Salesforce’s proprietary artificial intelligence tool Agentforce has been on the scene for exactly a year now, and there is no better time than the present to evaluate not only its performance, but what kind of earnings it’s bringing in for the ambitious cloud giant.
Assessing its existing achievements and capabilities is one thing, but for analysts, investors, and critics alike, the key information lies in the earnings of Salesforce’s AI suite – and how Salesforce’s numbers measure up against those of its competitors.
How Is Salesforce Doing?
In Salesforce’s latest earnings report, the SaaS leader watched its stock drop by nearly 7% due to weak forecasts, but the figures of the present spoke for themselves. Boasting a second-quarter revenue total of $10.2B, $1.2B of that was attributed to ‘Data Cloud and AI’, up 120% YoY.
Salesforce also reported that it had secured more than 12,500 Agentforce deals since its launch, with 6,000+ of them being paid. This is a strong jump from the 8,000 deals reported from the quarter before, and can arguably reflect the efforts that Salesforce has worked to address with their latest rendition of the tool, Agentforce 3.
Persistent Doubts
Although anyone can quite confidently argue that Agentforce has escaped its hype cycle, that does not necessarily mean it is out of the woods. In fact, we might well be in the most pivotal part of Agentforce’s journey – the part where Salesforce decides whether it continues to safely build upon the existing bones of Agentforce, or break out of the box with something big.
However, Agentforce is still surrounded by an air of hesitation, and this is being shown through the slower levels of adoption than Salesforce had possibly hoped for. It is evident that Salesforce still has considerable work to do to show that Agentforce is a fully fledged product – much like its competitors are actively trying to do.
The Competitor Stats
When looking at the numbers of the competitors, it is important to note that most vendors don’t break out a clean “AI revenue” or “AI earnings” line. Instead, they will often use proxies like ARR (annual recurring revenue) for AI/Data products, ACV (annual contract value) targets for AI add‑ons, or broader AI‑influenced ARR.
With this in mind, current competitor figures are as follows:
| Company | What They Disclose (Definition) | Latest Figure & Date |
|---|---|---|
| Salesforce | “Data Cloud & AI ARR” (Salesforce’s combined ARR for Data Cloud + AI/Agentforce SKUs) | $1.2B ARR, up 120% YoY (Q2 FY26, reported Sept 3, 2025). $100M Agentforce revenue (May 29, 2025) |
| ServiceNow | Now Assist AI ACV target (management goal for AI add‑ons, not revenue) | On track to hit $1B ACV by 2026”; “continued to surpass net new ACV expectations” (Q2 2025, Jul 23, 2025) |
| Anthropic | AI Annualized Revenue Run-Rate (ARR) | $5B (As of August 2025, having grown from about US$1 billion at the start of the year.) |
| OpenAI | AI Annualized Revenue Run-Rate (ARR) | $12B (~$12B ARR after doubling its income over the first seven months of 2025.) |
| Adobe | AI ARR (ARR where AI materially drives value) and AI‑first ARR (ARR from new, AI‑only products like ‘Firefly’ and ‘Acrobat AI Assistant’) | AI ARR: $5B; AI‑first ARR: $250M (Q3 FY25, Sept 2025) |
| Microsoft | No concrete AI revenue; management commentary on pricing/ARPU | “ARPU growth will again be driven by E5 and M365 Copilot.” (FY25 Q4 outlook, Jul 30, 2025) |
| Workday | No AI revenue figures; commentary on AI SKU demand | “Growing demand for our AI SKUs,” noted with results (Q4 FY25, Feb 25, 2025). |
| HubSpot | No AI revenue figures | Q2 2025 results show solid growth, no AI‑specific metric disclosed (Aug 6, 2025). |
| Freshworks | AI ARR for ‘Freddy Copilot/Agent’ products | $20M combined AI ARR (Q2 2025, Aug 2025) |
What Does This Mean?
Despite the criticism that Agentforce has garnered over the last year, Salesforce’s AI numbers are not feeble. In fact, by just numbers alone, Salesforce shows a growing market dominance in not only the CRM space but the fast-evolving AI space, too.
ServiceNow remains one to watch in this particular race, with a similarly scaled trajectory to Salesforce’s Data+AI run-rate, but the key difference here lies in the fact that ServiceNow’s figures are only forecasted, not current.
Adobe also shows promising numbers with the biggest ‘AI-adjacent’ figures, but the definition of its earnings is slightly different. However, analysts have already begun solidifying Adobe as an “AI single point of contact”, indicating a belief in the tools and their capabilities.
It becomes more difficult to compare the remaining software due to bundling – inadvertently making AI earnings much less measurable – or a lack of specific AI figures. Nonetheless, Salesforce is not alone in its AI efforts, and this much is clear from the direction that many of its competitors have chosen to go.
What About the Big AI Duo?
Although this particular comparison focuses heavily on Salesforce’s fellow SaaS competitors, it is impossible to ignore how the impacts of the big AI duo – OpenAI and Anthropic – come into play here.
In May, it was announced that Athropic hit $3B in annual revenue, indicating a rising demand for AI. The month after, OpenAI reported earnings of $10B – solidifying its place as the AI leader.
Although these two companies are in a league of their own, it does still pose questions for Salesforce.
These numbers show that businesses are already spending considerable amounts with OpenAI and Anthropic to build their AI tools and platforms, likely due to the range of vast abilities and continuous results. So, how is a company like Salesforce, with a packaged AI solution, going to compete with a completely different, customizable AI solution?
That is why it’s pertinent to not only consider the earnings of Agentforce going forward, but also how well it can snag the customers that want an AI solution that’s a bit more packaged and specific.
Summary
Looking at Salesforce’s AI earnings as a singularity – especially compared to the figures of its competitors – casts the CRM giant in a healthy light. Breaking through with numbers like this, despite the odds, might make anyone question why the resounding doubt still hasn’t been quelled.
That is where considering the tool as a whole is so vital, because although closed deals show promise, they do not show use. They do not show Agentforce’s capabilities in their full splendor, and they do not show how any stage of these capabilities performs against those of its competitors.
As we head into the next few months of Agentforce’s run, we will all be paying close attention to both sets of data: the earnings and the tangible results. Only this way will we understand what Salesforce has in store next for the tool, and how long it will take for investors and analysts to take on a more confident stance when it comes to the tool.
