By Belle Lin

International Business Machines said Tuesday it closed its roughly $11 billion acquisition of the data-streaming company Confluent.

The deal, first announced in December, is intended to help businesses access their data for AI agents, or bots that can take action on their own.

At the moment, corporate data is spread across multiple information-technology systems, from software applications to private data centers and cloud platforms. To make AI agents work, "you need to be able to get data wherever it is" and get it instantly, IBM Chief Executive Arvind Krishna told The Wall Street Journal.

That's why Confluent "has been something that we have been keeping an eye on for a long time," Krishna continued.

Armonk, N.Y.-based IBM's acquisition comes as technology vendors are increasingly selling tools to help businesses use, manage and create AI agents. The bots are sweeping through American corporations even as business tech leaders are struggling to oversee and safely scale those agents.

For IBM, Confluent's technology will become the "backbone" of its own platform for helping business clients access their data for various AI uses, Krishna said. That's the same tack IBM took with its $33 billion acquisition of Red Hat, which became IBM's default technology to help customers update their software applications, he said.

The acquisition is also central to Krishna's vision of positioning IBM as a leading player in hybrid cloud-computing and AI. It is IBM's second-largest deal in history, and aims to give businesses a way to make use of their old-school IT systems and data in the AI era.

The Challenges AI Poses for IBM

About three years ago, IBM began adding AI and AI agents to its internal operations which has since added roughly $4.5 billion in productivity to its bottom line, Krishna said. Over $3 billion of that total has been reinvested in research and development and other areas, he said.

At the moment, AI agents are best suited to "low-risk" areas like software development, customer support, enterprise operations, HR and payroll, Krishna said. Over time, they could be used to directly market a product to a customer.

"There is a whole set of mundane AI use cases that we can keep ourselves busy for the next two, three years just getting those done," he said.

Looking ahead, rote white-collar work will be "displaced" by work requiring greater human interaction and more decision-making, according to Krishna.

IBM itself has been sorting through the impact of AI agents on its workforce. Last year, the company said that bots had begun to do the work of a couple of hundred of its human-resources workers.

Krishna said he expects IBM will maintain or grow its head count in the next five years -- mostly in areas like consulting, sales and coding. But, roughly 20% of the company's enterprise operations and customer-support roles "could see a change" over the next two years, he added.

In February, IBM stock posted its worst decline in 25 years -- tumbling sharply on news that Anthropic had released AI tools that could help with modernizing Cobol, a programming language mainly run on IBM mainframe computers.

Krishna said IBM has its own tools to help modernize Cobol applications, and noted that its stock has since rebounded: "I think people also realized that was way exaggerated in terms of its impact," he said.

More largely, Krishna said he believes IBM is insulated from the perceived risk of AI and AI agents to software businesses. "A lot of what we do is enabling middleware, enabling software. We are not that much in application software, and so I think [volatility] is actually a tailwind for us," he said.

Write to Belle Lin at belle.lin@wsj.com

(END) Dow Jones Newswires

March 17, 2026 09:25 ET (13:25 GMT)

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