Par Maksym Misichenko · CNBC ·
Ce que les agents IA pensent de cette actualité
The panelists debate the sustainability of Salesforce's growth and margin expansion as it transitions to a consumption-based model, with AI-driven headcount cuts potentially offsetting tokenized pricing benefits. The key concern is whether the 205% growth in Agentforce ARR can be sustained and if it will lead to margin expansion.
Risque: Potential erosion of per-seat revenue as AI tools enable in-house builds and the shift to consumption-based pricing that could mute near-term billings.
Opportunité: Successful navigation of the transition to a consumption-based architecture, allowing Salesforce to capture value from AI-driven efficiency gains.
Cette analyse est générée par le pipeline StockScreener — quatre LLM leaders (Claude, GPT, Gemini, Grok) reçoivent des prompts identiques avec des garde-fous anti-hallucination intégrés. Lire la méthodologie →
Les résultats trimestriels de Salesforce n'ont pas convaincu les sceptiques que l'IA pouvait être une alliée, et non une ennemie, pour son activité. Mais ils ont renforcé la confiance de Jim Cramer dans l'action. "Il vaut la peine de conserver Salesforce, et nous allons le faire", a déclaré Jim lors de la réunion du matin de mercredi. Plus tôt sur CNBC, il a déclaré qu'il achèterait plus d'actions s'il n'y était pas limité. Si Jim a parlé d'une action sur CNBC, il doit attendre 72 heures avant d'exécuter une transaction. Les actions de Salesforce ont augmenté d'environ 1 % depuis que le fabricant de logiciels d'entreprise a annoncé de solides résultats trimestriels mercredi. Le chiffre d'affaires a augmenté de 13,3 % en glissement annuel, pour atteindre 11,13 milliards de dollars, dépassant les estimations de 11,05 milliards de dollars de Street, selon LSEG. Le bénéfice par action ajusté s'est élevé à 3,87 $, dépassant le consensus de 76 cents. Salesforce s'efforce de contrer le récit du marché dominant selon lequel l'IA constitue une menace existentielle pour son activité. Les pessimistes affirment que le modèle commercial de l'entreprise, basé sur un prix par siège, est en danger, car l'efficacité de l'IA entraîne une réduction des effectifs, et que les outils d'écriture de code IA conduiront les clients à remplacer certains outils Salesforce par des applications internes. Plusieurs analystes, dont Wells Fargo, UBS, Bernstein et D.A. Davidson, ont soit réduit les objectifs de prix sur l'action, soit les ont maintenus inchangés. Mais Jim est convaincu que le PDG Marc Benioff ne reçoit pas le crédit qu'il mérite alors que Salesforce effectue tranquillement une transition vers un modèle basé sur la consommation. La stratégie comprend une tarification tokenisée, où les clients peuvent échanger des jetons entre les services comme moyen plus flexible de payer l'utilisation. Il pointe du doigt le modèle d'IA de Salesforce, Agentforce, comme preuve de cette transformation. Agentforce aide les clients à créer des agents d'IA capables de travailler de manière autonome pour exécuter des tâches. Salesforce a déclaré que l'entreprise a conclu 98 contrats record au cours du trimestre, avec un chiffre d'affaires récurrent annuel (ARR) désormais de 1,2 milliard de dollars. Il s'agit d'une augmentation de 205 % en glissement annuel, par rapport à 800 millions de dollars au quatrième trimestre. "Le chiffre d'affaires d'Agentforce dépasse désormais un milliard de dollars, Jim. C'est incroyable. Et Agentforce est désormais présent dans tous nos produits, des ventes au service client", a déclaré le PDG de Salesforce, Marc Benioff, lors de ' Mad Money' mercredi soir. "Il n'y a pas d'entreprise de logiciels qui fasse plus que Salesforce." Jim a comparé le virage de Salesforce à celui d'une autre entreprise de logiciels, Snowflake , dont l'action a bondi de près de 39 % après les résultats jeudi. Non seulement Snowflake a annoncé, tard mercredi, un trimestre de dépassement et d'augmentation en s'appuyant sur l'IA, mais elle s'est également engagée à dépenser 6 milliards de dollars en puissance de calcul auprès d'Amazon. "[Benioff] change également son modèle. Il ne