The $2 Trillion Selloff: How Anthropic's AI Tools Rattled Software Stocks

In just eight trading sessions, roughly $2 trillion in market value evaporated from global software and IT services stocks. The catalyst wasn’t an earnings miss, a regulatory crackdown, or a recession warning. It was a series of product announcements from a seven-year-old AI startup that convinced investors the entire software industry might be obsolete.

This is the story of how Anthropic’s Claude Cowork—and the enterprise plugins that followed—triggered the most dramatic reassessment of software company valuations since the dot-com bubble. For investors trying to separate signal from noise, understanding what actually happened, what the market got right, and what it got wrong has never been more important.

The Genesis: Claude Code’s Quiet Revolution

To understand the February 2026 selloff, you need to start with Claude Code.

Anthropic launched Claude Code publicly in May 2025 as an AI-powered coding assistant. Within six months, it had reached $1 billion in run-rate revenue—faster than ChatGPT, faster than Slack, faster than almost any software product in history. By February 2026, that figure had more than doubled to $2.5 billion.

The numbers were staggering. Business subscriptions quadrupled since January 1. Eight of the top ten companies by Fortune 500 revenue became Anthropic customers. More than 500 enterprises were spending over $1 million annually on Anthropic products.

But Claude Code was just a developer tool. What came next would target everyone else.

January 12-30: The Cowork Launch and Plugin Expansion

Week 1: The File Manager That Thinks

On January 12, 2026, Anthropic announced Claude Cowork—what the company described as “Claude Code for the rest of your work.” Unlike typical AI chatbots that answer questions and disappear, Cowork could:

Read, write, and create files on a user’s computer
Manage documents and folders autonomously
Execute multi-step projects without constant oversight
Operate in an isolated virtual machine for security

Scott White, Anthropic’s head of enterprise product, framed it as a transition from Claude being “a helpful assistant to a full collaborator.” Users could now “delegate entire projects… that are hyper-specific to your company and your role.”

For Pro subscribers, access began January 16. Team and Enterprise plans followed on January 23. The market barely noticed.

Week 2: MCP Apps Go Live

On January 26, Anthropic announced MCP Apps—direct integrations with nine popular workplace tools including Slack, Figma, Asana, and Canva, all built on the Model Context Protocol that Anthropic open-sourced in late 2024.

Now Claude could draft Slack messages, create Figma diagrams, and manage Asana projects—all without leaving the chat interface. The protocol had already been adopted by OpenAI, Google, Microsoft, AWS, and dozens of major enterprises.

Still, the market shrugged.

Week 3: The Plugins That Changed Everything

Then came January 30. Anthropic released 11 industry-specific plugins for Cowork:

Legal Plugin: Contract reviews, compliance checks, document screening Finance Plugin: Journal entries, account reconciliations, financial statement preparation Data Analysis Plugin: SQL queries, database pulls, dashboard generation Sales Plugins: CRM integration, lead tracking, pipeline management Marketing Plugin: Campaign analysis, content optimization, audience research

Anthropic open-sourced all 11 plugins and emphasized that custom plugins were “easy to build, edit, and share” without technical expertise.

For the first time, Wall Street paid attention.

February 3-4: The First Wave of Panic

Monday: The “Get Me Out” Trade

On Monday, February 3, 2026, the selloff began.

A Goldman Sachs basket of US software stocks dropped 6%—its biggest single-day decline since April’s tariff-driven selloff. An index of financial services firms tumbled nearly 7%. The Nasdaq 100 fell as much as 2.4% before trimming losses.

The damage was concentrated but devastating:

Stock	Single-Day Loss	Context
Thomson Reuters	-15.8%	Largest single-day drop on record
LegalZoom	-19.7%	Legal services automation fears
RELX (LexisNexis parent)	-14.0%	Data analytics exposure
FactSet	-10.0%	Financial data terminal concerns

Asset managers weren’t spared. Apollo, Ares, Blackstone, Blue Owl, Carlyle, and KKR all fell between 3% and 11% as investors questioned whether AI could automate the due diligence and analysis these firms charge premium fees for.

