Is AI Going to Crash the Stock Market? SaaSpocalipse Explained.

Summary

• A new AI tool called Claude Cowork launched in January and wiped-out hundreds of billions of dollars in stock value in weeks. People panicked.

• Two viral posts, one viewed 80 million times, warned that AI is about to take everyone's jobs and crash the economy.

• Those writers aren't completely wrong. But they're also not completely right.

• The software companies most likely to survive are ones that have real protection: they work in heavily regulated fields, they're so deeply built into how businesses run that switching them out is a nightmare, or they own data that AI can't just copy off the internet.

• Smart economists say AI will actually create more jobs than it destroys over the next five years. But the transition will be painful for some people.

• For investors: don't panic sell everything. But definitely think harder about what you own and why.

Imagine you've been the best note-taker in school for years. You charge classmates $5 per set of notes. Then one day, an app comes out that takes perfect notes automatically, for free. Your business is in trouble. Now imagine that app doesn't just take notes. It writes essays, does research, it creates customized study guides and index cards, generates tailored practice tests, and, eventually, perhaps takes the exam. All overnight. All for almost nothing. That's basically what happened in January when Anthropic, the company behind an AI called Claude, launched a new product called Cowork. And the stock market completely freaked out. The shares of many software companies offering Software-as-a-Service (Saas) crashed. This phenomenon was quickly nicknamed SaaSpocalipse.

SaaSpocalipse: How Bad Was the Panic?

Pretty bad. Here's what happened to some well-known companies in just a few weeks:

• A major index of software stocks lost 21% of its value.

• Salesforce, a giant software company that helps businesses manage customers, dropped 28%.

• Intuit, which makes TurboTax, fell 34%.

• On one single day, $285 billion in stock value disappeared.

Then two writers made things worse. Matt Shumer, a tech entrepreneur, wrote an essay that 80 million people read, comparing this AI moment to the early days of COVID, when most people didn't realize yet how much things were about to change.² Then a newsletter by Citrini Research wrote a fictional story set in 2028, where AI had caused 10% unemployment and the stock market had crashed 38%.³ Michael Burry, the investor from "The Big Short," shared it. More panic followed.

Are They Right to Be Scared?

Partly. Here's what's actually true: AI is getting really good, really fast. It can now complete tasks that would take a human expert five hours, and that ability is doubling every few months. Some jobs and some companies are genuinely in danger.

Think about any business that makes money by doing something complicated that most people find annoying or confusing. Parsing the fine print in a contract. Comparing insurance prices. Processing paperwork. AI can do all of that faster and cheaper than humans. If your whole business model is built on being the middleman for those kinds of tedious tasks, you have a real problem. Economists call this the "tollbooth economy," businesses that collect fees for navigating complexity. AI just found a way around the tollbooth.

But Here's What the Panic Gets Wrong

The ominous articles assume that humans just sit there and do nothing while AI takes over everything. That's not how technology has worked in history. Think about the calculator. When calculators became cheap and common, you might have expected all the jobs that involved doing math to disappear.⁴ Instead, more people started doing more math, because suddenly it was easy and affordable. Accounting firms got bigger. Financial analysis expanded. New jobs were created that didn't exist before. Economists call this the demand expansion effect: when something gets cheaper, people find more ways to use it.

There's also a natural speed limit on how fast AI can take over. If every company tries to replace its workers with AI at the same time, the demand for the computers that run AI explodes. That makes AI more expensive. At some point, keeping a human employee actually becomes the cheaper option again. The system naturally slows itself down.

Nobel Prize-winning economist Daron Acemoglu, who has studied this carefully, says that only about 5% of work tasks can realistically be automated by AI over the next ten years.⁵ That's significant, but it's not the apocalypse. The World Economic Forum predicts 92 million jobs lost by 2030, but 170 million new ones created.⁶ Net positive. Though, yes, the transition will be rough for people in the middle.

Which Software Companies Are Actually Safe?

Not all companies are equally at risk. The ones most likely to survive have what investors call "moats," real protections that make them hard to replace. There are three big ones.

