Competition

Competition

In the market as it has been defined for a decade, Salesforce is not just the leader — its lead is widening. It holds roughly a fifth of worldwide CRM-application revenue, close to five times its nearest rival. Whether that market is still where the contest will be decided is the less settled question. The contest is moving toward autonomous agents, where the distribution advantage tilts toward Microsoft's installed base, and where Salesforce's own 10-K now names, for the first time, rivals built to "displace established user interfaces." The moat holds the seat-based base; the shift to agentic software is real and unsettled.

The incumbent market: a widening lead

Salesforce competes in a market its own 10-K calls "highly competitive, rapidly evolving and fragmented, and subject to changing technology with low barriers to entry" [1]. That framing is honest about the threat surface, but it understates where Salesforce actually sits today. On IDC's measure of the CRM-applications market — the category the company was built to own — Salesforce held about 20.0% of worldwide revenue in 2025, its thirteenth consecutive year at the top and more than four times the share of its closest competitor.

Source: IDC Semiannual Core AI Software Tracker, H2 2025, worldwide CRM-applications revenue share — as reported by IDC; not in the filing corpus.

Two things follow from this shape. The lead is not a rounding advantage: at 20.0% against Oracle's 4.1% and Microsoft's 4.0%, Salesforce earns more CRM revenue than its next several rivals combined, and the gap has been widening, not closing, for over a decade. And the incumbents most often named as threats — Microsoft, Oracle, SAP — are, in the CRM category proper, minor. None appear in Salesforce's indexed peer set at all; the auto-selected comparables (ServiceNow, Adobe, Snowflake, plus Shopify, Uber and ADP) are enterprise-software adjacencies, not CRM rivals, so the competitive picture has to be assembled from Salesforce's own filings and outside data rather than the peer folder.

The point for a durability investor is narrow but real: within the game as scored today, no rival is close enough to take the installed base by direct competition. That is the evidence behind the switching-cost floor established earlier (Switching Costs). It is also why the more consequential threat is not a share war in CRM but a change in what customers are buying.

The new arena: where agentic budgets are landing

The reset is a shift from software priced by the seat to software priced by the task an agent performs. In that arena the scoreboard looks different, because the entry point is distribution, not CRM share. The clearest contrast is Microsoft. Its overall AI business was running at roughly a $37 billion annualized rate by the quarter ended March 2026, up about 123% year over year, and Microsoft 365 Copilot alone reached about 15 million paid seats — on the order of $5.4 billion of annual revenue at list — sold into an installed base of some 450 million Microsoft 365 seats. Salesforce's Agentforce, by contrast, crossed roughly $1 billion of annual recurring revenue in early fiscal 2027, growing faster in percentage terms but from a base several times smaller, and the broader Agentforce and Data 360 franchise stood near $2.9 billion (Growth and Agentforce).

Sources: company disclosures as reported (Microsoft fiscal Q2 2026; Salesforce Q4 FY2026 / Q1 FY2027 results; ServiceNow reporting); Agentforce and Data 360 ARR per Growth and Agentforce. Traction metrics are not comparably defined across vendors and are shown as reported.

The table mixes metrics that are not strictly comparable, and that is the finding: there is no agreed unit for "who is winning" in agents yet. But the direction is legible. Microsoft's advantage is that an agent is a feature it can switch on inside seats a customer already pays for; Salesforce's advantage is that the agent runs on the customer's own system of record and 92%-retained data, which is where an autonomous workflow has to reach to be useful. ServiceNow enters from workflow incumbency; the AI-native entrants enter with nothing to protect and therefore nothing to slow them down. Salesforce is a strong participant, not the runaway leader it is in CRM — and Wall Street has noticed the gap between the ARR growth rate and the depth of actual deployment, with more than one broker downgrading the stock in July 2026 on soft Agentforce adoption feedback.

The seat-based model under agentic AI

The bear case is not that a rival takes Salesforce's customers. It is that agents reduce how many seats those customers need. Salesforce's revenue is overwhelmingly per-user subscription; if a company can run its service desk or its sales operations with a handful of agents instead of scores of licensed employees, the seat count — and the bill — falls even with perfect retention.

