Is anyone else getting a little fatigued by the narrative that AI could one day take everything from you – your earning power, your ability to do your job, maybe even your life? We recently covered the uncomfortable reality that, technically, no one in the Salesforce ecosystem is completely safe from being replaced. The conversation usually focuses on entry-level roles, but we also explored the possibility that organizations may eventually need fewer senior professionals too.
But what happens if we flip that conversation on its head? What would happen to the Salesforce job market if AI itself ran into trouble? We’re seeing trillions of dollars being poured into AI infrastructure and products, while companies race to bolt AI onto almost every part of enterprise software (hello, Claudeforce).
At the same time, leaders of some of the world’s biggest AI companies are openly discussing the possibility of catastrophic outcomes and “pacing the frontier”, while economists, central banks, investors, and researchers are asking whether parts of the AI economy have become increasingly speculative.
So, could that enthusiasm eventually be pushed too far? And if it is, could we see an AI bubble burst over the next few years? Even if that happened, AI wouldn’t suddenly become useless. I imagine it would look much more like the dot-com crash: the internet didn’t disappear, but the market became far more discerning about which companies had genuine technological value and which valuations actually made sense.
If AI goes through something similar, where does that leave Salesforce professionals? Would they be relatively well insulated from the fallout, or dragged down with the rest of enterprise tech?
What Would an AI Bubble Bursting Actually Mean?
Before we get too carried away with the idea of an AI bubble bursting, it is worth being clear about what that would actually mean. Because I do not think the realistic version of this story is everyone waking up one morning and deciding AI was terrible after all.
There is plenty of evidence that something real is happening underneath the hype. Enterprise adoption is growing, models keep getting more capable, and genuine productivity gains are starting to show up. One recent paper, ‘Boom, Bubble, or Buildout?’, argues that AI can become a genuine technological revolution while parts of the market around it still become speculative, which feels like a more sensible way to look at it.
If we look back at the dot-com crash, Amazon surviving did not mean Pets.com deserved the valuation it had been given. The internet was obviously going somewhere, but investors had just become far too enthusiastic about where every company attached to it was going – and AI could end up looking similar.
Barak Goldstein, Co-founder of June and previously Bonobo AI (which was acquired by Salesforce), thinks this contrast matters. He told SF Ben: “I think some valuations reflect expectations that will be difficult to meet. But that can be true while the technology continues to create real value. A correction in valuations wouldn’t mean enterprise adoption has stopped.”
He also made another point that is easy to overlook – a pullback in funding for AI companies does not necessarily mean businesses suddenly stop spending on AI themselves. What may change is what gets funded.
“Within enterprises, I’d expect overlapping tools, pilots without a clear business owner and projects without a credible path to measurable value to face pressure first,” Barak explained. “At the same time, successful deployments could receive substantially more funding.”
And as many of you may know, those signs of tougher attitudes towards AI may have already arrived. Brett Carneiro, who works with enterprise clients on Salesforce and AI strategy, told me there has been a noticeable change in the questions businesses are now asking.
“A year or two ago, a lot of AI projects were about adoption for its own sake, corporate-wide pushes to get teams hands-on, figure out ROI later,” Brett told SF Ben. “Now clients want the ROI framework built before anyone touches a keyboard.”
Alex Borland, an independent Salesforce consultant, has seen something similar among his own clients: “With budgets being scrutinized for a lot of businesses, it has made phased delivery more important,” he told SF Ben. “Clients want to see a practical outcome from each stage rather than commit immediately to a large program with benefits that may be a year or more away.”
So this is probably what an AI correction would actually look like. AI wouldn’t disappear, but patience definitely would. This could mean weak pilots getting cancelled, companies questioning overlapping tools, or infrastructure spending facing more scrutiny. It would, for example, give your CFO more ground to start asking hard questions around what something actually saves and earns.
As Brett put it, the burden of proof has moved from “prove it worked” to “prove it before we start”, and if that becomes the norm, Salesforce enters that environment from a very different starting point than a company whose entire existence depends on the AI boom continuing.
Salesforce Was Strong Before AI
Now this is where Salesforce looks quite different from a lot of the companies that risk sitting at the centre of the AI boom.
It goes without saying that Salesforce was the biggest CRM company before generative AI took over every earnings call, keynote, and product roadmap, and it still is now. Agentforce (and now Claudeforce) may dominate the conversation, but Salesforce as a business is not dependent on Agentforce suddenly becoming the future of work.
The company’s latest numbers make that pretty clear. In Q2 FY27, Salesforce generated around $10.8B in subscription and support revenue (including Agentforce), which represented roughly 95% of its total revenue for the quarter. It also had around $66.3B in remaining performance obligations. Agentforce and Data 360 are growing quickly, but they are still sitting on top of a much larger existing business.
This matters because most Salesforce customers are not logging in every day wondering whether they should keep Salesforce because Agentforce exists. They’re still very much using it because sales teams, service teams, customer data, integrations, and years of business processes already live inside it.
Our own Admin Survey reflects this, with Sales and Service Cloud still the most widely used products among respondents, while Agentforce usage, although growing quickly, is still much lower.
