TNW’s investigative piece exposes the gap between Salesforce’s Agentforce narrative and reality. The numbers (29,000 deals, $800M ARR, 2.4B agentic work units) are real — but flagship deployments at Williams-Sonoma, UChicago Medicine, and SharkNinja were presented using future tense at Dreamforce rather than as live systems. Revenue growth has slowed from ~25% to ~10%, and the stock is down 30% in 2026. The consumption model (“agentic work units”) must scale fast enough to offset structural seat-compression. The May 28 earnings will be the first concrete test.

Salesforce confirmed the new AI revenue disclosure structure debuts on May 28: all AI-related revenue consolidates into two reporting buckets — Agentforce (agent platform) and Data 360 (data cloud/analytics). This is a high-conviction structural move — you don’t create disclosure lines for underperforming businesses. The simplified structure makes it easier for investors to track AI contribution quarter-over-quarter. With BofA arguing Agentforce is <2% of revenue, this first print will either validate the bears or force a narrative reset.

UBS analyst Karl Keirstead trimmed his Salesforce price target from $200 to $185 (effectively at current price), maintaining a neutral stance. The cut came on May 21 amid continued choppiness in legacy software stocks. CRM fell 2%+ on the day despite broader market gains. The timing — one week before earnings — signals that even neutral-rated analysts are unwilling to put their necks out ahead of the print. The pattern: every analyst who updates pre-earnings is cutting, not raising.

Zacks expects Salesforce to beat Q1 estimates based on its consistent track record (beat 4 of last 4 quarters). Company guidance: $3.11–$3.13 EPS, $11.03–$11.08B revenue. Street consensus is squarely in the guided range at $3.12/$11.06B. FY2027 full-year guidance: $13.11–$13.19 EPS, $45.8–$46.2B revenue. But as HubSpot and Cloudflare proved earlier this month, a beat means nothing if the AI narrative doesn’t land. The stock trades at 22x trailing earnings — cheap for software, expensive if growth decelerates further.

Salesforce’s largest product release of the year (GA June 15) delivers three headline capabilities: (1) Multi-Agent Orchestration — agents collaborate as unified teams with shared context across channels; (2) 50+ specialized IT agents deployed in Slack, Teams, and service desks out of the box; (3) Tableau MCP — AI agents query Tableau analytics directly via secure open integration. Also shipping: Customer Engagement Agent (24/7 lead qualification), Agentforce Self-Service (setup in 10 clicks), and real-time offer management. The volume of shipping product is the counter-argument to the “hype vs. delivery” critique.

Salesforce published internal results of dogfooding its own AI stack: standardizing on Claude Code and removing token limits allowed the Agentforce Commerce team to complete a migration estimated at 231 person-days in just 13 days — 18x acceleration. Fewer incidents, fewer bugs, more shipped. This is the strongest internal proof point yet: if Salesforce’s own engineering teams achieve these results, it validates the thesis that agentic workflows compress development timelines dramatically. It also explains why Bhusri at Workday said he wants headcount “as flat as possible” — the productivity gains are real.

New integrations allow customers to deploy AI agents across Slack and Google Workspace with Agentforce and Gemini Enterprise providing intelligence behind the scenes. The “deep context” architecture means agents access both Salesforce data and Google Workspace signals simultaneously. This matters strategically: rather than competing head-to-head with Google/Microsoft on AI, Salesforce is positioning as the orchestration layer that works WITH hyperscaler AI. It’s the “Switzerland of agents” play — and it hedges against customer lock-in concerns.

Workday reported Q1 FY2027: EPS $2.66 (beat by $0.15), revenue $2.54B (+13%), subscription revenue +14%. Best first quarter of new ACV growth in five years. Agentic AI ARR approaching $500M with agent adoption doubling QoQ (4,000+ clients using at least one agent). Raised full-year margin guidance to 30.5%. Co-founder Aneel Bhusri returned as CEO and declared: “The next agentic application will move the needle, not the 150th HR feature.” He wants headcount flat — betting Workday’s own AI tools replace incremental hires. Stock jumped 5% Friday.

Bank of America analyst Tal Liani reinstated Salesforce coverage with an Underperform rating and $160 price target (implying ~11% downside) — the first outright bearish call from a major bank in years. Previously rated Buy. Three concerns: (1) muted net new customer additions, (2) limited upsell potential as AI compresses seat counts, (3) “underwhelming AI monetization pathway.” BofA forecasts revenue growth slowing to 9.6% in FY2028 and warns that AI-driven workflow automation increases competition. Notably, CRM rose 3% on the day — contrarian investors appear to disagree.

ServiceNow has recovered 17.6% since its controversial Q1 report that initially triggered the “SaaSpocalypse” selloff in April. The stock’s recovery despite posting the worst single-day drop in company history (-18%) tells an important story: the market eventually rewarded the strong underlying metrics (19% subscription growth, raised guidance). For Salesforce, this is both encouraging and cautionary — a beat eventually gets recognized, but the initial reaction can be violent. Knowledge 2026’s “autonomous workforce” thesis is gaining traction.

CIO.com’s landmark essay argues the central premise of enterprise SaaS — that building was too expensive — is breaking down as AI compresses development costs. Key data: Retool’s 2026 survey found 35% of teams have already replaced at least one SaaS tool with a custom internal build; 78% plan more. The moat “was never the code — it was the cost of writing the code.” The counter-argument: switching costs are measured in quarters of lost productivity, not license fees. Vendors who defend legacy margins will “find themselves defending empty castles.” This is the structural thesis behind BofA’s bearish Salesforce call.

Zacks highlights the extreme dispersion in CRM analyst targets: 49 estimates with a mean of $270, but the range spans from $160 (BofA’s new bear case) to $475 (most optimistic). Standard deviation of $60.63 — unusually wide for a mega-cap. This dispersion reflects genuine uncertainty: if Agentforce is real, CRM is deeply undervalued. If AI compresses the SaaS model, it’s fairly valued or overvalued. The May 28 earnings will narrow this range dramatically — expect a wave of analyst revisions within 48 hours of the print.