Jefferies trader Jeffrey Favuzza coined a term that stuck: “SaaSpocalypse.”

Tuesday: $285 Billion in Two Days

By close on Tuesday, February 4, the two-day rout had erased $285 billion from software, financial services, and asset management stocks. The Indian IT sector was hit particularly hard:

Infosys: Down 8%, its steepest decline in over two years
TCS: Down 6.5% to ₹3,014
Tech Mahindra: Down 5.8%
Nifty IT Index: Would ultimately fall 19% in eight trading sessions, wiping out $50 billion

In the US, the carnage continued:

ServiceNow: Down 22% since January 29
Intuit: Down 26%
Salesforce: Down 20%
Snowflake: Down 18%

The narrative was simple and brutal: if an AI tool could handle legal reviews, financial reconciliations, and CRM management, what exactly were enterprise software companies—and the armies of consultants who implement them—actually selling?

February 5-6: Opus 4.6 and the Goldman Revelation

Thursday: A More Powerful Model Arrives

Just as investors hoped the bleeding might stop, Anthropic poured gasoline on the fire.

On February 5, the company released Claude Opus 4.6, featuring:

A 1 million token context window (up from 200K)
128K max output tokens (double the previous limit)
“Agent teams”—AI agents that could coordinate to split large tasks into parallel workstreams
Top performance on the Finance Agent benchmark
A 50% time horizon of 14 hours 30 minutes on METR’s task completion test—the longest of any AI model

Financial data providers bore the brunt. FactSet dropped another 10%. S&P Global, Moody’s, and Nasdaq all saw sharp declines.

Friday: The Goldman Sachs Bombshell

On February 6, CNBC reported that Goldman Sachs had been working with embedded Anthropic engineers for six months to develop autonomous AI agents for:

Accounting for trades and transactions
Client vetting and onboarding
Compliance workflows

Marco Argenti, Goldman’s Chief Information Officer, said the bank was “surprised” at how capable Claude was beyond coding—particularly in areas combining large document parsing with rules-based judgment.

The message was clear: this wasn’t theoretical disruption. One of the world’s most sophisticated financial institutions was already building AI systems to replace human-intensive back-office operations.

Argenti called it “premature” to expect job losses. CEO David Solomon had already announced plans to limit headcount growth during the AI transition. The market read between the lines.

February 7-12: The Numbers Behind the Narrative

Anthropic’s Explosive Growth

As software stocks cratered, Anthropic’s metrics told a different story:

Metric	Figure
Annual Revenue Run Rate	$14 billion
Claude Code Run Rate	$2.5 billion
YoY Revenue Growth	10x+
Customers >$100K/year	7x growth in 12 months
Customers >$1M/year	500+
2026 Revenue Target	$20-26 billion

On February 12, Anthropic closed a $30 billion Series G funding round at a $380 billion post-money valuation—more than double its September 2025 valuation of $183 billion.

The round, led by GIC and Coatue, marked the second-largest venture funding deal of all time, behind only OpenAI’s $40 billion raise. Separately, Microsoft invested $5 billion and Nvidia invested $10 billion, with Anthropic committing to $30 billion in Azure compute purchases.

The contrast was stark: software companies losing trillions while the AI company allegedly disrupting them raised billions.

February 17: The Infosys Pivot

On February 17, one of the IT services companies hit hardest by the selloff announced it was joining the other side.

Infosys partnered with Anthropic to develop “enterprise-grade” AI agents for telecommunications, financial services, manufacturing, and software development. The collaboration would integrate Claude models with Infosys’s Topaz AI platform.

Dario Amodei, Anthropic’s CEO, was direct: “There’s a big gap between an AI model that works in a demo and one that works in a regulated industry—and if you want to close that gap, you need domain expertise. Infosys has exactly that kind of expertise.”

