AI Is Disrupting Software Companies—but Not as Fast as Many Feared

Not too long ago, Salesforce CRM -4.72%decrease; red down pointing triangle and other corporate-software companies were getting punished by investors convinced that their expensive off-the-shelf products would soon be replaced with cheaper AI-concocted versions.

That hasn’t happened. It isn’t likely to happen any time soon, either.

There was some logic behind the swoon. Shares of Salesforce, which sells software with which companies manage their sales and marketing functions, fell about 30% early this year after the release of surprisingly adept new coding tools from Anthropic and OpenAI.

It was becoming clear then that one of AI’s early strengths was going to be software development. In a podcast last year, Jared Friedman, a managing partner at startup-incubator Y Combinator, said a quarter of its current crop of companies were already generating almost all their code with AI. The “SaaSpocalypse,” a play on the acronym for software as a service, became the watchword for the meltdown.

While it is true that AI is disrupting corporate software, the incumbents are mounting a defense. And they have a lot more staying power than it might seem.

One reason is that most software development isn’t about writing code, where AI excels. It is about maintaining, updating and improving on existing software.

AI has made it easier and cheaper for companies to create new programs and features from scratch. But AI coding agents can’t yet easily be tasked with keeping software up-to-date and adjusting it to new business priorities.

The incumbents’ deep integration with their customers’ IT systems provides another bulwark against AI disruption. Most large companies have a tangle of software from different providers that they’ve knit together over time. To replace existing software with their own AI-generated code, companies must handle integrating it themselves, which is no easy engineering task.

“Most buyers of technology are not engineering organizations—they are operators,” said Arun Chandrasekaran, a Gartner analyst who focuses on AI. “And there is a world of difference between building something versus operating something.”

Sheer inertia is also giving software giants more time to adapt to AI’s march. OpenAI Chief Executive Sam Altman said on a recent podcast that disruption of the software industry hadn’t happened as quickly as he thought it would. A big part of the reason was that habits don’t easily change, he said.

“People keep doing the same things they’re doing,” he said. “They keep buying from the same company. They keep wanting to use their tools in the same way.”

Even as their shares got pummeled, software companies have been showing their financial strength.

Salesforce last month reported better-than-projected sales and profits and raised its outlook, citing an AI tailwind. It also debuted an expanded partnership with Anthropic to integrate its Claude AI model with Salesforce’s software. Initially, a plug-in will allow Salesforce users to apply Claude’s reasoning capabilities to their corporate data and automate certain sales functions, the companies said. For Anthropic, the deal brings Claude “into the systems where much of the world’s commercial activity happens,” Anthropic CEO Dario Amodei said in a statement.

That provides a preview of what could become a model for collaboration with AI heavyweights where Salesforce remains relevant. Its shares rose more than 22% the day following its earnings.

Other big software companies have also reported strong results this summer. ServiceNow, which helps companies handle IT support and automates business tasks, reported better-than-expected sales growth in July. Snowflake, a software and data-storage company, cited AI in lifting its full-year sales forecast last week. Human-resources software player Workday also reported strong results last month as adoption of its AI agents grew.

Of course, corporate software’s resilience doesn’t mean it can’t be disrupted. One fear is that Anthropic and OpenAI may see corporate software as lucrative new ground for themselves—especially as they court investors in their coming IPOs with grand growth visions.

Gil Luria, an analyst at D.A. Davidson, said in a recent note that corporate software might be too small for Anthropic to dip into given that it already generates more revenue than Salesforce. Working with Salesforce may help it keep customers from embracing a bigger threat—cheaper, open-weight models that are gaining traction in businesses.

For software’s incumbents, the biggest challenge ahead may be new software-as-a-service companies that are built from the ground up with AI in mind. That, Chandrasekaran suggested, could lead software incumbents to accede to client demands for shorter contract terms and pricing that is tied more directly to business goals.

Against that backdrop, Salesforce’s stock looks like an especially good bargain. Even after its post-earnings rise, its shares are trading at only around 16 times forward earnings—well below its 10-year average of 43 times, and the 24-times average for an S&P index of software companies.

AI is making the corporate-software business more difficult. That is entirely different than killing it off.