You can also listen to this podcast on iono.fm here.
Welcome to the Supernatural Stocks podcast on Moneyweb, with your host The Finance Ghost – your weekly fix of local and international insights for investors and traders.
At the Barclays 23rd Annual Global Technology Conference in early December 2025, Miguel Milano, president and chief revenue officer of Salesforce, waxed lyrical about the performance of the company. He’s a sales guy and a passionate Argentinian, so he gets as excited about the company as he does about football.
At the 44th Annual JPMorgan Healthcare Conference in January, Salesforce sent Mark Sullivan along to represent them. He leads the Life Sciences vertical at the company. There’s already a lesson in that around how Salesforce has built a multi-industry moat, but more on that later.
These conferences tend to feature panel discussions where the moderator only throws softballs that are easy to hit out the park, but the transcripts are still a helpful opportunity to understand more about the bull case.
And in a world that seems to be assuming that software-as-a-service [SaaS] is now a broken business model, we need to dig everywhere we can for nuggets of information.
The last formal earnings release by the company was at the beginning of December. And if you really want to dig deeper and deeper, there are other transcripts available from around that time.
Read/listen:
But before you wonder about whether Salesforce is on a charm offensive right now, it’s worth noting that they regularly attend conferences and host special calls. There’s a lot to read, should you so desire.
But why do I care so much about Salesforce right now?
Finding the baby in the bathwater
Salesforce is recognised as the company that showed the world how to turn SaaS into big business.
The company was founded in 1999 – just before the peak of the dot-com bubble – and it figured out how to deliver enterprise-grade solutions through a web browser, rather than via software that you install on your device.
Founder Marc Benioff is still the CEO of the company. He is regarded as something of an internet guru. We aren’t talking about an obscure or niche company here.
Despite this track record, the onset of the AI era – and particularly the promise of ‘vibe coding’ – has obliterated returns.
Over 10 years, the compound annual growth rate [CAGR] in Salesforce’s share price has been just 11.5%. That’s still good compared to many other investments, sure, but an S&P 500 ETF would’ve achieved a CAGR of 13.6% over this period – and that’s without dividends.
Fund management careers are made and broken over just 50 or 100 basis points of outperformance.
To see more than 200 basis points of outperformance by the index over a decade is significant. Or, put differently, 200 basis points of underperformance by Salesforce. Ouch.
The problem is that over the past five years, the CAGR for Salesforce’s share price has been -5.6%.
ADVERTISEMENT
CONTINUE READING BELOW
While Big Tech has pulled away and names like Nvidia have become bigger in the S&P 500, Salesforce has fallen by the wayside.
And it’s not just Salesforce.
Adobe has a 10-year CAGR of 12.4%, splitting the difference between Salesforce and the S&P 500. But the five-year CAGR at Adobe is -11.6%, so you can see that it’s suffered even more pain in recent years than Benioff’s creation.
Adobe just happened to have a much stronger first half of the decade under review, as their adoption of the SaaS model came much later than what we saw at Salesforce.
Read: AI disruption fear sparks investor scrutiny of software stocks
What about Intuit, the owner of excellent platforms like QuickBooks, Mailchimp and TurboTax? This is a great business, with a 10-year CAGR of 14.7%. We’ve finally found a name that has beaten the index!
But brace yourself, because although the five-year CAGR is only -1.5%, and thus the best of the trio, the year-to-date performance will shock you.
Intuit has shed 39% of its value thus far in 2026. When people describe a bear slashing its paws at a share price chart, this is what they mean.
Adobe is down 22% this year and Salesforce has received a 27% beating. All bad, but Intuit by far the worst.
In summary, the sector is being taken to the woodshed. Or, in a more South African context, it’s been bliksemed.
Fine. But does that mean they all deserve it, or is there an opportunity here to bravely buy the sector or pick a couple of stocks?
Read/listen:
Because if there’s one thing the market knows how to do, it’s to throw babies out with the bathwater. And those babies can be very good investments at the cheaper multiples dished up by market panic.
Note the careful use of the word “cheaper” rather than “cheap” – but also more on that later.
What is the investment case?
I’m going to focus on Salesforce here. This isn’t because Salesforce is the only name worth considering, but rather because it feels most appropriate to base this discussion on the company that created SaaS as we know it.
At the Barclays event, Miguel Milano put forward a bull case based on the “last mile” in the agentic enterprise era. To understand this, we need to know something about what agentic enterprise means.
The reference to “agents” has nothing to do with black suits in the Matrix, even if it does feel like we are living in the future right now. No – it relates to the use of different large language models [LLMs] to do specific things for you, with agents trained for a particular purpose and with strict rules.
That’s the theory, at least.
And with the reference to agentic enterprise, Milano is reminding the world that Salesforce is embedded in huge companies worldwide, not just your friendly local start-up.
ADVERTISEMENT:
CONTINUE READING BELOW
For these agents to exist, they need to be trained on data. For them to subsequently be useful, they need ongoing access to that data. Enterprises care deeply about cybersecurity, so this creates entirely new layers of risk.
