Anthropic vs. Wall Street: How One Button Burned $285 Billion

A minus $285 billion is not a rounding error. It is the price of realizing that AI did not come to help office drones, but to replace them along with their desks and their software. When Bloomberg reported the drop in market value across the software and financial services sectors, one thing became clear: investors no longer buy the fairy tale of "peaceful coexistence" between old corporations and new neural networks. Anthropic launched a tool that automates contract review and legal analysis, and that pulled a trigger that had been cocked for a long time.

For a long while, Anthropic was seen as OpenAI's "younger but very ethical brother." While Sam Altman was building multimodal castles in the air, Dario Amodei's team focused on accuracy and reliability. And here is the result: its new business tool does exactly what large companies have spent years paying specialized software vendors millions of dollars to do. Why pay for an expensive subscription to a legal platform or an audit service if Claude can read a thousand pages of a contract and find critical errors in seconds? The market answered that question unambiguously, kicking off a mass sell-off.

Software developers were not the only ones hit; asset management took a blow too. Here lies the central irony: the industry that should have been first to calculate the risks of adopting AI ended up in the eye of the storm. Investors realized that a significant share of operational work in finance is routine that AI cracks like nuts. Where analyzing a portfolio or checking compliance once required staffs of analysts and complex enterprise systems, the barrier to entry has now dropped to the cost of an API call. That destroys business models built over decades.

We are watching the beginning of the end of the era of "middleware software." For the past ten years, venture capital poured money into companies that built convenient interfaces on top of databases for lawyers, accountants and financiers. Now those interfaces are becoming unnecessary. The intelligence of Claude or GPT-4 lets users work with data directly. That poses an existential threat to SaaS giants: if your product can be replaced by a single well-written prompt, you no longer have a product. You are just a redundant link in the value chain.

The $285 billion panic is only the start of a repricing of value. Money is now flowing out of companies that "use" technology and into companies that "create" it. Investors are betting on raw intelligence, not on wrappers. Anthropic has shown that even conservative fields such as law and finance are no longer a safe harbor. If you sell your employees' time or access to closed software, your business is in the blast zone.

In the coming months, expect desperate attempts by corporations to "implement AI" in their products in order to justify current share prices. The problem is that bolting a neural network onto old software often looks like attaching a jet engine to a cart. Anthropic and other LLM market leaders are building new infrastructure from scratch, and it is inherently more efficient than anything created before 2023. A big shake-out is coming, and only those who can offer something more than automation of routine work will survive.

The bottom line: the era of "software as a service" (SaaS) is rapidly turning into the era of "intelligence as a service." If your company's value lay in processing documents, be prepared for the market to price you at zero.