Over on Semi Insider, and later in the video here https://youtu.be/aEcd5uVYkww, we discussed with our members some changes we made to our enterprise SaaS stock portfolio for 2025. This includes a small change with our long-time holding in Salesforce (CRM). Here are those research notes in blog form.
Let’s get you up to speed if you missed our video in September, after CRM stock had another sizable selloff and we reiterated our bullish view. It was a bumpy ride for CRM in 2024, but it eked out respectable returns anyways, about in-line with or just above the Nasdaq Composite.
The worry over the summer 2024 was that enterprise SaaS (software-as-a-service) providers were toast. The Klarna (a European fintech) story exemplified this. The Swedish buy-now-pay-later company said it was ripping out Salesforce and Workday (WDAY) software and replacing them with its own in-house developed AI software. However, we argued that Klarna is doing IPO window-dressing by dramatically cutting expenses, as well as drumming up some positive PR in the process, rather than Klarna being indicative of a broad-based trend.
The investor worries were laid to rest in the weeks following, though. How?
First, Salesforce went back on the acquisition hunt. Cash acquisition expense is only about half a billion dollars this last year, most of it the Own company in a bid to up its game in data management. Salesforce generates plenty of free cash flow to take care of such a small purchase in a single quarter.
And then hot on the heels of the Own purchase announcement, at the Salesforce Dreamforce annual dev conference, new data and AI deals were announced with Nvidia and Google. More bolstering of the data infrastructure software over at Salesforce is clearly in the works. The fruit of this labor, at least the latest flavor of it, is Agentforce. Agentforce is the AI agents software product that co-founder/CEO Marc Benioff has been trumpeting, while dogging on Microsoft‘s abilities in AI agents. The old Salesforce Einstein AI product isn’t out, of course, but Salesforce is really trying to make good use of (or help its customers make good use of) all the disparate data they may have stored in their CRM databases.
Hold up, what’s an AI agent? Is it like RPA?
No, RPA (robotic process automation) is a pre-defined repetitive task that a software-based bot performs. Like data entry on an electronic form, for example. An AI agent, powered by something like an LLM (large-language model), has a bit more autonomy in arriving at a solution or answer. A simple example of an AI agent would be a chatbot on a website asking if you need help finding anything.
Also FYI, part of our September update, we said we were parting ways with UiPath (PATH) stock. In case you were wondering, that position is indeed now gone from our portfolio.
What is Salesforce’s goal for the AI era?
Salesforce needs to build out its infrastructure software and data management capabilities for the AI era. And so, that’s what they’ve been doing with the Own acquisition and the dev programs with Nvidia and Google.
You can see Salesforce moving DOWN the software stack towards software infrastructure, so they can build more useful CRM data products for their customers. And as they build out the infrastructure software, they can do fancier things moving UP the stack, like the new Agentforce AI agent product.
This explains Salesforce’s Data Cloud development and marketing push. It seems they’re trying to get customers to use Data Cloud as a type of singular repository for both CRM and other data sources (like Google Worksuite) as a place to store and structure data — especially customer relationship data.
Will Agentforce be a success? Maybe. Benioff and company sure talk a good game.
It’s early days for this endeavor, and it certainly seems like Data Cloud is the most notable organic product launch at Salesforce for some time. But there’s also a lot of developer skepticism on the Agentforce product atop Data Cloud, at least on Reddit: https://www.reddit.com/r/salesforce/comments/1fzi0b4/is_agentforce_really_the_future_of_salesforce/
One of the issues with Salesforce is the constant upsell after the initial subscription deal is signed with a customer.
It also seems like Data Cloud isn’t a standalone product either, at least not yet. It’s more of an add-on. That makes sense for these early stages, and was hinted this past spring (see the below slide from our June update).
Once a customer is onboarded, Data Cloud can be used for a growing line-up of AI products, including Agentforce — which seems to be really a type of RAG (Retrieval-Augmented Generation). Think of RAG as like traditional data search, but with an LLM layer.
But for a company with vast stores of private data, an AI agent utilizing RAG can be powerful as a means to speed up employee productivity on whatever specific domain of work they’re doing.
To date, a lot of this development seems to be geared towards Salesforce’s Marketing Cloud segment, and perhaps a bit in Sales Cloud. Integration and Analytics, at some point, really ought to accelerate again — or it will be a missed opportunity.
The story gets a bit murkier still. Data Cloud — and the RAG Agentforce product built atop it — are a consumption-based pricing model. It’s not a SaaS subscription model. The more you use (or more data you store), the more Data Cloud costs. These are an upsell from the subscription plans (SaaS) that Benioff helped pioneer 25 years ago. There’s a free-to-use tier, but the cost of all the Data Cloud features ramps up quick. Clearly, AI cost controls in the data center are still a work-in-progress, making this new Data Cloud and Agentforce product best suited for large enterprises only.
The reason for the CRM trim in our portfolio
The ~40% run in the last few months certainly indicates Wall St’s warming optimism that Salesforce will be an AI software winner. However, as you likely know, revenue has slowed from a more-than-20% growth rate a couple years ago, to just 8% YoY growth in Q3. Wall St. is starting to price in some upside to that figure, factoring for a return to 10% revenue growth, or a little more, within the next two years.
Part of our thesis for the last five years in particular has been that Salesforce’s margins would increase, resulting in adjusted earnings per share increasing at an even faster pace than revenue. This trend has slowed, though, although GAAP profitability still has a lot of room for improvement.
CRM looked cheap earlier in 2024 after another selloff, but the recent run gets it back closer to a fair value — at least based on an assumption for low- to mid-teens % adjusted EPS and FCF-per-share growth over the next three years.
Which brings us to a somewhat unpopular opinion: It’s definitely time for another splashy CRM acquisition. AI, and perhaps more work on the data and infrastructure side, seems like it would be a good fit right now. Perhaps Benioff and co. are just being patient and waiting for the right company at the right price. If and when CRM finds the right deal, they have the cash to spend. This could really help speed up the Data Cloud and AI development, and put a path to accelerating revenue and earnings growth back in play.
But in the meantime, in this new era of “AI eating software,” our default on software businesses is always back to Vanguard IT ETF (VGT), which has averaged a mid- to high-teens % growth CAGR for some 20 years now. If we don’t think the software company can outperform this index, why not just add to the index fund instead? And given where Salesforce and CRM stock are right now, we’re starting to think it will perform in-line with VGT at this valuation.
Also, we are making some final moves to get ourselves to our 5% cash/eq goal for Q1 2025. This last action, trimming a small portion in CRM, should get us there. We will be keeping the bulk of our Salesforce position, but trimming will move it back down our non-chip stock holding list somewhere into position #6 – #10.
Which brings us to our final point: It’s a popular request, so we’re revealing our top 10 non-chip stock positions at the end of 2024 (and of course, you already know, our top long-time ETF holdings, VGT and VUG). Please don’t ask us for regular updates on this, we’re sticking with our rules and objectives for CSI. This isn’t meant to help you place big sweeping bets in your own portfolio. In fact, this list doesn’t look entirely different from what you’ll find in the top holdings of VGT. This is for a reference point, which we hope will become clear as we outline our “top picks” for 2025. We’ll also give a first look at our top chip stock holdings next week. From there, we’ll update as usual via videos on YouTube, with a lot more detail on Discord (via Semi Insider https://ko-fi.com/chipstockinvestor/tiers).
We’ll have more inital 2025 outlooks for you soon, as well as some new features we’re working on for Chip Stock Investor for the new year. Stay tuned!