Salesforce is trying to prove Agentforce spend becomes software revenue—its FY27 guide pins the answer to the subscription line

IBM’s filing gives the bluntest version of the argument: customers are “redirected” toward AI infrastructure (servers, storage, memory) to secure scarce capacity, and that behavior shows up as software deal slippage.

That matters because enterprise software investors are not debating whether AI is real—they’re debating when AI spend converts into subscription and support revenue versus when it stays outside the software P&L.

In other words, IBM is not claiming AI destroys software demand; it’s claiming the path from AI intent to paid software seats gets warped when hyperscalers and OEM supply chains are tight. The next bellwether read on whether that path is getting shorter is Salesforce’s upcoming FY27 report (Agentforce/Data 360).

Salesforce’s most investor-relevant move here isn’t the Agentforce headline—it’s the way the company is presenting recurring revenue. In its FY27 materials, Salesforce updated disaggregated revenue reporting into two categories aligned to agentic enterprise: Agentforce Apps and Data 360, Headless Platform, & Other.

For analysts trying to separate “AI as a feature” from “AI as an added monetization layer,” this reclassification is the key because it puts AI-adjacent products into the subscription and support line where SaaS investors usually underwrite growth.

Because Salesforce is telling investors which portion of recurring revenue sits in the agentic apps bucket versus the data/headless/platform bucket, it creates a clean test for whether AI monetization is flowing through the software P&L—or whether it’s mostly financing infrastructure and services that don’t expand recurring billings fast enough.

Here’s the causal lens investors should use once results print: if customers are paying for agentic workflows and data/headless platform use (i.e., recurring subscription and support), then Salesforce’s recurring growth should hold up even if AI infrastructure remains a near-term procurement bottleneck.

Conversely, if the IBM-style pattern dominates, Salesforce could still report revenue growth—but the mix could shift toward non-seat monetization (e.g., services, term-like components, or delayed deals) and margin could wobble as infrastructure-related costs rise.

IBM’s description of customers redirecting spending to “servers, storage, and memory” is a capacity-layer story. In that world, enterprise buyers may want AI functionality, but they can’t deploy it fast without compute.

So the key investment question is not “is software demand present?” but whether the software attachment to that compute deployment happens fast enough to show up in subscription and support growth, rather than being delayed.

Salesforce’s two-bucket disaggregation reduces ambiguity: Agentforce Apps represent AI-native CRM motion, while Data 360/Headless represents the data and platform layer agents rely on. If AI infrastructure procurement delays deployment, Salesforce’s FY27 recurring growth and margin target become the easiest places to see whether monetization is getting pulled forward or stretched out.