In recent weeks, more than $1T was wiped from SaaS company valuations. Wall Street investors, spooked by the rapid rise of AI agents, began pricing in a dramatic shift. What if artificial intelligence doesn’t just enhance software, but replaces large parts of it? For companies like Salesforce – long considered the backbone of enterprise CRM – the narrative quickly became existential. If AI can reason, automate, and orchestrate workflows on its own, what happens to the platforms those workflows run on?
What followed this was a pretty quick whiplash. Anthropic’s enterprise event showcased deeper integrations with tools including Slack, Salesforce, and other major business platforms. Software stocks then rebounded, and analysts changed their tune, arguing that AI tools are only as useful as the systems they connect to. Almost overnight, the conversation shifted from “AI will replace SaaS” to “AI runs on SaaS” – so what did the sell-off get wrong?
The answer certainly isn’t that Salesforce is in any way immune to AI disruption, but that the market may have misread where the real change is happening. The more interesting question isn’t whether Salesforce will survive in an AI-first world, but how the balance of power, pricing, and control within the enterprise stack is evolving.
Wall Street Priced Extinction, Enterprises Didn’t
As SaaS stock has stumbled, the narrative around it is simple – AI agents will automate workflows, collapse seat-based pricing, and gradually erode the need for large enterprise software platforms.
On paper, that logic might make sense. If AI can actually build apps, orchestrate tasks, and reason across data, why keep paying for expensive enterprise licenses? Unfortunately for those who are team “SaaS is dying”, this assumption overlooks something fundamental about how enterprise tech actually works.
Tom Harding, VP of Contract (USA) at Smart4 Cloud, spends his working days helping large businesses hire Salesforce, ServiceNow, and GCP professionals. From his vantage point, the idea that enterprise SaaS is about to be “unpicked” feels disconnected from the reality on the ground.
“If you’ve spent years embedding Salesforce into your business-critical workflows, it’s incredibly difficult to remove,” Tom explained. “You can say whatever you like about agentic AI, but how are you realistically going to rip out Salesforce? They own the data – Data Cloud is being actively pushed and adopted – and that’s the key. If they control the data and are already deeply integrated into your systems, they have a significant first-mover advantage over anyone else, because they already have it all.”
This is where we can start to recognize the clear disconnect happening between theory and reality. The sell-off assumed AI would displace platforms like Salesforce quickly, but Tom believes it’s the opposite. Enterprise incumbents, if anything, are protected. Not because AI isn’t advancing, but because large companies are fundamentally risk-averse.
Tom said: “I honestly believe the limiting factor is going to be people in leadership – their beliefs and their risk appetite – rather than the technology’s capabilities. I don’t think the technology is fully there yet, and it will take a while to get there. But while these people are in leadership positions, it’s going to take time for them to want to do it and to feel comfortable taking the kind of risks their shareholders will be happy with.”
The intangibles mentioned by Tom don’t show up in stock charts. Large enterprises aren’t making architectural decisions based on purely what’s technically possible – they have to factor in roadblocks like integration complexity, governance, brand assurance, among many other things.
Tom pointed to a familiar enterprise pattern, where companies often choose global consulting firms over boutique specialists, simply because they’re a safer bet (rather than being better per se). As he says quite firmly: ”No one ever got fired for hiring IBM.”
This same logic can be applied to AI adoption. While smaller, seat-based SaaS companies serving SMBs may be more exposed to disruption, Tom sees enterprise platforms as fundamentally “sticky”, given that they own the workflows, integrations, and (increasingly) the data.
“I think building agents inside Salesforce will be easier for these companies than going externally and creating agents that interact with Salesforce,” Tom explained. “If you’re a SaaS-based company serving small and medium businesses where users just log in, then you might be in trouble.”
That stickiness doesn’t mean Salesforce is immune to change. Tom is clear that the seat-growth era is likely over, saying: “Salesforce will need to adapt its pricing model. They’re not going to suddenly grow seat-based revenue.”
But that’s more of an evolution problem than the extinction event being so heavily marketed. The sell-off treated AI and Salesforce as mutually exclusive outcomes. Enterprise reality suggests they are more likely to coexist – at least for now.
Is SaaS Being Replaced, Or Repriced?
If Tom’s view reflects enterprise caution, Senior Industry Analyst Vernon Keenan’s perspective adds another layer – that the sell-off wasn’t just about AI hype, but also about economics.
