Salesforce just dropped its latest earnings report and the numbers look fine. Actually, they look better than fine. Revenue is up, margins are healthy, and the feared "AI wipeout" simply didn't happen. If you listened to the bears six months ago, you’d think ChatGPT was about to eat Marc Benioff’s lunch by Tuesday afternoon. Instead, the company is still growing.
But let's be real. The anxiety hasn't left the room. It’s just hiding in the corners of the balance sheet.
Investors are staring at a massive gap between what Salesforce says AI will do and what customers are actually paying for today. We’re in a strange limbo. The old world of per-seat licensing is starting to feel dusty, yet the new world of autonomous "agents" isn't quite paying the bills yet. You can see the tension in every analyst question and every executive pivot.
The disruption isn't a single event. It’s a slow erosion of the old ways of doing business.
The Revenue Reality Check
Salesforce reported a beat, sure. But look at the professional services growth. It's sluggish. This is where the first cracks usually show up. When companies start thinking they can use Large Language Models to automate their own workflows, the first thing they cut is the army of consultants they used to hire to click buttons in a CRM.
We're seeing a shift from "how many people use this software" to "how much work does this software do." That sounds great in a keynote. It's terrifying when your entire valuation is built on selling seats to human beings. If an AI agent can do the work of five sales reps, does the company still buy five licenses?
Benioff is betting everything on Agentforce. He wants you to believe that instead of paying for a tool, you’re paying for a digital employee. It's a bold pivot. It also changes the fundamental math of the enterprise software industry.
Why the Fear Persists Despite the Beats
If the results were good, why is everyone still twitchy? It’s because the barrier to entry for building software has collapsed.
Ten years ago, if you wanted to build a custom CRM layer, you needed a team of developers and eighteen months. Now? You need a smart prompt and a few weeks. Salesforce isn't just competing with Microsoft or Oracle anymore. They’re competing with "good enough" internal tools built by AI.
I've talked to CTOs who are looking at their seven-figure Salesforce contracts and wondering if they're paying for a Ferrari to drive to the mailbox. They see the AI features being added, but they also see the price tags. There's a growing sense of fatigue.
Companies don't want to pay an "AI tax" on top of their already expensive subscriptions. They want the AI to justify the cost they’re already paying. That’s a tough needle to thread.
The Agentforce Gamble
The strategy now is all about autonomy. Salesforce is moving away from the "copilot" model—where the AI sits on your shoulder and makes suggestions—to the "agent" model. These agents are supposed to handle customer service, lead qualification, and data entry without a human in the loop.
It’s a smart move. It’s also a defensive one.
By becoming the platform where these agents live, Salesforce hopes to remain the "system of record." If they own the data and the agents, they stay relevant. But this assumes that customers want their agents locked inside the Salesforce ecosystem.
Open-source models and specialized AI startups are circling. They offer flexibility that a massive, legacy platform sometimes can't match. Salesforce has the advantage of the "installed base," but inertia only carries you so far when the technology is moving this fast.
The Problem With Middle Management AI
One thing nobody is talking about is the data mess. AI is only as good as the data it digests. Most enterprise Salesforce instances are a disaster of duplicate records, outdated leads, and "dirty" data.
Giving an AI agent access to a messy CRM is like giving a high-speed blender to a toddler. It’s just going to make a mess faster. Salesforce is pushing "Data Cloud" as the fix, but that's another product to buy, another integration to manage.
The friction is real. Businesses are tired of "more." They want "better."
What This Means for Your Tech Stack
If you’re running a business, you can't ignore the Salesforce numbers, but you shouldn't let them lull you into a false sense of security. The disruption is happening at the workflow level, not the earnings level.
- Audit your seat count. Don't just renew because that's what you did last year. Look at who is actually using the software and who is just a passenger.
- Test the agents. Don't buy into the hype of a $2 trillion market. Run a small pilot. See if the "autonomous" agent actually solves a ticket or just frustrates your customers.
- Clean your data now. Whether you stay with Salesforce or move to a leaner AI-first competitor, your data is your only real asset. If it's trash, your AI will be trash too.
- Demand outcomes, not features. Stop paying for "AI-enabled" versions of things. Ask for a price that reflects the productivity gain. If the vendor can't prove it, don't pay the premium.
The big software companies are in a race to redefine themselves before the technology makes their old models obsolete. Salesforce is winning the first few laps, but the race is a marathon, and the track is still being built.
Watch the margins, not just the headlines. When the cost of serving an AI agent drops but the price to the customer stays high, that's where the real profit—or the real vulnerability—lives.
Stop waiting for a "Big Bang" moment where AI replaces everything. It's a slow grind of optimization. You either lead that optimization or you get optimized out of your own budget. Start by looking at your most repetitive manual process in your CRM today. If you haven't automated that by next quarter, you're already behind the curve.
