RPT-BREAKINGVIEWS-Salesforce's problem is not enough AI upheaval

By Pranav Kiran

TORONTO, Feb 25 (Reuters Breakingviews) - Salesforce CRM.N is stuck in a technological limbo. Fears that its core business model of selling software subscriptions will be displaced by artificial intelligence are rising. The company bets instead that machine learning will eventually redound to its benefit by allowing bots to work on its products in place of humans. That future isn't here yet, though. Disruption is too fast and too slow at the same time.

The tech giant run by Marc Benioff is making the case for AI as an opportunity rather than a threat. On Wednesday, it unveiled a new metric to investors, saying that 2.4 billion distinct tasks were done on its platform by so-called AI “agents” in the last two financial years. As it moves to arguing that it is delivering outcomes, rather than tools, the company foresees organic revenue growth accelerating in the second half of 2027. That would be a relief to shareholders who have watched the pace of expansion in annual recurring revenue slow for three years now.

Software stocks have been squeezed by two big shocks: rising interest rates since 2022 and the breakneck progress of AI labs like Anthropic. Companies handled the first hit by slashing costs to boost margins. This latest upheaval, which vaporized over $800 billion in market value of software and services companies, is trickier. Selling a new machine-learning-tuned business model isn’t entirely in Benioff’s control. It depends on customers to embrace novel ways of working. At the same time, success is double-edged, since moving to a task-based system could imperil traditional subscriptions.

It'll be a while coming yet. Research firm Gartner estimates that most of the $2.5 trillion in AI spending will be on infrastructure like massive data centers this year, meaning that capabilities are still under development. Large firms are slow to adopt new tech that could break important functions like finance and payroll.

If autonomous bots really are the future though, it’s also unclear who will derive the most value from them. So far, public and private market valuations favor the developers of large-language models like Anthropic. The platforms with which their models will interact, like Salesforce, are still arguing their case.

CONTEXT NEWS

Salesforce on February 25 said that revenue rose 10% in the year ending January 31, to $41.5 billion, in line with analyst estimates compiled by Visible Alpha data.