Please refer to our disclaimers, which can be found in the footnote of this page and here.
17 October 2025
We wrote about our Salesforce investment in our 4Q 2023 Investor Letter and have owned it since 2020 (in various sizes). Despite the stock being down -30% YTD through September, it has contributed +30bps to performance as we added to the position during periods of weakness. It is now our largest position.
The stock has declined due to several factors: it started the year with a stretched valuation, negative sentiment around SaaS, a slowdown from low double-digit to high single-digit growth as the business matures (prompting shareholder turnover), and a notable drop in Marketing & Commerce revenue. Each is discussed below.
The “SaaS is dead” narrative has emerged in the AI hype cycle. The view is that traditional software is now much quicker to create with the assistance of AI leading to faster disruption times, that AI agents will do the heavy lifting creating little need for the application layer, that SaaS businesses will struggle versus AI-native competitors, and that AI will cause a decline in seat-based subscriptions as knowledge workers are made redundant with AI. This view is too simplistic.
We are AI-believers and are already implementing many AI use cases for our own business. However, we believe its impact on SaaS is far more nuanced. As was the case for the transition from on-premise computing to the public cloud, the AI transition will threaten weaker business models and enable many new business models, but the strongest and best-placed incumbent businesses also have an opportunity. Deeply entrenched, mission-critical, enterprise-focused SaaS businesses have an advantaged position in the agentic AI era. They own or control the most important functions and datasets for an enterprise and are a natural integration point for AI agents. They also have deep enterprise sales relationships. AI is both a revenue and a margin opportunity for these businesses as they will also use AI to become more productive and reduce headcount themselves.
The concern regarding reduced knowledge workers and seat-based subscriptions is misguided and short-term orientated. While subscriptions might largely be based on seats today, the reality is that customers look at their overall spend versus their overall value received. If Salesforce’s AI innovations can help reduce their customers’ headcounts and costs, then it is also the case that the value being delivered is significantly greater. In our experience, more value to the customer usually means higher revenue capture not lower. That is, the number of seats might decline, but the price per seat (or per consumption) would go up. This might not be immediate, or linear, but over the long-term we have high conviction that higher value delivered to the customer means higher revenue and profits, not less.
The better question is whether Salesforce can deliver higher value to its customers in the AI era. Salesforce has a critical advantage here given the depth and breadth of the customer data it controls (and sometimes owns) for a wide range of enterprises in mission critical, system-of-record applications. These touch sales, service, marketing, commerce, employee messaging, and analytics. Salesforce has long had a vision to provide a single 360-degree view of an enterprise’s customers. As outlined in our 4Q 2023 letter, its early efforts were failing due to its various acquired systems not being well integrated with its core products. Recognising this, Salesforce began a multi-year rebuild of its technology platform well before the generative-AI wave, a decision that proved fortuitous once ChatGPT launched. Progress here has been an ongoing diligence item for us, and we have spoken with several former software engineers at the company for an update. The initiative is designed to unify Salesforce’s sprawling mix of legacy and acquired products onto a common architecture with shared data, security, and workflow services. It has become central to Salesforce’s AI roadmap. The transition, however, has been bumpy. For example, the re-write of Marketing Cloud requires customers to undergo a significant migration from the old to the new. Customers are frustrated with this process and that the new Marketing Cloud does not yet have feature parity.
On the plus side, Salesforce’s Data Cloud and Agentforce initiatives are scaling rapidly. Data Cloud and AI ARR reached $1.2b last quarter (120% year-over-year growth); within that, generative AI generated about $440 million of ARR (400% growth). Since launch, the company has closed more than 12,500 Agentforce deals (6,000 paid) and half of the Fortune 500 are Data Cloud customers. This has the potential to add 1%-3% to overall annualised growth over the next 5 years. Salesforce has all but promised a reacceleration of growth above 10% at its Dreamforce analyst session in mid-October. They reported a reacceleration of net new average order value (“NNAOV”) in excess of average order value, which implies a future revenue acceleration if this is sustained (NNAOV went negative for two years prior). Importantly, adoption of Data Cloud and Agentforce also drives higher spending on the core platforms. Large language models, and therefore AI agents, are probabilistic in nature. They require traditional tools like process automation and mining among others to help set the rules and boundaries. On balance, AI represents more of an opportunity for Salesforce than a risk.
Concerns about slowing Marketing & Commerce revenue are partly due to Salesforce moving Data Cloud revenue out of Marketing Cloud without restating previous figures, resulting in unclear communication and confusion among investors. Additionally, Commerce Cloud revenue is normalising as outsized post-pandemic contracts with upfront GMV commitments wind down. Commerce Cloud also faces competitive pressure from Shopify and still requires a significant platform rebuild. A major recovery in Commerce is not in our base case.
At Salesforce’s current price of $243, it trades on a 5.6% FCF yield (4.2% net of SBC). We think this is very reasonable for a business that can compound FCF in the mid-teens. Our base case IRR is 20%+[1].
Footnotes
[1] Refer to our disclaimers, we make no guarantees of future performance, and our projections should not be relied upon