By Adam Levine

Business customers have been slow to adopt artificial-intelligence tools in a systematic way, but some of that conservatism is beginning to fade. At Salesforce -- one of the few software companies willing to open up about its AI business -- core AI revenue is small but rising quickly.

Salesforce's AI momentum could come as a surprise to investors, who have been dumping software stocks for much of the year. The iShares Expanded Tech-Software Sector exchange-traded fund is down 17% on the year. A narrative has arisen that AI will badly disrupt software, and that some of these companies, even ones that are thriving now, won't make it out alive.

AI agents are the focus of the worries. Agents are software that can use an AI language model to accomplish a complex series of tasks in much the same way a human worker would. They are still at an early stage, but, projecting forward, one can imagine a world in which a large portion of knowledge work is being done by machines, not people. And these machines may not need software in the way we think of it today.

The two primary sources of worries come from AI start-ups OpenAI and Anthropic, especially the latter. ChatGPT, OpenAI's general purpose chatbot, had a head start and is more popular with consumers, so Anthropic has focused its attention on software developers and business customers. Its two agentic products for those markets are called Claude Code and Claude Cowork.

In late January, Anthropic released Cowork tools for specific enterprise functions like finance and legal. Software stocks tumbled in the wake of those launches. But neither company has experience in selling to enterprises, and we can infer from some of their statements that it's slower going than they'd like.

"The limiting factor for seeing value from AI in enterprises isn't model intelligence, it's how agents are built and run in their organizations," began a Feb. 23 press release from OpenAI announcing a new partnership with IT consultants. OpenAI needs their expertise to sell and implement AI in enterprise markets.

Anthropic was even more blunt in a business-focused presentation last month. "2025 was meant to be the year where AI agents transformed the enterprise. But the hype turned out to be mostly premature. Many pilots started and many failed," Kate Jensen, Anthropic's head of Americas, said in a livestream. "There was a growing sense that the technology was moving faster than the ability to actually deploy it well. It wasn't a failure of effort, it was a failure of approach."

It turns out the approach can't simply cut out existing software makers and their tightknit relationships with business customers.

As OpenAI and Anthropic have indicated, we are still near the beginning of a process described by economists like Erik Brynjolfsson, who have outlined the sometimes halting progress of new technologies. They have to be accompanied by new business processes, new worker skills, and new co-inventions.

Evidence from the biweekly U.S. Census Bureau's Business Trends and Outlook Survey shows that AI uptake is still low. An average of 18% of U.S. businesses were using AI in the four 2026 surveys. But that rises to 32% for firms with 250 or more employees. The numbers have been rising slowly since the Census Bureau began asking questions about AI in 2023.

During the cloud disruption two decades ago, many incumbent enterprise software makers looked down on a host of new browser-based applications. This time around, software companies are taking the threat seriously.

Like many of its competitors, Salesforce is trying to sell agents to its customers. The company began to report annual recurring revenue for its agent software, Agentforce, in the second fiscal quarter, which ended in July. Back then it was $440 million, only about 1% of annual revenue. But when Salesforce reported its fourth quarter last week, Agentforce recurring revenue had jumped to $800 million, an 82% rise in six months.

This is still a relatively small number, but Salesforce's agent trajectory is an important metric as enterprise AI adoption creeps higher. Since Salesforce is one of the few public software companies willing to put a number on AI revenue, it's also a barometer for enterprise adoption more generally.

Salesforce stock is down 25% this year.

Write to Adam Levine at adam.levine@barrons.com

This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

March 06, 2026 16:37 ET (21:37 GMT)

Copyright (c) 2026 Dow Jones & Company, Inc.