Two-thirds of top SaaS companies won't survive the AI age, an analyst says. Here's how to tell who will win and lose.

• AI is transforming software-as-a-service, just as SaaS disrupted traditional software companies.
• Infrastructure companies like Twilio can thrive as AI transforms call centers and communication.
• AI-native SaaS firms face challenges, though companies like Navan have been successful so far.

The famous VC Marc Andreessen once said that software would eat the world. Now, AI is coming for software, and a lot of today's biggest SaaS companies may not survive the meal.

Out of the top 20 software companies in 2000, less than half of them survived the cloud era, says Pat Walravens, analyst at Citizens. AI could weed out top SaaS companies in the same way, he says. Walravens predicts that two-thirds of these companies will not emerge from the AI era alive.

What AI is doing to software-as-a-service (SaaS), or cloud software, is similar to what SaaS did to older software companies years ago, Walravens said.

Some of the top software companies in the early 2000s, such as Sun Microsystems, Computer Associates, PeopleSoft, and Siebel Systems, have since been acquired and lost relevance as cloud software companies took over the industry.

"No one went bankrupt," Walravens said. "Eventually, they got acquired or got rolled into something else."

Which SaaS companies will win or lose?

Walravens says software can be sorted into two categories: infrastructure and applications.

Infrastructure includes back-end tasks like data management and communications.

"Generally speaking, the infrastructure names are going to have a better time of it," Walravens said.

He pointed to cloud communications companies Twilio, whose stock soared last week after reporting its fastest growth rate in three years, and Bandwidth. That's because companies are rushing to replace their call centers with AI.

"AI makes software easier to build and replace, but it makes the underlying infrastructure more essential," Bandwidth CEO David Morken told Business Insider in a statement. "This is because the 'intelligence' of an AI agent is only as good as its connection to the real world."

Likewise, Twilio's chief product officer, Inbal Shani, told Business Insider that the company "has always operated at the infrastructure layer, and that is becoming more critical as AI moves into real customer interactions every day."

It's harder for application companies to survive in the AI era, Walravens said. Application companies sell software or apps that perform specific tasks for companies, such as payroll processing, task tracking, or IT support ticket management.

Application-based software companies, including SAP, Workday, ServiceNow, Atlassian, and Adobe, have seen their stock prices decline sharply over the past year (though Atlassian's stock saw a boost since reporting earnings last week).

Walravens said that these companies have parts of their business that can be disrupted by AI, while others are AI-native. The question is whether the AI-native parts can offset the losses on other parts of the business.

"They all have a dynamic where there is some part of the business, which is a leaky bucket," Walravens said.

The companies that could remain unscathed by the AI era have likely been adopting AI for years before the SaaSpocolypse, Walravens added. For example, he said Navan has been disrupting SAP's Concur travel booking software and travel agencies because it has been "very AI-forward from the beginning."

"The old corporate travel tools are basically dinosaurs — they haven't changed much since the '90s," Michael Sindicich, president of Navan, told Business Insider in a statement. "They're still clunky, offline, and built on outdated software. Navan is a fundamentally different offering because we built our product with AI from day one, which means we can deliver superior value and experiences for both travelers and finance teams."

A litmus test for SaaS

There's a quick, easy litmus test investors can apply, Walravens said. If companies see their revenue growing faster organically, then it's a good sign that their AI strategy is working.

Two other tests are its business model and the ease with which a company's products can be recreated. Walravens said that a consumption-based pricing model will generally work better than a seat-based model, which is how SaaS companies typically charge for their products.

And if your company's product can be easily vibe-coded with Claude, OpenAI, or Cursor's coding tools, "you're in a really bad spot," Walravens said. He said that's why companies like Asana and HubSpot are under so much pressure.

"There are so many ways you can vibe code a decent, simple solution," Walravens said.

An Asana spokesperson said that "AI makes our operating system more foundational, not less. We're the layer that gives agents the context to act with strategic intent, the controls to do it safely, and the multiplayer collaboration needed to work across enterprise teams."

They added that customers like Morningstar have achieved over $600,000 in efficiency gains using Asana.

"AI is making it easier to build software, and HubSpot is embracing that shift," HubSpot's chief product and technology officer, Duncan Lennox, said to BI in a statement.

Something more complicated, like a supply chain application, will be harder to vibe code.

Customers still have a lot of trust in many of these large SaaS companies and rely heavily on them, especially since they store lots of sensitive company data.

"If these enterprises can get the AI solutions they want from an existing vendor, they're going to use the existing vendor," Walravens said. "There's less risk."

Have a tip? Contact this reporter via email at rmchan@businessinsider.com, or Signal at rosal.13. Use a personal email address, a nonwork WiFi network, and a nonwork device; here's our guide to sharing information securely.