The Accidental Monopoly-Killer

• Anthropic’s product launches wiped an estimated $2 trillion from software stocks in under eight weeks — the steepest AI-driven selloff in market history.

• Three sectors hit in 30 days: Claude Cowork triggered a $285B single-day SaaS rout. Claude Code Security crashed cybersecurity stocks 8–18%. Claude Opus 4.6 sank financial data firms by double digits.

• Anthropic then defied the Pentagon over autonomous weapons, was blacklisted — and saw Claude hit #1 on the App Store as users rallied behind the company.

• Revenue hit $14B (10x annual growth for 3 consecutive years). Raised $30B at $380B valuation. Eight of the Fortune 10 are Claude customers.

• Below: the mechanics, the market impact, the contrarian case, and your action plan.

At 9:31 AM on February 3, one minute after the market opened, a trader on the Jefferies equity desk watched Thomson Reuters drop 8% in a single candle. Then LegalZoom. Then the London Stock Exchange Group. By 9:45, $285 billion in software market cap was gone. He turned to a colleague and said something that would become the term of the year: “It’s a SaaSpocalypse.”

The trigger was a minor product update. Anthropic had released 11 open-source plugins for its Claude Cowork AI suite — tools for legal discovery, tax accounting, sales prospecting, and customer support. Not a new model. Not a breakthrough. Plugins. But what traders saw wasn’t a feature announcement. They saw the moment AI stopped augmenting software and started replacing the humans who use it.

What followed over the next 30 days was the most consequential month in enterprise AI history. Three product launches. Three sectors rattled. A standoff with the President of the United States. And the fastest-growing enterprise company ever recorded coming out the other side stronger than it went in.

By the Numbers

Three Shockwaves in 30 Days

Shockwave 1: The SaaSpocalypse (January 30 – February 6)

The plugins themselves were modest — tools that let Claude handle contract review, NDA triage, compliance workflows, and sales prospecting. But the market’s reaction wasn’t about what the plugins could do today. It was about what they implied for tomorrow.

The fear wasn’t that Claude would replace Salesforce next quarter. It was subtler and more devastating: if AI agents do the work, companies need fewer people. Fewer people means fewer software seats. Fewer seats means the entire SaaS pricing model — built on per-user subscriptions — breaks.

“If 10 AI agents can do the work of 100 sales reps, you don’t need 100 Salesforce seats anymore. You need 10.”

— Jason Lemkin, SaaStr founder

The numbers told the story. Thomson Reuters plunged 16%. LegalZoom dropped 20%. The London Stock Exchange Group fell 13%. By mid-February, the iShares Expanded Tech-Software Sector ETF (IGV) had fallen 30% from its September 2025 peak — its worst performance since the 2008 financial crisis. Forward earnings multiples for the entire software sector collapsed from 39x to 21x, levels not seen since the mid-2010s. Piper Sandler downgraded Adobe, Freshworks, and Vertex, citing what analysts called “seat compression” and “vibe coding” narratives.

Shockwave 2: The Cybersecurity Flash Crash (February 20)

To understand why this shockwave hit differently, you need to understand the mechanism.

Traditional vulnerability scanners — the tools CrowdStrike, Zscaler, and Fortinet sell — work by matching code against libraries of known patterns. They’re sophisticated, but they’re fundamentally rule-based: if the code matches a known vulnerability signature, they flag it. This approach catches known bugs efficiently. But it’s blind to novel logic flaws — the kind where two individually safe components interact in a dangerous way that no one has catalogued.

Claude Code Security doesn’t scan for patterns. It reasons. It examines how components work together, how data flows through an application, and where the logic itself breaks down. This is the difference between a spell-checker and an editor who understands your argument. And in internal testing, Opus 4.6 identified over 500 high-severity vulnerabilities in production open-source projects — bugs that had survived decades of expert human review.

That’s what spooked the market. A pattern-matching tool is a complement to human security teams. A reasoning-based tool is a potential replacement for parts of them. Traders priced it accordingly.

“It’s a mini-flash-crash, triggered by fear of disruption.”

— Dennis Dick, Head Trader, Triple D Trading (via Bloomberg)

CrowdStrike plunged up to 18% in the days following the launch, wiping out $20 billion in market capitalization alone. Palo Alto Networks, Fortinet, Cloudflare, and Zscaler all fell 9% or more. IBM recorded its biggest daily drop since 2000. The Global X Cybersecurity ETF hit its lowest level since November 2023.

