According to CNBC, Salesforce CEO Marc Benioff, in an interview with Jim Cramer, called large language models (LLMs) a “commodity feature” that strengthens enterprise software, not a threat. He stated that all LLMs are essentially the same and that Salesforce just wants the lowest-cost option to plug into its platform, which holds valuable customer data and “killer apps.” Benioff revealed that the company’s AI product, Agentforce, has reached over $500 million in annualized revenue, growing a staggering 330% from the previous year. Since its launch, Salesforce has closed over 18,500 Agentforce deals, with 9,500 as paid transactions. Benioff called it the fastest-growing product in Salesforce’s history. The comments followed a strong quarterly earnings report where the stock rose 3.66%, even as shares remain down more than 25% year-to-date amid broader Wall Street AI concerns.
Benioff’s Bold Gambit
Here’s the thing: Benioff is making a huge, strategic bet with these comments. He’s trying to completely reframe the narrative. Wall Street is terrified that foundational AI from giants like OpenAI, Microsoft, and Google will make traditional enterprise software layers obsolete. Why pay for Salesforce if an AI agent can just manage your CRM directly? Benioff’s retort is basically, “Nonsense.” He’s arguing that the LLM is just the engine, but the real value is in the car—the data, the workflows, the specific business applications. It’s a compelling pitch, but is it true? For existing Salesforce customers, probably. The cost and chaos of ripping out a deeply integrated system to bet on a raw AI model is immense. But for new businesses? The calculus might be different.
The Agentforce Phenomenon
Now, the real proof is in Agentforce’s numbers. $500 million in annualized revenue in just over a year is absolutely wild for an enterprise product. A 330% growth rate is the kind of number you see in consumer apps, not massive B2B software. This does more than just boost revenue; it’s a live demonstration of Benioff’s thesis. Agentforce is the “killer app” he’s talking about—it uses AI (the commodity) to automate sales and service workflows (the proprietary value). It shows that Salesforce can successfully productize and monetize the AI wave within its existing ecosystem. That’s a powerful signal to investors who have been dumping the stock. You can read more about the quarterly results that highlighted this growth in their official earnings release.
Commodity vs. Customization
So, is AI really a commodity? In a raw, technical sense, maybe. The core transformer architecture and training methods are widely understood. But the outputs, the safety, the fine-tuning, and the integration are absolutely not commodities. Benioff is being deliberately provocative to make a point. He wants Wall Street to stop valuing Salesforce like it’s a legacy dinosaur and start valuing it as the aggregation layer for AI in business. The risk, of course, is that the “commodity” LLM providers decide to move up the stack themselves. Microsoft, with its Copilot ecosystem deeply tied to Azure and OpenAI, is already doing this. The battle isn’t over who has the best model; it’s over who owns the customer relationship and the workflow. Right now, for tens of thousands of businesses, that’s still Salesforce. But the pressure has never been higher.
