BREAKINGVIEWS-AI apocalypse trade has a buggy logic

By Pranav Kiran

TORONTO, Feb 4 (Reuters Breakingviews) - Fears of an artificial intelligence-led corporate apocalypse are causing market hallucinations. Large language model (LLM) developer Anthropic recently launched new tools that promise to automate tasks ranging from sales to legal and financial analysis. It caused a stock selloff on Tuesday and Wednesday for IT and professional services firms, the magnitude of which implies that some widely used products will disappear overnight. Investors are overreacting.

Incumbent software companies have had a difficult time on the stock market since ChatGPT's release in November 2022. The BVP Nasdaq Emerging Cloud Index is barely up since then, whereas AI-heavy technology giants have surged. The fear is that LLMs could make trivial work of many enterprise tasks that currently require dedicated software, like managing expenses.

With the new additions to Anthropic's Claude Cowork product, these fears are spreading. Data and publishing group RELX REL.L plunged 14% on Tuesday, while consultancies and advertising agencies like WPP felt the pain too. Thomson Reuters, the parent of Reuters Breakingviews, dropped 16%. The worry is that companies will eventually just deploy bespoke AI tools, generated through Claude or similar LLMs, rather than using external services.

Just looking at revenue growth rates, it's easy to see the cause for concern. The constituents of the BVP Nasdaq Emerging Cloud Index, for example, grew sales by 16% on average, according to Bessemer Venture Partners. Professional services firms grow typically well below even that level. Anthropic and OpenAI on the other hand, are multiplying their top lines each year. Also, Gartner estimates that IT spending on AI models will grow 1.5 times faster than on software between 2025 and 2027.

Still, some of the stock moves look too extreme. Take London-listed RELX, whose value fell by $9 billion to $55 billion on Tuesday. It sells data and analytics tools for customers covering the legal and risk management businesses, and also publishes academic journals.

According to UBS analysts, 88% of the UK group's revenue is subject to very little LLM disruption. RELX's legal database division LexisNexis, where AI risks could arise, made up only 12% of overall operating profit in the most recent full financial year. Apply a 15-times multiple to the roughly $660 million of adjusted operating profit that Bank of America analysts expect for the at-risk legal segment in 2026. The resulting $10 billion of implied value is close to the market capitalization that RELX lost. In other words, investors are assuming LexisNexis is worthless as a business division.

In fact, the pace of revenue growth in the segment has picked up from low single digits in 2019 and 2020 to mid-single digits now. RELX and its legal-services peer Thomson Reuters have rolled out AI tools trained on their respective databases. The general lesson is that valuable business data still sits with the incumbents. Nvidia NVDA.O boss Jensen Huang on Wednesday called the notion that AI will replace software and related tools "illogical". The market is yet to hear that message.

CONTEXT NEWS

AI developer Anthropic on January 30 launched plugins for its Claude Cowork product. The new additions are designed to help users automate tasks across legal, sales, marketing and data analysis.

The BVP Nasdaq Emerging Cloud Index fell 6.4% on February 3. Shares of data firms RELX and Thomson Reuters fell 14% and 16% respectively on February 3.