Shockwave 3: The Pentagon Standoff (February 24 – March 2)

On February 24, Defense Secretary Pete Hegseth summoned Dario Amodei to the Pentagon and delivered an ultimatum: remove Claude’s restrictions on mass domestic surveillance and autonomous weapons, or face consequences. The meeting lasted less than an hour. Amodei held firm.

On February 27, Trump ordered every federal agency to cut ties with Anthropic within six months. Hegseth designated the company a “supply chain risk” — a label never before applied to an American company. Within hours, OpenAI CEO Sam Altman announced a Pentagon deal claiming the same red lines Anthropic had fought for. OpenAI employees had signed an open letter supporting Amodei’s stance just days earlier.

The cancelled contract was worth $200 million — roughly 1.4% of Anthropic’s $14 billion revenue run rate. The consumer response dwarfed the loss. Claude rocketed to #1 on the App Store. Free users surged 60% since January. Paid subscribers doubled since October. On Monday, March 2, the servers buckled under what Anthropic called “unprecedented demand.”

Principles, it turns out, sell.

The Case Against Panic

Here’s where we push back on our own narrative.

The $2 trillion selloff is real. The fear is not entirely irrational. But some of the sharpest minds in technology think the market has overcorrected — dramatically.

“This notion that the software industry is in decline and being replaced by AI is the most illogical thing in the world, and time will prove itself.”

— Jensen Huang, CEO, NVIDIA

Huang’s argument: AI agents won’t reinvent software tools — they’ll use them. ServiceNow, SAP, and Salesforce aren’t going to be replaced by Claude. They’re going to be operated by Claude. The tools survive; the human operators get augmented or reduced.

The data supports caution over panic. Gartner’s February 2026 forecast projects worldwide software spending will grow 14.7% in 2026 to more than $1.4 trillion — that’s roughly $180 billion in net new software spending in a single year. Companies aren’t abandoning software budgets. They’re reallocating them. Wedbush called the selloff an “Armageddon scenario that is far from reality,” noting that enterprises won’t overhaul trillions of data points embedded in their software infrastructure overnight.

And SaaStr founder Jason Lemkin — whose seat-compression math we cited above — explicitly argues this isn’t the death of SaaS. It’s a repricing of growth expectations that were already overdue, compounded by AI fear. The comparison he draws: the 2016 software crash, where LinkedIn dropped 44% and Tableau dropped 50% — and both recovered within months.

The truth, as usual, lives in between. AI isn’t going to kill enterprise software next quarter. But it is structurally changing who uses it, how it’s priced, and what justifies a premium multiple. The companies that survive will be the ones that integrate AI deeply enough to expand wallet share from fewer users — not the ones pretending nothing has changed.

The Power Realignment

First, Anthropic is destroying the old software business model. Claude Cowork, Claude Code Security, and Opus 4.6 aren’t competing with individual SaaS vendors. They’re undermining the premise that enterprise work requires specialized software at all. The $2 trillion selloff isn’t about market share — it’s about category collapse.

Second, it’s building a trust moat that compute can’t replicate. Defying the President is spectacularly risky. But in a market where users increasingly choose AI providers based on values, Anthropic just created differentiation that no amount of training compute can match.

Third, AI companies are now geopolitical actors. When the Pentagon labels an AI company a supply chain risk and the President orders a ban, these companies have crossed from the technology sector into the power sector. This changes everything about how enterprises should evaluate, procure, and depend on AI.

How Key Players Responded

What This Means for Your Career

Jack Dorsey didn’t fire 4,000 people because Block was struggling. He fired them because the market rewarded it. The stock surged 24% overnight. He then predicted most companies would reach the same conclusion within a year. Whether that’s accurate or performative, the signal it sends to every CEO in America is unambiguous: AI-driven headcount reduction is now a shareholder value event.

This changes the career calculus. The question is no longer “will AI affect my job?” It’s “will my CEO cite AI when they restructure?” Those are different questions with different timelines. The first depends on AI capability. The second depends on market incentives. And as of last Thursday, the market incentive is live.

The roles that survive restructuring aren’t the ones that produce the most output. They’re the ones that produce output AI can’t. Judgment calls under ambiguity. Stakeholder alignment across conflicting interests. Creative direction that requires taste, not just competence. If your day is mostly tasks a well-prompted AI could handle, the displacement audit in this week’s newsletter isn’t optional — it’s urgent.

