The Week Anthropic Tanked the Market and Pulled Ahead of Its Rivals

As you may have noticed I use the pro level version of ClaudeAI, Anthropic's main product. BAM!

The Week Anthropic Tanked the Market and Pulled Ahead of Its Rivals

Once a distant second or third in the AI race, the company is moving to the front with a focus on caution, coding and business clients

By Bradley Olson, WSJ

Feb. 5, 2026

Anthropic’s recent product enhancements, including industry-specific add-ons and advanced models, caused a global stock selloff across various sectors.

Anthropic once appeared as an also-ran in the chaotic race for AI supremacy. This week, the sophistication of the startup’s products upended the stock market.

A simple set of industry-specific add-ons to its Claude product, including one that performed legal services, triggered a dayslong global stock selloff, from software to legal services, financial data and real estate. Then, Anthropic unveiled Super Bowl ads that taunt rival OpenAI.

On Thursday, Anthropic unleashed its most advanced model yet, capable of synthesizing data and analysis, running teams of coding assistants, and functions akin to product management. Shares of software companies including Salesforce, Intuit and others fell again Thursday, although less precipitously than earlier in the week.

The viral moment for Anthropic’s models “is the most important thing that’s happened in AI since ChatGPT’s launch,” said Dean Ball, a senior fellow at the Foundation for American Innovation who writes an artificial-intelligence newsletter. “It’s infinitely interesting.”

Industry prognosticators and AI evangelists have spent months foretelling the toll Anthropic and other sophisticated AI tools would exact on software-as-a-service companies that were darlings of the previous internet era. Tools made by such companies as Workday, Monday.com and Adobe have become the digital backbone of much of corporate America.

Anthropic’s tools, which include so-called agents that can act autonomously to carry out increasingly complex user requests for hours, have offered a preview of the threat sophisticated AI models pose to entire companies. The startup has fought its way to the forefront of the AI conversation with a novel strategy focused on safety, software engineering, and business customers.

The result: a broader business reckoning that has left corporate leaders asking what it will mean when an AI system can easily replicate expertise developed over a lifetime of coding or, in the case of companies, years of corporate development.

Tech titans such as Nvidia Chief Executive Officer Jensen Huang and others have warned that the correction is an overreaction, because many platforms are far more complex and intricate than spinning up simple websites and apps. Many companies also give priority to investing in their businesses rather than building internal software tools, analysts have said.

Still, tech giants including Microsoft, Amazon.com, Meta Platforms, Oracle and Alphabet’s Google are planning more than $600 billion in 2026 capital spending, an amount that approaches the 2026 spending budget of Japan and exceeds that of Germany and Mexico. Anthropic—and all the ways the world might use its tools—are crucial drivers of that spending.

Playing it safe

Co-founded in 2021 by Dario Amodei, a former Google researcher who left OpenAI after a feud with CEO Sam Altman, Anthropic maintains a focus on safety that is often credited as a primary reason it initially fell behind OpenAI.

The company delayed publicly releasing its AI models in 2022 out of a concern they would lead to an AI arms race. That November, OpenAI kicked off the AI boom when it unveiled ChatGPT, and Anthropic followed suit a few months later.

Early on, Anthropic focused on business customers and software engineering, reasoning that establishing itself in those areas would provide a foundation for stable growth.

For years, it looked as though its decision to delay the launch would keep Anthropic permanently behind OpenAI, which saw its monthly active users soar as millions of people began experimenting with its flagship model, which was presented in chatbot form. Anthropic’s focus on safety has played an important role in its recent resurgence.

The company pioneered a technique called “reinforcement learning from AI feedback” to ensure that its models avoid harmful or unsafe answers. Human feedback on the quality and safety of AI answers helped build ChatGPT. Anthropic took steps to use AI systems to check AI answers, with humans providing guiding principles.

One advantage of its reinforcement-learning method is that it would help avoid human bias, but it has also sped up Anthropic’s ability to improve its models, company executives have said.

Software engineering was at the center of its appeal to business customers, and company executives saw it as an eventual bridge to additional capabilities. A model that can master coding can do relatively anything on a computer.

Early releases wowed software engineers, with some saying the latest models enabled them to do projects in weeks that might have otherwise taken a year. Usage metastasized from there, spreading to noncoders and tech enthusiasts who raved about how it could take control of a computer, use a web browser and complete tasks far afield from coding.

“Anthropic has been going for the enterprise game, which they discovered is all about coding,” said David Hsu, the chief executive of Retool, a startup that helps companies build and manage internal AI tools.

Pulling ahead

By some measures, Anthropic has pulled ahead in the business market. Data from expense-management startup Ramp shows that Anthropic in January dominated so-called API spending, which occurs when users access an AI model through a third-party service. Anthropic’s models made up nearly 80% of the market in January, the Ramp data shows.

Some broad surveys of companies in the past six months suggest OpenAI has more enterprise users.

Anthropic’s success in enterprise has put it on a course to turn a profit earlier than OpenAI. The Claude maker expects to break even for the first time in 2028, two years before OpenAI, which is facing ballooning costs because of its computing needs, The Wall Street Journal reported.

The Journal’s parent company, News Corp, has a content deal with OpenAI.

OpenAI, which initially focused mainly on consumers, remains the top AI company in terms of users, and the company released a suite of powerful products in recent weeks that are capable of feats similar to those of Anthropic’s most-advanced tools. On mobile devices, ChatGPT had nearly 900 million monthly active users, more than the combined total from Google’s Gemini, Anthropic’s Claude, xAI’s Grok and Chinese models from DeepSeek and Alibaba, according to Sensor Tower.

Google, which managed a comeback of its own with the release of its Gemini 3 model, has narrowed the gap, and the company said this week that it now has more than 700 million monthly active users.

The race is far from over, and AI improvements are coming so quickly that the winner in one week or month might fall behind, said venture capitalists, executives and AI researchers.

This year, companies are likely to approach AI spending with far more discipline, making winners and losers hard to predict, said Neeraj Agrawal, a general partner at the venture firm Battery Ventures, which isn’t an investor in Anthropic or OpenAI.

“We’re in peak AI experimentation mode,” he said