From his standpoint, the reaction in markets has often been “overdone”.
“It seems like there’s a bit of a roller coaster happening on Wall Street,” he said, pointing to the sharp swings between “AI bubble” fears and the sudden belief that AI is powerful enough to wipe out entire industries.
One core driver of the volatility, he argues, is uncertainty around pricing models, due to the wider industry not adequately addressing the transition from seat-based pricing to consumption-based pricing.
This is a legitimate concern – if companies become more efficient and reduce their headcount, then they’ll need fewer software seats. But that’s more of a monetization issue than it is a displacement issue.
Where Vernon sees investors misreading the situation is underestimating the complexity of enterprise systems. It’s quickly grown from a CRUD database (a system that Creates, Reads, Updates, and Deletes records) to a much bigger beast – which AI is likely unable to replicate.
He said: “You have Data 360, MuleSoft, and other features of Salesforce that get put together. The way those things operate underneath is not going to be vibe-coded. For people to take a blog post from an investor that says, ‘I made a CRM for my startup in a weekend’… let’s take those with a grain of salt.”
Vernon is particularly skeptical of claims that lightweight AI tools can simply replace deeply embedded enterprise platforms as well.
“I saw somebody put Stripe on a list of potential ‘SaaSpocalypse’ victims,” he said. “That’s as dumb as you can get!”
In essence, the dividing line, in his view, isn’t “SaaS vs AI”, but whether a company owns a true system of record.
“People who don’t really carry much of a system of record right now are the ones that are vulnerable,” he said. Overlay AI tools that sit on top of existing infrastructure are already showing higher churn rates, in some cases losing as much as 75% of customers within a year.
Meanwhile, platforms that collect, secure, and structure original data – like Salesforce – retain structural durability. That doesn’t mean nothing changes, but it does indicate that the battleground may shift upward.
If AI becomes the reasoning layer, it still needs a place to live, to store data, and to execute workflows. The question, then, is not whether Salesforce disappears, but whether it successfully evolves from a seat-based SaaS vendor into a platform that captures value in an AI-orchestrated stack.
Structural durability, however, doesn’t automatically mean structural acceleration.
As Salesforce Architect Timo Kovala puts it: “There are some similarities to mainland China if you think about it. They rose from dismal performance to become one of the world’s biggest economies in just a couple of decades. But such growth cannot be sustained indefinitely.”
That perspective reframes the sell-off. The market may not be pricing extinction, and instead may be pricing maturity.
Salesforce Q4 Earnings: A Reality Check for the “SaaSpocalypse”?
Salesforce’s Q4 results arrive at an interesting moment in the debate. On the one hand, the company reported 29,000 Agentforce deals and $800M in Agentforce ARR, up sharply quarter-over-quarter. Agentforce and Data 360 bookings were heavily driven by expansion within the existing customer base, and 2.4 billion Agentic Work Units (AWUs) have now been delivered.
On the other hand, the market reaction was relatively muted. Shares dipped in extended trading after Salesforce issued fiscal 2027 guidance slightly below Wall Street expectations. Revenue is still growing at roughly 10% year-over-year, but investors appear focused less on survival and more on acceleration. What does this really show?
First, it suggests that AI is not replacing Salesforce but instead is being layered into it. Agentforce growth is coming largely from existing enterprise customers expanding usage inside the platform, not abandoning it. That aligns far more with a repricing narrative than an extinction one.
Second, it reinforces the transition question. The market is asking how quickly AI-driven products can offset slower seat-based growth. In that sense, Salesforce’s Q4 results definitely don’t validate the “SaaSpocalypse”, but it does complicate it. AI momentum is real, but so is investor caution.
Final Thoughts
The SaaS sell-off assumed speed – that AI would quickly unpick enterprise software and collapse the old model. What both enterprise hiring signals and structural analysis suggest is something slower and more complex.
Salesforce isn’t immune to disruption, but neither is it easily displaced. The real shift is the economic impacts, away from seat-based growth, and toward consumption and AI-enabled workflows.
As Timo puts it: “AI may be a catalyst in the current turbulence, but some of this performance deterioration was likely inevitable. Salesforce is a mature platform, and mature companies naturally become more risk-averse – they simply have more to lose.”
The question now is this: Who captures value in the next layer of the stack, and how quickly are enterprises willing to adapt?
