so high that human labor could become increasingly redundant and trigger a recession. The report lays out a chilling June 2028 scenario, in which the Standard & Poor's 500 is down 38% from its all-time high, unemployment is over 10%, private credit is unraveling and prime mortgages are cracking -- all while AI didn't disappoint, exceeding every expectation. Citrini said the term "Ghost GDP" could emerge, describing it as output that shows up in the national accounts but never circulates through the "real economy." "A single GPU cluster in North Dakota is generating output previously attributed to 10,000 Manhattan office workers," Citrini theorized in a potential June 2028 scenario. The result: a massive white-collar layoff, far less consumer spending and a recession, Citrini said. The macroeconomic uncertainty from AI and other issues, such as US President Donald Trump's tariffs, has not been taken well in the crypto market over the past few months, with Bitcoin (BTC) falling nearly 50% from its $126,080 all-time high in early October, while safe havens like gold continue to rise. Computing and AI company IBM saw its largest single-day drop in 25 years on Monday, tumbling 13.1% to $223.35, while Microsoft, Oracle and Accenture fell 3.21%, 4.57% and 6.58%, Google Finance data shows. Credit card platforms Visa, Mastercard and American Express also fell 4.5%, 5.77% and 7.2% as Citrini said private credit and software-backed loans would face cascading defaults. Investor anxiety was compounded by warnings from renowned risk theorist Nassim Taleb, who said AI could make some software companies bankrupt, while Anthropic said its Claude Code tool can be used to modernize software written in the COBOL language, which handles large transactions for many governments, banks and airlines. Related: How SocialFi, memecoins and AI pushed Base to the top of the L2 ladder Anthropic's findings appeared to affect IBM's share price directly, as COBOL is mostly run on IBM's systems. Citrini said the rise in agentic AI tools like Anthropic's Claude Code or OpenAI's Codex will drive the broad economic shift, reducing the need for human labor and forcing companies to reinvest savings into ever-more capable AI, essentially creating a feedback loop that accelerates workforce displacement and consumer spending decline. However, three multimillionaire tech investors recently said the high costs of deploying AI agents still don't justify replacing many humans, who can perform tasks just as well, more cheaply. Tech investor Jason Calacanis said he is spending $300 per day to run a single AI agent despite it only operating at 10% to 20% of full capacity, while Social Capital CEO Chamath Palihapitiya noted the same problem and said his AI agents need to be at least twice as productive as employees to justify the costs. Billionaire investor Mark Cuban said the economic-constraint argument of AI agents raised by Calacanis and Palihapitiya was the smartest counterargument that he had seen to AI replacing humans.

AI Unease Leads Software Firms to Pause Debt Deals | PYMNTS.com

By completing this form, you agree to receive marketing communications from PYMNTS and to the sharing of your information with our sponsor, if applicable, in accordance with our Privacy Policy and Terms and Conditions. This trend is happening amid mounting pressure from artificial intelligence, which has threatened the business model of these software companies, Reuters reported Monday (Feb. 23), citing industry sources. According to the report, software companies have already paused or held off on fundraising because lenders and investors expect industry upheaval from AI. "We expect AI disruption risk to be increasingly reflected over 2026 to early 2027, particularly for lower‑quality credit sectors with elevated refinancing needs -- and more so in the U.S. than in Europe," Matthew Mish, UBS' head of credit strategy, told Reuters. Leveraged loans, particularly for American tech companies, have begun to price modestly higher defaults, the report said. UBS expects defaults to tick up 3% to 5% in a scenario of faster market disruptions, compared with market expectations of a 1% to 2% increase. "The disruption is going to play out over two years," Mish said. "We ultimately think that the market will price in a majority, but not all of the defaults that we're forecasting." An example of the way AI appears to be disrupting the software sector played out last week, when Anthropic released a new security feature for its Claude model, and several cybersecurity firms saw their share prices fall. A report on this trend from Bloomberg News noted that investors are concerned that the ability to "vibe code," or use AI to write software code, will allow software customers to develop their own applications, weakening demand for legacy products. But as PYMNTS covered last year, research has found that vibe coding is not about to supplant human software developers anytime sooner. According to that report, researchers have shown that agentic AI models like Claude operated best when developers reviewed outputs after once key checkpoints had been reached, instead of running fully autonomous sessions. "Without those checkpoints, the models produced longer, less maintainable codebases and missed security constraints," PYMNTS added. Meanwhile, PYMNTS wrote recently about the role AI agents are playing in the software-as-a-service space. As these agents perform the duties of human employees, the connection between worker headcount and software revenue is growing weaker. "Instead of charging per user, vendors are increasingly experimenting with pricing tied to tokens consumed, workflows executed, transactions processed, or measurable business outcomes delivered," PYMNTS wrote. "The shift represents a structural recalibration of SaaS economics rather than its demise."