Infosys shares jumped 4.8%—their best day in two weeks. The message was clear: adapt or die. And adaptation meant partnering with the company causing the disruption.

February 23: The AI Scare Trade Returns

Just when markets thought the worst was over, today brought a fresh reminder that AI disruption fears haven’t gone away.

On Monday, February 23, the “AI scare trade” resumed with a vengeance. IBM shares plunged 13%—one of the steepest single-day drops in years for the legacy tech giant. Accenture and Cognizant Technology also fell sharply as investors reassessed consulting and IT services valuations.

The catalyst was a combination of factors: renewed attention to Anthropic’s enterprise automation capabilities, concerns about the upcoming Nvidia earnings (which will reveal just how much hyperscalers are spending on AI infrastructure), and broader market jitters from Trump’s 15% global tariff announcement following Friday’s Supreme Court ruling.

Wall Street analyst Adam Crisafulli of Vital Knowledge captured the mood, noting that AI—“which until recently had propelled the stock market to a succession of record highs”—is now “increasingly a net negative for the equity market.” The concern isn’t just about software anymore; it’s about whether AI will compress margins across the entire services economy.

The IBM selloff is particularly notable because Big Blue has been aggressively repositioning itself as an AI company. If the market is punishing even the companies trying to ride the AI wave, it suggests investors see disruption risk as systemic rather than selective.

The Counter-Narrative: Why Some Analysts Say It’s Overblown

Not everyone bought the doomsday scenario.

The Bull Case

JPMorgan strategists argued software stocks had “scope to rebound” because the market was “pricing in unrealistic near-term disruption.” Mark Murphy, head of US enterprise software research, called it “an illogical leap” to assume an LLM plugin could “replace every layer of mission-critical enterprise software.”

Wedbush Securities said the selloff reflected an “Armageddon scenario for the sector that is far from reality.” Dan Ives noted: “Enterprises won’t completely overhaul tens of billions of dollars of prior software infrastructure investments to migrate over to Anthropic, OpenAI, and others.”

Nvidia CEO Jensen Huang weighed in directly, calling fears that AI would replace software “illogical.” His view: AI will use existing software platforms as tools, not replace them entirely.

The Structural Arguments

Bulls pointed to several factors the market may have ignored:

Switching costs: Enterprise software is deeply embedded in workflows. Ripping out Salesforce or SAP isn’t a weekend project.
Data gravity: Companies have decades of proprietary data in existing systems. AI tools need that data to function—often through the same software being disrupted.
Regulatory requirements: Regulated industries need audit trails, compliance certifications, and accountability structures that chatbots don’t yet provide.
Integration complexity: The “plumbers of the tech world” (JPMorgan’s phrase for IT services firms) will be essential for integrating AI into existing stacks.

Early Signs of Recovery

By mid-February, some stocks began stabilizing. Firms like DA Davidson and Wedbush upgraded several names and added them to AI-focused lists. Options implied volatility remained elevated at 41%, but the freefall had slowed.

What the Market Got Right and Wrong

What the Market Got Right

The market correctly identified that AI capabilities have reached a genuine inflection point. The Goldman Sachs partnership wasn’t a pilot program—it was six months of embedded engineering with production deployment “soon.” Anthropic’s revenue growth isn’t hype; it’s $14 billion in annualized run rate.

The market also correctly identified which segments face the most pressure: legal document review, financial data terminals, basic compliance workflows, and commoditized IT services. These are the tasks Claude Cowork’s plugins target directly.

What the Market May Have Gotten Wrong

The selloff may have conflated long-term disruption risk with near-term revenue impact. Even Goldman’s Argenti said job losses are “premature” to expect. Enterprise AI adoption takes years, not quarters.