The career moat is shifting from “I can do this work” to “I can orchestrate AI doing this work while managing the decisions AI can’t make.” That’s not a slogan. It’s a concrete skill set: knowing which model to use for which task, building fallback workflows, evaluating AI output quality, and making the judgment calls that sit between what AI generates and what ships to clients.

90-Day Predictions

1. Anthropic will launch a product that directly replaces at least one major SaaS category. The trajectory is clear: Cowork for workflows, Code Security for cybersecurity, Opus 4.6 for financial analysis. Legal research or HR software is next. Watch for a dedicated Claude plugin that makes an entire incumbent’s core product redundant — and for the stock selloff that follows within hours.

2. At least two more S&P 500 companies will announce Block-style AI restructuring. Dorsey’s playbook worked. Cut headcount, cite AI, get rewarded by markets. The 24% stock surge is the most powerful incentive signal a CEO can receive. Expect copycats in professional services, financial services, and media within 90 days.

3. The Pentagon designation will be challenged in court — and Anthropic will win. The designation lacks clear statutory authority, Anthropic’s restrictions haven’t affected a single government mission, and OpenAI secured functionally identical terms days later. Anthropic will likely be back on classified networks by Q3 with stronger contractual protections.

4. Multi-vendor AI strategy will become an enterprise procurement requirement. The Pentagon blacklisted Anthropic on Friday. Claude crashed on Monday. Two vendor disruptions in four days. By mid-2026, any CIO presenting a single-vendor AI strategy will face the same question: what happens when your provider disappears overnight?

Your Action Plan

Whether you’re an individual contributor, a team lead, or a CIO — this month’s events demand immediate action. Work through this checklist this week.

This Week (March 3–7)

Run the AI Displacement Audit (see this week’s newsletter prompt) on your team’s tasks. Map every recurring workflow to automation risk, specific AI tools, and the human element that survives.

Set up a secondary AI provider. If you use Claude, register for OpenRouter or get a Gemini API key. Test your three most critical prompts on the backup.

Write a fallback doc for your most AI-dependent workflow. Include primary tool, backup, switch time, data needed, and owner. Then test it — actually run one task cycle on your backup.

This Month (March)

Audit your team’s SaaS subscriptions. For each tool, ask: could Claude Cowork or a competing AI agent handle 80% of this workflow? If yes, start a pilot.

Brief your leadership on the Anthropic-Pentagon situation and what it means for vendor risk. Frame it as business continuity, not politics.

Identify the 3–5 tasks in your role that become more valuable when everything around them is automated. Invest your development time there — that’s your career moat.

This Quarter (Q2 2026)

Renegotiate SaaS contracts. The $2 trillion selloff gives you leverage. Vendors are nervous. Push for AI-inclusive pricing, shorter terms, and agent-compatible APIs.

Build a cross-functional AI resilience team. Include IT, procurement, legal, and operations. Meet monthly. Their job: ensure the company can survive any single AI vendor disappearing overnight.

Start building your orchestration skill set. Learn to evaluate model quality for different tasks, build multi-model workflows, and make the judgment calls that sit between AI output and what ships to clients.

Go Deeper

→ Anthropic Series G announcement (full revenue data): anthropic.com/news/anthropic-raises-30-billion-series-g-funding-380-billion-post-money-valuation

→ TechCrunch on the SaaSpocalypse: techcrunch.com/2026/03/01/saas-in-saas-out-heres-whats-driving-the-saaspocalypse

→ Claude Code Security details: anthropic.com/research/claude-code-security

→ Gartner 2026 IT spending forecast: gartner.com/en/newsroom/press-releases/2026-02-03-gartner-forecasts-worldwide-it-spending

→ Jensen Huang on the software selloff (CNBC): cnbc.com/2026/02/26/nvidia-jensen-huang-gpu-ai-threat-software-companies

→ SaaStr analysis — “The 2026 SaaS Crash: It’s Not What You Think”: saastr.com/the-2026-saas-crash-its-not-what-you-think

This deep dive accompanies the iPrompt newsletter for the week of March 3, 2026. If someone forwarded this to you subscribe below to get these every week.

Stay curious — and stay paranoid.

— R. Lauritsen