Salesforce Is Terrified, and It Shows

When a company’s CEO mentions an existential threat six times during an earnings call, you know something is wrong.

Salesforce reported solid numbers this week. $10.7 billion in quarterly revenue, up 13% year-over-year. Guidance is strong. The fundamentals look fine. Yet Marc Benioff spent his time on stage fighting a narrative that won’t die: that AI agents are about to render the entire SaaS industry obsolete. The market has dubbed this fear the “SaaSpocalypse,” and it’s been hammering technology stocks lately.

So what did Salesforce do? It panicked. Not in a way that admits panic, of course. Instead, it threw the kitchen sink at investors.

The Full Court Press

Salesforce increased its dividend by nearly 6%. It launched a $50 billion share buyback program. It revamped the entire earnings call format into part podcast, part infomercial, with CEO interviews from satisfied customers lined up like testimonials at a revival meeting.

The SharkNinja CEO was there. The Wyndham Hotels CEO too. Even the CEO of SaaStr, the software industry conference itself, showed up to tell everyone how great Salesforce’s AI agents are. It felt less like an earnings report and more like a coordinated PR blitz.

But here’s where it gets interesting. Salesforce also did something clever: it introduced a new metric called “agentic work units” or AWU. The company logged 19 trillion tokens last quarter, which sounds massive until you realize that in the AI world, it’s barely a sneeze. So instead of bragging about tokens, Salesforce created a new way to measure whether agents actually complete tasks like writing to a database record, not just generating text.

“You can ask it a question and it can write you a poem, but that’s not really all that valuable in the enterprise world,” Salesforce’s Patrick Stokes said. Fair point.

The Architecture Wars

What’s really revealing is that Salesforce felt compelled to present its own architectural vision of the future. The company is basically saying: SaaS platforms like us will own the stack. AI model makers like OpenAI? They’ll be commoditized engines underneath.

This is a direct response to OpenAI’s architectural vision, which shows the opposite: OpenAI at the top, database companies relegated to the bottom as interchangeable engines. That vision spooked investors earlier this month when OpenAI released Frontier.

It’s architecture theater. Both companies are fighting for investor confidence about who controls the future of enterprise business software. The answer is probably: whoever can prove they actually solved a real problem for customers.

The Jacket Moment

And then there’s the detail that nobody will stop talking about: Benioff wore a black leather jacket during the call. The same signature look of Jensen Huang, Nvidia’s CEO, who’s been crushing it in the AI world.

It’s almost sad. The CEO of one of enterprise software’s most powerful companies decided to dress like someone else to send a signal to investors. That’s not confidence. That’s mimicry born from fear.

The real question isn’t whether Salesforce’s AI agents are good. The customer testimonials suggest they are. The real question is whether investors believe the company can maintain its pricing power and market position when AI agents start proliferating. Salesforce clearly doesn’t think the numbers alone will convince anyone.

Written by

Adam Makins

I can and will deliver great results with a process that’s timely, collaborative and at a great value for my clients.