The market may also have underestimated how much software companies will benefit from AI. ServiceNow, for example, is integrating AI into its own products. So is Salesforce, Intuit, and virtually every major SaaS provider. The question isn’t whether AI disrupts—it’s whether incumbents can adapt faster than they lose.

Finally, the “all software is dead” narrative ignores specialization. An AI that can write SQL queries doesn’t replace Snowflake’s data infrastructure. An AI that can draft contracts doesn’t replace the workflow management systems that route those contracts through legal review.

Investment Implications: Navigating the New Landscape

Near-Term (3-6 Months)

Expect continued volatility. Options markets are pricing in 40%+ implied volatility for software ETFs. Any new Anthropic or OpenAI announcement will trigger sharp moves in both directions.

Watch earnings calls closely. Listen for management commentary on AI impact, pilot programs, and capex shifts. Companies that frame AI as opportunity (not just threat) may outperform.

Monitor the Infosys model. If other IT services firms announce similar partnerships, it could signal the sector is finding a path forward.

Medium-Term (6-18 Months)

Differentiate between commodity and specialized software. Basic document processing and data entry tools face real pressure. Complex workflow systems with deep integrations have more time.

Track enterprise AI adoption curves. Goldman’s deployment timeline will be instructive. If production rollouts happen faster than expected, the bear case strengthens.

Follow the consulting spend. If enterprises shift budgets from IT services to AI subscriptions, it shows up first in consulting firm bookings.

Long-Term (18+ Months)

The “picks and shovels” thesis may invert. In prior tech transitions, infrastructure providers (cloud, chips) benefited while application layers churned. This time, the application layer (AI models) may capture more value while infrastructure becomes commoditized.

Regulatory clarity will matter. AI governance frameworks are still forming. Companies that solve the compliance and auditability problems will have moats.

Winner-take-most dynamics are possible. Anthropic and OpenAI are both pursuing enterprise markets aggressively. Network effects from MCP adoption could create lock-in.

The Bigger Picture

The February 2026 software selloff isn’t just a market event—it’s a signal about how quickly investor expectations can shift when AI capabilities cross visible thresholds.

Claude Cowork’s plugins didn’t introduce fundamentally new technology. They packaged existing capabilities (document parsing, code generation, workflow automation) into formats that non-technical users could immediately understand and executives could immediately fear.

That’s the real lesson here: AI disruption isn’t always about what’s technically possible. It’s about what becomes obviously possible to people making capital allocation decisions.

For the software industry, the question is no longer whether AI will reshape their markets. It’s whether they can reshape themselves faster than investors lose patience—and faster than Anthropic ships its next product.

TL;DR

$2 trillion wiped from software stocks in early February 2026 after Anthropic released Claude Cowork plugins targeting legal, finance, sales, and data workflows
The selloff accelerated with Claude Opus 4.6’s agent teams and Goldman Sachs revealing six months of embedded AI development for accounting and compliance automation
Anthropic is now valued at $380 billion with $14 billion in annual revenue, while software stocks face ongoing volatility
Bulls argue the selloff is overdone—enterprise software switching costs, regulatory requirements, and integration complexity provide more runway than the market assumes
Investors should differentiate between commodity tools facing near-term pressure and specialized platforms with deeper moats

Sources

Bloomberg — Anthropic AI Tool Sparks Selloff From Software to Broader Market
CNBC — Goldman Sachs taps Anthropic’s Claude to automate accounting, compliance roles
CNBC — Anthropic closes $30 billion funding round at $380 billion valuation
TechCrunch — Anthropic releases Opus 4.6 with new ‘agent teams’
TechCrunch — As AI jitters rattle IT stocks, Infosys partners with Anthropic
Fortune — Anthropic launches Cowork, a file-managing AI agent
Bloomberg — JPMorgan Strategists Say AI Fears on Software Stocks Overblown
Yahoo Finance — Selloff wipes out nearly $1 trillion from software and services stocks
Business Today — Why Anthropic’s new AI tool is spooking India’s IT services industry