29,000 Agentforce deals and a 30 per cent stock decline: Salesforce’s AI paradox

WorkAI.TV Editorial Desk
4 Min Read

Share with your CRO

Salesforce is betting its entire growth story on Agentforce, its AI agent platform, and the early commercial numbers are hard to dismiss: 29,000 deals closed, $800 million in ARR, and live production deployments at Williams-Sonoma, the city of Kyle, Texas, and the University of Chicago Medicine. Yet the stock is down 30 percent in 2026 as investors price in the SaaSpocalypse, a market repricing built on one brutal question: if AI agents replace human users, why does anyone pay for per-seat software licenses at scale?

What this means for your business

Revenue growth decelerating from 24 percent to 10 percent is a signal, not a data point. If you’re a CRO whose commercial engine runs on Salesforce, you’re sitting inside the specific customer segment that Salesforce needs to retain while simultaneously asking you to adopt a consumption-based pricing model, where you pay roughly $0.10 per agent action instead of a fixed seat fee. Whether that shift lowers your total cost depends entirely on how much your teams actually use the platform, which means your renewal conversation is about to get a lot more complicated.

The Williams-Sonoma deployment resolving 60 percent of chat inquiries autonomously is the number Salesforce wants you to anchor on. But SharkNinja’s headline, a 20 percent reduction in support calls, was described as a forward-looking expectation, not a reported outcome. That distinction matters when you’re evaluating whether to consolidate more of your customer engagement stack onto Agentforce or hedge with a competing platform from ServiceNow or Microsoft. The deployments that are genuinely in production are largely in customer service and 311-style triage, not complex sales workflows, which is the territory most relevant to a CRO.

The consumption model Salesforce introduced is the clearest sign that the company knows seat-based pricing is structurally threatened, and pricing architecture changes at a vendor this size tend to outlast any single contract cycle. If Agentforce delivers on autonomous resolution rates, your per-seat headcount in service and inside sales shrinks, which cuts Salesforce’s own license revenue unless consumption volume fills the gap. That’s a bet you’re being asked to share the risk on. The falsification condition is straightforward: if Agentforce ARR crosses $2 billion by fiscal 2027 with documented enterprise outcomes, the model works and the pricing shift is permanent. If it stalls, expect Salesforce to quietly reintroduce hybrid seat floors.

Concept deep-dive: Consumption-based pricing

Consumption-based pricing charges customers for what they actually use, measured in actions, API calls, or tokens, rather than a flat fee per user per month. Think of it like a utility bill versus a gym membership. For enterprise software, it removes the floor on revenue predictability for vendors while theoretically aligning cost to value for buyers. The business risk is that low-usage periods generate almost no revenue, making it hard for either side to forecast reliably without solid deployment data.

Based on reporting from 29,000 Agentforce deals and a 30 per cent stock decline: Salesforce’s AI paradox, originally published 2026-05-22 03:00:00.

TAGGED:
Share This Article