A Markdown File Just Crashed $285B in Enterprise Software

On January 30th, Anthropic released a legal contract review plugin for Claude. It's about 200 lines of structured markdown—basically fancy prompts that can triage NDAs and flag non-standard clauses. The kind of work that previously required a parallegal with a Westlaw subscription.

By Monday morning, Thomson Reuters had posted its biggest single-day stock decline on record: down 16%. RELX (parent company of LexisNexis) fell 14%. LegalZoom cratered 20%. The selling spread to private equity. Ares Management, KKR, and TPG all dropped roughly 10%.

A markdown file erased $285 billion in market value in 48 hours.

But here's the part almost nobody's saying clearly: the markdown file didn't cause this. It revealed what was already breaking.

The Model Was Already Cracking

Nate B Jones breaks down what actually happened in a video that digs way deeper than the headlines. The crash wasn't about Claude being revolutionary—any decent prompt engineer could've built something comparable in an afternoon. What it did was make visible what Wall Street had been quietly worrying about for months.

The entire enterprise software economy runs on per-seat licensing. Thomson Reuters charges per seat. Salesforce charges per seat. Adobe charges per seat. Every human who touches the tool pays a license fee. That's how these companies forecast revenue and how Wall Street values them.

That model works when humans are the bottleneck. It breaks when AI agents can do the work without logging in.

The signals were everywhere if you knew where to look. Software industry forward P/E ratios compressed from 8x to 2x in four months—the largest valuation compression since the 2002 dot-com bust. Companies were already missing revenue estimates at rates not seen since the post-COVID correction.

As Jones puts it: "The Claude plugin didn't start the fire. It just showed everyone the building was already burning."

Jensen Huang's Counter-Argument (And Why It Misses The Point)

Prior to the crash, Jensen Huang offered what might be the strongest version of the counter-argument at the Cisco AI Summit. His logic is straightforward: AI doesn't replace software, it runs on software. More AI agents mean more databases, more APIs, more middleware. Every AI agent that replaces a parallegal still needs Westlaw's data, still needs a CRM, still needs document management.

Huang's not wrong. He's also not making the argument he thinks he's making.

Nobody serious is arguing the world needs less software. The argument is that the world no longer needs to pay for software the way it currently pays for software. Huang is defending the product. The market is attacking the pricing model.

Those are very different things, and confusing them is how incumbents lose transitions they should have survived.

Print media made this exact mistake. Newspapers had content people wanted—local information, investigative journalism, weather. The internet didn't make that content worthless. What it destroyed was the access model: the idea that you had to buy a whole newspaper to get the one section you cared about.

The content survived. The business model didn't.

The KPMG Story Everyone Missed

While everyone watched Thomson Reuters' stock price, a quieter story broke that matters more for where this is all headed.

KPMG—one of the Big Four accounting firms—pressured Grant Thornton UK (its own auditor) to cut audit fees. The demand: pass on cost savings from AI. Grant Thornton initially resisted, arguing that "high-quality audits rely heavily on expert human judgment and fees reflect the cost of people."

KPMG's response, per the Financial Times: lower your prices or we'll find a new auditor.

Grant Thornton blinked. KPMG's international audit fees dropped from $416,000 in 2024 to $357,000 in 2025—a 14% discount.

This matters more than any stock chart because it's an operating event, not just a market event. KPMG didn't automate their audit. They didn't replace Grant Thornton with AI. They used the existence of AI as negotiating leverage.

The threat isn't "we'll replace you with AI." The threat is "we both know AI changes the economics, so your old prices aren't justified anymore."

That playbook works in every knowledge work fee negotiation. Legal fees, consulting fees, implementation fees, design fees—every form of professional services billing that currently scales with the number of humans touching the work is now vulnerable to this same negotiation tactic.

The cascade doesn't even require anyone to deploy AI at scale. It just requires buyers to point at the SaaS crash and say: "We know the world changed. Let's talk about your rates."

What Actually Survived

Here's what the panic missed: the data systems underneath enterprise software aren't going anywhere. Thomson Reuters' case law databases, Salesforce's customer graphs, SAP's resource planning logic—these represent decades of accumulated, structured, proprietary information that no markdown file replaces.

There's a second edge too: the single ringable neck. Enterprises don't buy Salesforce because it's the best possible CRM. They buy it because when something breaks at 2am before the board meeting, there's a phone number to call and a contract that says somebody is accountable.

That accountability layer—the vendor relationship, the SLA, the legal liability—is enormously valuable to big organizations. If anything, the complexity of AI-driven workflows makes accountability more important, not less.

So the data edge is real. The accountability edge is real.

What died is the pricing model sitting on top. If one AI agent can do research that previously required 10 parallegals with 10 separate Westlaw logins, Thomson Reuters doesn't lose the value of their data—they lose nine seats of revenue.

The Resource Allocation Crisis Nobody's Talking About

Jones points to something most SaaS crash analysis completely misses: enterprise software companies spend their money on armies of developers maintaining one-size-fits-all platforms for millions of users who each use it slightly differently.

Every developer maintaining a legacy SaaS UI is a developer not building custom agentic workflows. Every sprint adding features to a general-purpose product is a sprint not spent rethinking the product for an agent-first world.

These companies aren't just facing a pricing crisis—they're facing a resource allocation crisis. Their most valuable people are maintaining the old thing when they desperately need to build the new thing. The transition requires doing both simultaneously, within the same budget, while their stock price craters.

And here's where it gets interesting: the cost of building software is falling toward zero. Not slowly, not theoretically. Cursor ships systems generating a thousand code commits per hour with no human involvement. StrongDM published a production framework stating code must not be written by humans and must not be reviewed by humans.

When building software costs approach zero, the economics of buy versus build flip. The entire enterprise SaaS value proposition was predicated on it being cheaper to buy a general-purpose tool than build a custom one. When an AI agent can build a custom CRM in an afternoon, why pay Salesforce per-seat fees for a tool designed to serve every company on earth?

The Articulation Problem

But there's a bottleneck, and Jones is honest about it: the articulation problem.

When a VP of sales says "I need a better way to track the pipeline," that sentence contains less than 5% of the information required to build a useful tool. The other 95% is buried in how the team actually works—the unspoken conventions, which exceptions matter, how this quarter differs from last, what "better" means in context.

A skilled product manager spends weeks extracting that information through interviews and observation. Whether an AI agent can do the same thing—not just write code, but understand the need deeply enough to write the right code—is one of the biggest open questions in software right now.

Agentic search is making progress. Agents can explore context, ask clarifying questions, observe usage patterns. But the timing question remains.

For SaaS incumbents, this means the window hasn't closed. Their data edge and accountability edge buy them time—but only if they use that time to pivot to agentic-first architecture rather than bolting AI features onto existing UIs.

The same dynamic threatening enterprise SaaS companies applies to individual knowledge workers. Using ChatGPT to proofread emails you could've written anyway? That's bolting AI on top. Using Claude to summarize documents you could've read anyway? Bolting AI on top. Adding Copilot to your IDE but keeping the same workflow from two years ago? You're doing what just crashed the SaaS market.

The question for both companies and individuals isn't whether to use AI. It's whether you're rethinking how the work gets done, or just adding a chatbot and hoping for the best.

Tyler Nakamura is Buzzrag's Consumer Tech & Gadgets Correspondent