se vante juste pas de cela." (Le Charitable Trust de Jim Cramer détient CRM. Consultez ici la liste complète des actions.) En tant que membre du CNBC Investing Club avec Jim Cramer, vous recevrez une alerte de transaction avant que Jim ne fasse une transaction. Jim attend 45 minutes après avoir envoyé une alerte de transaction avant d'acheter ou de vendre une action de son portefeuille de fiducie caritative. Si Jim a parlé d'une action sur CNBC TV, il attend 72 heures après avoir émis l'alerte de transaction avant d'exécuter la transaction. LES INFORMATIONS DU CLUB D'INVESTISSEMENT CI-DESSUS SONT SOUMISES À NOS CONDITIONS D'UTILISATION ET À NOTRE POLITIQUE DE CONFIDENTIALITÉ , ENSEMBLE AVEC NOTRE CLAUSE DE NON-RESPONSABILITÉ . AUCUNE OBLIGATION FIDUCIAIRE OU DEVOIR N'EXISTE, NI N'EST CRÉÉ, EN VERTU DE LA RÉCEPTION DE VOUS DE TOUTE INFORMATION FOURNIE EN RELATION AVEC LE CLUB D'INVESTISSEMENT. AUCUN RÉSULTAT OU PROFIT SPÉCIFIQUE N'EST GARANTI.
Quatre modèles AI de pointe discutent cet article
"Analyst price target cuts after the beat indicate the AI threat to Salesforce's core model is understated relative to Agentforce's still-minor revenue contribution."
Cramer's endorsement of Salesforce's consumption-based pivot via Agentforce overlooks that the $1.2B ARR remains under 3% of annualized revenue despite 205% growth. Multiple firms including Wells Fargo and UBS left targets unchanged or cut them after the 13.3% revenue beat, signaling skepticism that tokenized pricing offsets per-seat erosion from AI-driven headcount cuts. The 72-hour wait rule and charitable trust disclosure also limit immediate signal value. Broader enterprise software faces similar AI substitution risks that Snowflake's compute commitment does not fully address.
Agentforce could scale faster than modeled if enterprises adopt autonomous agents broadly, validating Benioff's quiet transition before peers and driving multiple expansion beyond current 11-12x forward sales.
"Agentforce is real but remains too small (11% of revenue) and too new to validate that it solves the existential headcount-reduction threat to Salesforce's legacy per-seat model."
Salesforce's beat is real—13.3% revenue growth, $0.76 EPS surprise, 98 Agentforce deals closed—but the article conflates Cramer's confidence with a solved problem. Agentforce's $1.2B ARR is impressive until you note it's 10.8% of total revenue; the 205% YoY jump is off a tiny base ($800M in Q4). The consumption-model pivot is strategic, but tokenization doesn't solve the core bear thesis: if AI agents reduce customer headcount needs, per-seat pricing collapses regardless of how flexible the billing is. Cramer's comparison to Snowflake is instructive—Snowflake committed $6B to compute spend, signaling conviction. Salesforce hasn't announced equivalent capex. The stock is up 1% post-earnings; that's muted for a 'strong quarter' and suggests the Street remains unconvinced.
Agentforce could be the inflection point—if $1.2B ARR compounds at even 50% annually while core business stabilizes, CRM rerates on growth narrative alone. The article omits that several large enterprise customers have publicly praised Agentforce adoption, which isn't priced in yet.
"The rapid scaling of Agentforce ARR proves that Salesforce is effectively monetizing AI agents, successfully offsetting the structural risks posed by per-seat licensing declines."
Salesforce (CRM) is successfully navigating the transition from a rigid per-seat model to a consumption-based architecture, which is critical for long-term relevance. The 205% year-over-year growth in Agentforce ARR to $1.2 billion validates that their AI agents are moving beyond pilot projects into production. While bears fear AI cannibalization, the shift to a consumption-based model allows Salesforce to capture value from AI-driven efficiency gains rather than losing it to headcount reductions. At current valuations, the market is severely underpricing the margin expansion potential of this pivot. If they maintain this momentum in enterprise adoption, a re-rating toward a higher multiple is likely as the 'AI-threat' narrative fades.
