The SaaSpocalypse: the market just caught up

On Tuesday, Anthropic released a legal plugin for Claude. Not a breakthrough product. Not a fine-tuned model trained on decades of case law. A set of structured prompts that automates contract review, NDA triage, and compliance tracking.

Thomson Reuters lost 18% of its market value in a single day. The worst drop in the company’s history.

RELX, the parent of LexisNexis, fell 14%. Its worst day since 1988. LegalZoom dropped 20%. DocuSign, 11%. Wolters Kluwer, 13%.

By the close, $285 billion had been wiped from software stocks in a single trading session. As of today, the seventh straight session of losses, the damage has surpassed $950 billion. The S&P 500 Software & Services Index is down 26% from its October peak.

The broad market is near all-time highs. This isn’t a general selloff. The market is surgically repricing an entire category of company.

Jefferies called it the “SaaSpocalypse.”

We predicted this in January. SaaS is dead. The market just caught up.

But the market is only seeing the surface.

A folder of prompts

The part that should concern you isn’t the scale of the selloff. It’s what triggered it.

Anthropic’s legal plugin is not a specialized legal reasoning engine. It’s Claude, the same model available to anyone with a subscription, wrapped in workflow instructions. The company was careful to note that outputs should be reviewed by licensed attorneys. It’s practically a disclaimer page with a chatbot attached.

And it was enough to trigger the largest single-day collapse in Thomson Reuters’ history.

A folder of prompts just did what no competitor could do in two decades of trying. It made the market question whether Thomson Reuters’ core business has a future.

This isn’t about one plugin or one company. It’s about what the plugin represents: the application layer is collapsing into the model layer. When Thomson Reuters builds CoCounsel on top of OpenAI, they’re adding value above the model. When Anthropic ships its own legal plugin, the model becomes the product. The middleman disappears.

The narrative just flipped

For the past two years, the consensus story went like this: AI helps software companies. They integrate AI features, charge more per seat, grow faster. AI is a tailwind.

That story died this week.

The new story: AI replaces software companies. Why pay $50 per seat per month for a SaaS tool when an LLM does the same job for a fraction of the cost? Why license Westlaw when Claude can review your contracts?

The numbers tell the story of how fast this repricing is happening:

Schroders analyst Jonathan McMullan put it plainly: “Investors are aggressively repricing these areas as the historical ‘visibility premium’ erodes. The speed of AI advancement makes long-term valuations harder to defend, particularly as AI tools allow businesses to do more with fewer staff, threatening the traditional model of charging per software user.”

The traditional model of charging per software user. That’s the entire SaaS business model in one phrase. And it’s what’s breaking.

We wrote about this a month ago

On January 5, we published Why SaaS Is Dead. The argument was straightforward: SaaS pricing depends on software being expensive to build. If AI makes software cheap to build, the per-seat pricing model collapses. The moats that software companies spent decades building become irrelevant when a model provider can replicate their core functionality in weeks.

That was 31 days ago.

Since then, Anthropic launched Claude Cowork on January 12, shipped 11 vertical plugins on January 30, and the market lost $830 billion in six trading sessions. The timeline from product launch to industry-shaking disruption was less than three weeks.

Jensen Huang called the selloff “the most illogical thing in the world.” JPMorgan’s Mark Murphy called it “an illogical leap.” Box CEO Aaron Levie said this is “the most exciting moment we’ve ever had” while his stock sits 17% below where it started the year.

They’re defending the old narrative. The market has already moved to the new one.

Follow the money to see where this goes

While software stocks crater, Big Tech is doubling down on the infrastructure that’s causing the damage:

Collectively, Big Tech will spend over $500 billion on AI infrastructure this year. That money is being spent to build the systems that replace existing software categories. The more they invest, the better the models get, the more application-layer companies become redundant.

This is not an anomaly. This is the infrastructure buildout for a new economic era. And the companies funding it are the same ones shipping the products that kill incumbents. The structural advantage loop is now visible at a macro scale.

The canary in the coal mine

Here’s where most market commentary stops. Software stocks are down, AI is disruptive, some companies will adapt and some won’t. Standard analysis.

But James St. Aubin, Chief Investment Officer at Ocean Park Asset Management, said something that deserves more attention: “My biggest fear is that this is a canary in the coal mine for the labor market.”

He’s right. Follow the logic:

Thomson Reuters employs 26,000 people. RELX employs 35,000. Salesforce employs 70,000. ServiceNow, 22,000. These aren’t warehouse workers or fast food employees. These are highly paid knowledge workers. Engineers, salespeople, account managers, lawyers, analysts.

When these companies contract, and the market is betting they will, those jobs disappear. When those jobs disappear, the spending power of an entire economic class drops. When spending power drops, the consumer economy feels it.

This morning we published No Roles Are Safe, arguing that the disruption isn’t limited to junior roles or entry-level positions. It reaches every level of knowledge work. The SaaSpocalypse is the corporate version of the same argument. It’s not just individual roles at risk. It’s entire companies. Entire categories of software. Entire segments of the economy that were built on the assumption that software was hard to build and knowledge work required humans.

This isn’t a tech correction. It’s the market beginning to price in the most significant economic restructuring since industrialization.

Dario Amodei, Anthropic’s CEO, has written that AI could displace half of all entry-level white-collar jobs within one to five years. Marc Benioff has already stopped hiring engineers and lawyers at Salesforce. Clifford Chance, one of the world’s largest law firms, cut London staff by 10% citing AI. These aren’t predictions anymore. They’re quarterly earnings decisions.

What the counter-arguments miss

The pushback is predictable. Jensen Huang says enterprises will always need specialized tools. JPMorgan says a single plugin won’t replace mission-critical software. Thomson Reuters’ CEO says the winners will be companies with “trusted content and domain expertise.”

These arguments are reasonable in the short term and wrong in the medium term. They assume the current capabilities are the ceiling. They’re not. Claude Cowork launched January 12. The plugins shipped January 30. That’s 18 days from platform to vertical disruption.

The iteration speed of AI-native companies makes legacy software release cycles look geological.

The question was never whether AI could replace these companies. It was how fast. The market just got its answer.

What this means for enterprises

If you’re running a company that depends on SaaS tools, the calculus has changed. The vendors you rely on are facing existential questions about their business models. Some will adapt. Many won’t. The ones that survive will look fundamentally different than they do today.

We wrote in December about the existential divide between intelligence-native enterprises and legacy-structured ones. That divide is now visible in stock prices. Palantir, which is genuinely AI-native, gained 6.6% on Tuesday while the rest of the software sector burned. The market is learning to tell the difference.

The enterprises that will thrive in this environment aren’t the ones shopping for the next SaaS tool. They’re the ones rebuilding their architecture around intelligence as a core capability. Not bolting AI onto legacy systems. Not buying another vendor’s AI wrapper. Building the capacity to deploy intelligence directly, on their own terms.

The SaaSpocalypse wasn’t a surprise. It was inevitable. The only question left is how fast the rest of the economy follows.