For the first time, Salesforce's FY2026 10-K names this directly. Its list of competitors now includes "AI-native companies and emerging startups that leverage generative AI and large language models as the core foundation of their architecture, offering highly specialized, autonomous, or automated solutions that may bypass traditional business process workflows or displace established user interfaces" [2]. That clause did not appear in any of the four prior annual reports (FY2022–FY2025); it is new this year. The matching risk factor is equally plain: "New AI offerings may disrupt our service offerings or transform workforce needs and negatively impact demand for our offerings" [3]. "Transform workforce needs" is the seat-displacement mechanism, stated by the company whose seats are at issue.

The most direct evidence that the mechanism is real comes from Salesforce running the experiment on itself. Its own support portal, handling "over 60 million sessions and more than 2 million support cases yearly, all powered by Agentforce," has let the company "shift our workforce toward growth areas like distribution, where we strive to hire thousands of additional salespeople" [4]. Management frames this as the proof point, not the warning. Confronting "this strange narrative that's out there that somehow enterprise SaaS or apps or something are going away," Marc Benioff argued that a company can "do what we did, which is reduce your support heads and have an agentic layer and have a more efficient company and make more money and do better for your shareholders" [5].

Both readings sit inside that same fact. Salesforce cut its own support headcount because Agentforce works — and if it works for Salesforce, the same arithmetic lets Salesforce's customers cut the seats they buy from Salesforce. The company's answer is consumption pricing, which bills the agent's activity instead of the vanished seat (Growth and Agentforce); whether that recaptures the lost seat revenue dollar-for-dollar is the open variable, and it is the same variable the valuation chapter isolated as the line between a cheap price and a fair one (What the Price Implies). Isolated public examples cut against the company — Klarna's move to strip back packaged SaaS in favor of in-house agents drew Benioff's direct pushback — but they remain anecdotes, not a trend visible in the retention data.

The engine meant to replace displaced seats — pure Agentforce ARR of about $800M, roughly 2% of revenue once $1.1B of acquired Informatica is stripped out — is still a rounding error against the ~95% of $41.5B revenue that is per-seat, even as Salesforce's own 10-K newly names AI-native rivals that can 'displace established user interfaces' and Benioff proves the mechanism by cutting his own support heads. Those figures, and the ten-year scenario that turns on them, are developed in the growth and outlook chapters (Growth and Agentforce; Ten-Year Outlook).

Bull and bear on the same facts

Each row below is a fact both sides accept; the disagreement is what it implies and what would settle it.

Sources: FY2026 10-K Competition and Risk Factors [6][7]; Q2 FY2026 call [8]; IDC H2 2025 and company AI disclosures as reported.

The evidence does not settle this today, and pretending otherwise would be the error. What it does establish is a clean separation: in the incumbent CRM market the moat is not in question, and in the agentic market it is genuinely contested, with the structural distribution edge sitting with Microsoft and the data-and-workflow edge sitting with Salesforce. The reader's durability test — is year-ten free cash flow larger than today's — turns less on CRM share, which looks secure, than on whether the seat model survives its own automation.

What would change the read

Four checkable items, each with the line to watch and the threshold that flips it:

Net revenue retention on the AI line. Salesforce discloses Agentforce and Data 360 ARR but not a net revenue retention rate for it. A disclosed AI-line NRR holding above roughly 100% would show consumption revenue is replacing, not merely accompanying, seat revenue; its absence is why the seat question stays open.

Organic cRPO. Sustained organic current-RPO growth above the ~11–12% constant-currency mark named earlier (Growth and Agentforce) would signal the AI franchise is adding to the book faster than automation subtracts seats. Stuck near 9% leaves the seat question unresolved.

IDC CRM share through FY2027. Holding near 20% while agent-native entrants scale would confirm the incumbent market is not being routed around. A visible erosion — not seen yet — would be the first hard sign the category itself is shifting.

Reference-customer displacement. A large, named customer publicly replacing Salesforce seats with a rival or in-house agent stack at scale, in the manner Klarna has floated, would move the debate from anecdote to trend. The retention data shows no such trend to date [9].