The first is regulation. Some industries are so heavily governed by laws that AI can't just walk in and take over. A doctor diagnosing a patient, a lawyer signing off on a contract, a financial advisor managing someone's retirement fund: all of these require a licensed human who can be held legally responsible. AI can help with the work, but it can't take the legal blame. Companies operating in healthcare, financial compliance, and legal services have this protection built in.⁷

The second is being deeply embedded. Some software is so woven into how a company operates that switching it out would cost more time and money than it's worth. Think about a hospital that has used the same patient records system for 20 years. Replacing it would take years of work, massive retraining, and enormous risk. These companies are sticky in the best possible way.⁸

The third is owning unique data. AI models are only as smart as the data they learn from. Companies that own data nobody else can get, decades of financial records, proprietary research, real-time market data, have something AI desperately needs and can't just find on Google. That data actually becomes more valuable the more powerful AI gets, not less.⁹

What Should Investors Actually Do?

The sell-off has created an opportunity, but only for people willing to think clearly instead of react emotionally. Stay away from or reduce exposure to software companies that do simple, repetitive tasks with no special data and no regulatory protection. If an AI demo can replace what they do, so can a competitor in a year or two.

Take a closer look at companies that have real moats, regulation, deep integration, or unique data, but got sold off anyway just because they're in the "software" category. Some of these are now priced like they're going out of business, when they're actually well-protected.

Keep an eye on companies that provide the underlying infrastructure AI needs to actually work like data storage, security, compliance tools, and the payment systems that process real money. AI agents still need pipes to run through.

The Bottom Line

Yes, AI is changing things fast. Yes, some businesses and some jobs are going to struggle. The writers who scared the market aren't making things up. But the doomsday scenario requires humans to just give up and stop adapting. And that has never happened in the history of technology, not with the printing press, not with electricity, not with the internet. People figure it out. Companies evolve. New industries emerge.

We're in a rough transition period. Not the end.

References

¹ "The SaaSpocalypse: Nasdaq Hits Year Lows as Anthropic's Claude Cowork Dismantles the Software Moat," FinancialContent/MarketMinute, February 6, 2026. https://markets.financialcontent.com/stocks/article/marketminute-2026-2-6-the-saaspocalypse-nasdaq-hits-year-lows-as-anthropics-claude-cowork-dismantles-the-software-moat

² Shumer, Matt. "Something Big Is Happening." shumer.dev, February 2026. https://shumer.dev/something-big-is-happening

³ Citrini Research and Alap Shah. "The 2028 Global Intelligence Crisis." Citrini Research Substack, February 22, 2026. Summarized in: "A Viral Substack Post Is Tanking the Stock Market," Fintool News, February 23, 2026. https://fintool.com/news/citrini-ai-apocalypse-software-selloff.

⁴ Levie, Aaron, as cited in "Jevons Paradox for Knowledge Work," Torq Software Reading List, January 20, 2026. https://reading.torqsoftware.com/notes/software/ai-ml/enterprise-ai/2025-12-28-jevons-paradox-ai-knowledge-work-oversight-bottleneck.

⁵ Acemoglu, Daron. "The Simple Macroeconomics of AI." Economic Policy, vol. 40, no. 121, 2025, pp. 13-58. NBER Working Paper: https://www.nber.org/papers/w32487. Accessible summary: "A Nobel Laureate on the Economics of Artificial Intelligence," MIT Technology Review, February 25, 2025. https://www.technologyreview.com/2025/02/25/1111207/a-nobel-laureate-on-the-economics-of-artificial-intelligence.

⁶ World Economic Forum. Future of Jobs Report 2025. Geneva: WEF, 2025. Figures cited in: "AI Job Displacement Statistics (2026 Data & Trends)," Click Vision, February 2026. https://click-vision.com/ai-job-displacement-statistics.

⁷ For context on AI's limited ability to assume fiduciary and licensed liability, see: "Anthropic Updates Claude Cowork Tool," CNBC, February 24, 2026. https://www.cnbc.com/2026/02/24/anthropic-claude-cowork-office-worker.html.

⁸ Gartner Research Note, February 2026, as cited in "Anthropic's Claude Triggered a Trillion-Dollar Selloff," Fortune, February 6, 2026. https://fortune.com/2026/02/06/anthropic-claude-opus-4-6-stock-selloff-new-upgrade.

⁹ The proprietary data moat thesis is discussed in: "Jevons Paradox for Knowledge Work: AI Abundance vs. Human Oversight Bottleneck," Torq Software Reading List, January 20, 2026. https://reading.torqsoftware.com/notes/software/ai-ml/enterprise-ai/2025-12-28-jevons-paradox-ai-knowledge-work-oversight-bottleneck.