Microsoft’s argument in favour of Copilot is that it uses the “walled garden” that is your Microsoft account and your critical business data.
In other words, security is one of the key differentiators in this space. It’s one of the reasons why I’m not sure that a company like Salesforce just disappears in this era.
Now, back to the last mile, which Milano reckons has four components. I’ll run through them quickly.
The first component is whether the data is actually correctly organised in terms of metadata. Don’t underestimate this. I’m certainly no techie, but I have learnt that AI reads data differently to how your eye may see it, as confirmed by me asking my Copilot about how it ingests information from certain websites.
It’s possible to do this yourself with your data, but Salesforce has teams of engineers making sure that this works properly for all clients. It doesn’t seem very efficient or reliable for clients to do that themselves.
Read:
The second element is execution. If I read between the lines, he’s referring to things like AI hallucinations and how different models will do different things when presented with similar facts and prompts.
As SaaS applications are effectively a type of standard operating procedure for a company, it stands to reason that such platforms are the right foundation on which to build automations, rather than hoping that AI will apply the right logic every time if given enough freedom.
Read/listen:
The third component is the humans involved. AI can’t do everything. Salesforce already has over 150 000 customers worldwide and many more paid seats, serviced by 76 000 employees. That’s a pretty big ecosystem of people who know and understand the platform.
It doesn’t seem very efficient for companies to vibe-code an alternative and then train staff on it.
And if you believe that this is possible and that AI will be able to do literally everything, including at enterprise levels, then I’ve gotta tell you that your returns on your Salesforce investment will be the least of your problems – I suspect we would then be living in a dystopian catastrophe.
Read: Asking ChatGPT vs Googling
On a lighter note, we now reach the fourth component: governance and compliance. This speaks directly to the cybersecurity point that I raised. Privacy, access rights and the like are all relevant here.
The Salesforce view is obviously that AI cannot do this last mile. That is clearly the turkey voting for Christmas. They have to make you believe this, otherwise they don’t have a business. But I do tend to agree with them.
And this doesn’t mean that Salesforce is blind to the AI opportunity.
No, their efforts to be the last mile means that they need to plug beautifully into the agentic enterprise era, including with in-house agents offered to customers.
The sector moat
I don’t doubt that your friend who runs a techie start-up with a team of devs would love nothing more than to save on Salesforce fees. And if they have some spare development time, they just might build something.
At the margins, the AI era can hurt Salesforce, because the last mile considerations aren’t as important in start-ups looking to save money as they are in enterprises.
ADVERTISEMENT:
CONTINUE READING BELOW
In enterprises, people are using shareholder money to make purchase decisions in a classic cover-my-you-know-what strategy.
Remember how nobody gets fired for hiring IBM? People definitely get fired and probably sued for vibe-coding with sensitive data. But they don’t get fired for using slick new tools offered by Salesforce.
And when you add on the sector specialisation at Salesforce, the moat gets more interesting. Do you think that the attendees at that healthcare conference are vibe-coding solutions for patient data?
Read: Get me out’: Traders dump software stocks as AI fears erupt
When I record podcasts with execs at large companies, I always have to run a firewall test beforehand to make sure the recording software won’t be blocked by their systems. They have strict controls over their tech stack and they only stick to the big names.
And when the tech stack involves sector-specific solutions developed over the years and embedded in these organisations, it feels even harder for AI to disrupt this.
Risk, but at what price?
Is there a moat? I think so. Could I be wrong? Absolutely. It’s not called “taking on equity risk” for nothing.
And part of the reason for the sharp sell-off in this sector is that the valuations were steep. In the peak of the pandemic, Salesforce was on a price/sales multiple of over 12x. It then came down to a range of 6x to 8x. Currently, they are at just 4.3x.
Free cash flow margin has always been a sticking point, as Salesforce is one of the biggest offenders when it comes to using stock-based compensation instead of cash remuneration, leading to a significant increase in the number of shares in issue over time and artificially boosting the free cash flow margin.
Read/listen:
This trend has started turning recently though, with the number of shares outstanding actually dropping over the past couple of years.
And to be clear, all of Salesforce’s competitors also make use of stock-based compensation, so this is comparing apples with apples – and perhaps even to Apple?
With a free cash flow margin that has been as high as 38% recently, Salesforce generates cash. And thanks to the sharp drop in the valuation, you can buy it on a free cash flow multiple of below 12x. The average over the past five years? 26.2x.
Sure, SaaS faces some risks. But is it half as good as it used to be?
Does it deserve to trade at free cash flow multiples that I’m used to seeing at industrial companies? I’m not so sure.
If the market is wrong on SaaS, there is money to be made somewhere.
And if the market is right, I don’t think any of us have jobs anyway, so it kinda doesn’t matter. It does make for a pretty interesting analysis in this space, but I’ve also learnt that you don’t need to try and pick the bottom perfectly.
It can make sense to just wait for the momentum to start to the turn. For Salesforce, ramping up the share buybacks and driving the message home about the valuation might be the way to do it.
Follow Moneyweb’s in-depth finance and business news on WhatsApp here.
#Softwareasaservice #world