Software companies face higher borrowing costs, tougher scrutiny as AI threatens businesses

Feb 23 (Reuters) - Software companies are delaying debt deals as higher borrowing costs and tougher scrutiny from lenders weigh on the sector, at a time when mounting pressure from artificial intelligence threatens their business models, industry sources said. Software firms both in the U.S. and elsewhere have already paused or postponed fundraising efforts as lenders and investors expect AI to upend the industry. These concerns have been underscored in loan markets, where spreads for risky companies have started to price in more defaults. AI jitters also affected private capital manager Blue Owl, whose shares slid after its latest move to sell $1.4 billion in assets to return money to investors. "We expect AI disruption risk to be increasingly reflected over 2026 to early 2027, particularly for lower-quality credit sectors with elevated refinancing needs -- and more so in the U.S. than in Europe," said Matthew Mish, UBS' head of credit strategy. Leveraged loans, especially for U.S. tech companies, have begun to price modestly higher defaults. UBS expects defaults to rise 3% to 5% in a scenario of quicker market disruptions, compared with market expectations for an increase of 1% to 2%. "The disruption is going to play out over two years," Mish said. "We ultimately think that the market will price in a majority, but not all of the defaults that we're forecasting." Even those companies whose debt is deemed higher quality and less vulnerable to the impact of AI have held off on tapping markets until trading levels recover, one banker said. The market will closely watch investor reception to Qualtrics, a well-established software maker whose lenders will be in the market next month to raise a $5.3 billion acquisition financing package for its purchase of rival Press Ganey Forsta, a source familiar with the matter said. Qualtrics declined to comment. Press Ganey did not immediately respond to a Reuters request for comment. LEVERAGED LOANS The potential disruption from AI is having a bigger impact on more leveraged loan deals than high-yield bond deals, according to two bankers who declined to be identified discussing transactions. Technology industry borrowers, of which 60% are in software, account for the largest portion of leveraged loans, according to Brendan Hoelmer, head of U.S. default research at Fitch Ratings. Tech loans represent 17% of outstanding loans in the leveraged market, valued at $260 billion. Meanwhile, tech borrowers make up just 6% of outstanding high-yield bonds totaling $60 billion, Hoelmer noted. Of those, 70% are to software borrowers. A majority of the software sector's exposure is tied to lower credit ratings - with 50% of the loans holding a "B- or lower" credit rating - loans which typically denote a higher risk of default, Morgan Stanley estimates. Private credit software and services exposure is about 20%, BNP Paribas analysts estimate. U.S. stocks have also been roiled by AI, starting with investors dumping shares of software companies, then companies in sectors vulnerable to automation. The software index is down 20% so far this year. Only 0.5% of outstanding software sector loans are due this year, while 6% are due in 2027, Fitch's Hoelmer said. On the high-yield side, only 0.7% of software debt is due this year and 8% in 2027, he added. Still, companies in the sector that have tried to tap U.S. debt markets have faced significantly higher borrowing costs from banks to underwrite the debt. Banks marketing the loans also are facing more skepticism from potential investors, according to the two bankers. Banks likely will ask for higher yields on new debt and deeper discounts on earlier debt, said the first banker, who declined to be named discussing specific deals. Companies will come off the sidelines when prices improve, the first banker said. Future deals are also likely to include stricter covenants, or legal protections for investors, to get done, the second banker noted. These include maintenance covenants, which force borrowers to keep their debt-to-earnings ratios below specific levels, the banker added. Several planned deals in the tech sector have been pulled or delayed since late January. European digital service provider Team.blue postponed an extension of its 1.353 billion euro ($1.60 billion) term loan from September 2029 and a repricing of its $771 million term loan, according to the first banker. Team.blue declined to comment. There are currently no leveraged loan deals for software companies, as companies and banks wait for trading levels on existing debt in the sector to recover from their losses since late January, when AI disruption fears rose. Meanwhile lower-rated companies with upcoming maturities "are likely to face greater refinancing and default risk in 2026," according to a Moody's Ratings report published in January. "I don't really see software and business services as being hot sectors for issuance over the next year," said Jeremy Burton, portfolio manager on the leveraged finance team of asset manager PineBridge Investments. "The technology is changing so quickly that you've really got to be confident." (Reporting by Matt Tracy and Saeed Azhar, editing by Lananh Nguyen and Diane Craft)