KeyBanc’s Jackson Ader downgraded Salesforce to Hold from Buy after a string of partner and customer events surfaced the same complaint repeatedly: customer data isn’t organized enough to do meaningful AI work, and Agentforce partners are only now converting proof-of-concept pilots into real pipeline. Ader also said he cannot find evidence in Salesforce’s own filings for the “net-new AOV outgrowing overall AOV” claim management has repeated on recent calls. Bernstein followed with its own downgrade, noting customer checks “have not been strong” and that Salesforce’s valuation, while cheap versus its own history, now sits mid-pack versus peers once growth is factored in. The stock fell roughly 3–4% intraday to the ~$160 area. This is the most direct sell-side challenge yet to the thesis that AI monetization can offset seat-based revenue pressure.

A week before the KeyBanc/Bernstein downgrades, Guggenheim moved the opposite direction, upgrading Salesforce to Buy from Neutral with a $228 target, implying roughly 46% upside from the prior close. The firm’s case: at 3.7x recurring revenue and 11x forward enterprise-value-to-free-cash-flow, the stock already prices in a worst-case AI disruption scenario that Guggenheim views as overstated. The two calls, eight days apart, are effectively arguing about the same underlying question — whether Agentforce is real revenue or a narrative — and reaching opposite conclusions from similar starting valuations. That gap is the story heading into Q2 FY27 earnings in August.

Salesforce shipped what it calls its biggest Commerce release in five years: Shopper Agent, Buyer Agent, and Merchant Agent are generally available now across B2C, B2B, order management, and point of sale, with product catalogs syncing directly into ChatGPT (GA now) and Google Search/Gemini integration following this summer. Early customers cite double-digit conversion lifts from the new Agentic Commerce Search (built on the acquired Cimulate technology) and an 88% reduction in merchandising task time. The strategic logic: as product discovery shifts to AI assistants, Salesforce wants to remain the system of record for inventory, pricing, and fulfillment regardless of where the shopper started the conversation. The timing cuts both ways — it’s tangible proof of AI shipping velocity, which is exactly what KeyBanc says isn’t yet showing up in the numbers.

Folded into the Commerce release is Agentic Commerce Search — an AI-native discovery engine from Salesforce’s Cimulate acquisition, now fully integrated into the core platform and deployable even on non-Salesforce storefronts. Early adopters report double-digit improvements in both conversion and add-to-cart rates. It’s a small feature with a large strategic purpose: as generic chatbots commoditize conversation, Salesforce is betting that proprietary, business-specific search and inventory grounding is the layer that’s actually defensible. Whether that defensibility shows up as durable ACV growth — the exact metric KeyBanc says it can’t verify — is the open question heading into next quarter.

Salesforce signed on as a founding participant in a new AI for Good Global Commission aimed at expanding access, strengthening trust, and accelerating responsible AI impact worldwide. The initiative is soft-power positioning more than a product event, but it matters for enterprise buyers in regulated or public-sector accounts, where AI governance credentials increasingly factor into procurement. It also reinforces Salesforce’s consistent 2026 playbook of pairing every AI product push with a parallel trust-and-governance narrative — useful cover when the product story itself is under analyst scrutiny.

New Forrester research finds enterprises deferring roughly a quarter of planned 2026 AI spending into 2027 as finance teams demand clearer proof of return, while a separate Gartner survey found fewer than one-third of corporate decision-makers can point to a specific financial outcome from their AI investments so far. This is the macro backdrop that makes KeyBanc’s Agentforce critique land: it isn’t a Salesforce-specific problem so much as an industry-wide gap between AI deployment announcements and AI-driven P&L impact. Vendors that can show hard ROI numbers — not just usage metrics like Agentic Work Units — are going to separate from the pack over the next two quarters.

HubSpot reaffirmed an 18% year-over-year growth trajectory for both Q2 and full-year 2026, a notably steadier print than the AI-disruption anxiety hitting Salesforce and ServiceNow this year. HubSpot has leaned into margin expansion from AI cost efficiency (EPS guidance revised up to $3.73 from $2.91 earlier this year) even without yet showing dramatic AI-driven top-line acceleration. For enterprise buyers watching the SaaS peer group, HubSpot is the closest thing to a control case this week: mid-market-weighted, less exposed to the “seat cannibalization” narrative dogging the largest platforms, and growing at a consistent clip through the same AI disruption cycle that’s driving 40%+ swings in CRM and NOW.

Buried in Bernstein’s downgrade note is a peer-group observation worth flagging on its own: Salesforce’s multiple, while depressed against its own trading history, has drifted to roughly the middle of the enterprise SaaS peer pack once you adjust for growth. That’s a meaningfully different framing than the “statistically cheap” case Guggenheim made a week earlier using the same underlying numbers. The divergence is a reminder that 2026’s SaaS multiples compression has been uneven — some platforms (HubSpot) are holding growth steady, others (Salesforce, ServiceNow) are leaning on AI narrative to justify re-rating, and sell-side firms are increasingly split on which framework to apply.

Bain’s 2026 CFO survey finds 83% of finance leaders plan to raise enterprise AI investment by more than 15% over the next two years, with 42% planning increases above 30%. But the same survey wave shows innovation-budget line items for AI shrinking from 25% to just 7% of LLM spend, as AI moves from experimental fund to core IT budget — meaning it now competes directly with every other line item on ROI terms, not innovation goodwill. That shift in scrutiny is precisely the environment KeyBanc describes when it says CIOs are more likely to cut Salesforce spend than raise it next year: budgets are growing, but vendors now have to prove it, not just ship it.

Industry estimates put AI-specific enterprise software spending on pace to cross $450 billion in 2026, growing roughly 60% year-over-year and materially outpacing overall enterprise software growth. The scale of that number is exactly why the KeyBanc/Guggenheim disagreement over Salesforce matters beyond one stock: if AI spend is genuinely growing at 60%, the platforms that can prove they’re capturing a fair share of it re-rate quickly, and the ones that can’t — even ones with strong historical platform lock-in — risk a multi-quarter overhang while the market waits for evidence.