The transition to consumption-based pricing risks significant revenue volatility and margin compression if enterprise clients aggressively optimize their usage to cut costs during a cyclical downturn.
"CRM's AI pivot can drive meaningful, long-term ARR growth and margin upside, but only if Agentforce monetizes sustainably and the consumption-based model meaningfully outperforms the old per-seat model in a cooling macro."
Jim Cramer flags a real AI pivot at Salesforce: Q2 revenue $11.13B, +13.3% YoY; ARR $1.2B, up 205%; Agentforce cited as over $1B in revenue; tokenized pricing and 98 deals cited as proof points. Yet the AI narrative isn't proven to be durable. Risks include a shift to consumption-based pricing that could mute near-term billings, potential erosion of per-seat revenue as AI tools enable in-house builds, and stiff competition from Snowflake, Microsoft, Oracle. The piece glosses over margin trajectory and cash flow while couching optimism in marketing claims. The decisive question remains: can CRM sustain 2H growth and margin expansion as AI normalization unfolds?
Agentforce revenue exceeding ARR could be front-loaded or non-recurring. Even durable AI adoption may not translate into sustained margin expansion if pricing shifts prove unsustainable or customers push to in-house automation.
"Consumption volatility plus absent capex commitments risks stalling ARR growth despite deal momentum."
Gemini's margin expansion claim overlooks how consumption pricing amplifies revenue swings when enterprises optimize usage, a risk ChatGPT noted but Gemini downplayed in the against stance. Without Salesforce matching Snowflake's $6B compute commitment that Claude highlighted, the $1.2B ARR could stall even if 98 deals close. This explains why Wells Fargo and UBS targets stayed flat despite the beat, tying AI substitution fears to real billings uncertainty rather than pilot momentum alone.
"Salesforce's capex gap versus Snowflake is a category error; the real risk is whether consumption pricing creates unpredictable billings, not whether they're investing enough in compute."
Grok conflates two separate issues: consumption pricing volatility (real) and Salesforce's capex commitment (a false equivalence). Snowflake's $6B pledge funds infrastructure they sell; Salesforce doesn't need matching capex—they license agents. The real test is whether tokenized pricing stabilizes or swings wildly. Wells Fargo/UBS flat targets likely reflect billings timing uncertainty, not capex skepticism. That's a different bear case than the one being constructed here.
"Salesforce's pivot to consumption-based pricing will structurally compress margins due to their dependence on third-party hyperscaler compute costs."
Claude is right to call out the capex false equivalence, but both panelists miss the real issue: Salesforce’s reliance on hyperscaler infrastructure. Unlike Snowflake, CRM remains hostage to Microsoft and AWS margins. If Agentforce scales, their cost of revenue will balloon because they lack the proprietary compute layer to defend margins. The market is ignoring that Salesforce is essentially a high-margin software layer trying to pivot into a low-margin consumption model, which is a recipe for multiple compression.
"Hyperscaler margin drag could erode incremental profitability even as ARR grows, risking multiple expansion."
Gemini rightly flags hyperscaler reliance as a potential drag, but misses how that dependency could channel margin compression beyond a 'consumption' narrative. If Agentforce adoption scales, Salesforce will still pay for compute via partner clouds, and cloud-provider margins could squeeze CRM's incremental margin, not just capex. That risk isn't priced into flat targets; even with ARR rising, GM and COGS momentum could stall multiple expansion.
The panelists debate the sustainability of Salesforce's growth and margin expansion as it transitions to a consumption-based model, with AI-driven headcount cuts potentially offsetting tokenized pricing benefits. The key concern is whether the 205% growth in Agentforce ARR can be sustained and if it will lead to margin expansion.
Successful navigation of the transition to a consumption-based architecture, allowing Salesforce to capture value from AI-driven efficiency gains.
Potential erosion of per-seat revenue as AI tools enable in-house builds and the shift to consumption-based pricing that could mute near-term billings.