This leads us to another useful historical comparison from the dot-com crash. When the crash happened, Oracle and SAP both saw demand for new software take a huge hit. But the maintenance and support side of their businesses proved much more durable. It demonstrated that customers could stop buying something new far more easily than they could stop running the systems they already depended on.
I’m not saying Salesforce would follow the exact same path if AI markets corrected, but the basic logic still makes sense. Companies might delay a new transformation project, or cut an AI pilot. But ripping out the CRM that half the business relies on is a very different decision to make.
When I discussed this with Brett, he believed the argument goes beyond Salesforce simply being difficult to replace, saying: “It’s not just that everyone already uses Salesforce. It’s that the value proposition is quietly shifting from the dashboards and page layouts most people associate with it, to the engine underneath, the data and governance layer that makes it safe for any AI model to plug in and act.”
Barak made a similar point about the value of that installed base: “It provides a strong foundation. Companies depend on these systems, and there’s real value in understanding the processes, data and decisions built into them over many years.”
This becomes important also as we move from Salesforce the company to Salesforce careers.
Our Admin Survey also found that 72.9% of respondents work directly for Salesforce customers, while 83.5% are full-time employees.
An internal admin is not there because Salesforce is fashionable. They are there because somebody still has to manage access, fix automation, look after integrations, deal with technical debt, support users, and keep the CRM working.
That is also what Alex Borland sees from the client side. “Once Salesforce is embedded in a business, somebody still needs to keep it secure, reliable and aligned with changing business processes,” he told SF Ben. “Users will still need support, automations still need maintaining, data quality still matters, and management still needs accurate information.”
If budgets tighten, the question is less “AI or Salesforce?” and more “what can we actually afford to stop doing?” That probably gives core platform work a stronger case than speculative projects, but it does not make every Salesforce role untouchable.
Not All Salesforce Careers Would Be Bubble-Proof
There is a temptation at this point to take the logic a little too far – Salesforce is deeply embedded inside businesses, and companies have years of data, processes, and internal knowledge tied up in it. So it would be easy to look at that and conclude that Salesforce careers must be fairly well protected too, right?
In fact, I don’t think it’s that simple. A company may decide that Salesforce itself is essential while also deciding it needs fewer people around it, fewer contractors, fewer consultants, or fewer new projects – those things aren’t contradictory. As Barak told SF Ben: “I wouldn’t equate the resilience of a platform with job security for every role around it.”
Naturally, if budgets tightened, it may be the case that the pressure could fall first on work that is easier to delay, like new implementations, large transformation projects, or experimental AI projects. All three sit in a different category from the work needed to keep an existing org operating properly. They might not disappear all of a sudden, but they are easier to question.
Then there’s the AI angle. If the amount of manual effort needed to make changes in Salesforce falls, demand for Salesforce work could stay strong while the number of people needed to deliver that work changes. Barak made that point directly too: “Demand for changes to Salesforce could increase while the manual effort required for each change decreases. Both can happen at the same time.”
Alex sees the same distinction from a slightly different angle: “For me, the distinction is less about roles – admins vs. developers vs. consultants – and more about whether someone is helping the business get value from Salesforce.”
Brett also pointed out another skill becoming more valuable in this environment: “The clearest signal is knowing when not to use AI, and that’s actually the most defensible skill right now.”
That might sound slightly backwards after two years of being told everyone needs to become more AI-focused, but it makes sense. If a Flow or Apex solves the problem, the person who can recognize that may be more useful than the person trying to force AI into every use case. That is where the conversation starts to move away from job titles and towards judgment.
Our recent Architect Survey found that 42% of respondents work for Salesforce consultancies, which means a meaningful part of the market is still tied to project-based spending. It also found that architects spend around 45% of their time working with technologies outside Salesforce.
The second figure there is probably the more revealing one – the safer Salesforce professionals may not be the ones who know most about Salesforce or AI, but it could be the one who understands Salesforce alongside other key aspects, like data, security, and how the business really works day to day.
That fits closely with Alex’s view. He told SF Ben: “Technology is only one part of solving a problem. You also need to understand the process, the people involved, the risks, the data, and what success actually looks like,” he said. “Those wider skills have often proved as valuable in my Salesforce work as the technical knowledge.”
Salesforce can be sticky without every Salesforce role being sticky. The more defensive careers are probably the ones closest to decisions and systems a business still cannot afford to get wrong.
Final Thoughts
While these topics are always interesting to write about, we usually come away without a definitive answer. An AI bubble bursting, however realistic that scenario may be, would leave Salesforce professionals in a pretty strange place.
There is some reassurance in the fact that Salesforce is much bigger than its AI strategy. The platform existed long before Agentforce, and plenty of the work Salesforce pros do still would need doing. But there would also be an awkward sense of reversal. The ecosystem has spent years being told that AI would change how Salesforce is built, implemented, and used. People have earned certs, companies have invested in projects, and careers have completely shifted in that direction.
In some cases, that spending suddenly looked overdone, raising uncomfortable questions about how much time and money have been poured into projects that never really delivered what was promised.
For me, it’s also worth considering how the ecosystem would react. There is already some fatigue around Salesforce’s current direction, particularly the sheer amount of attention being given to AI. Could you imagine if the market did correct badly and customers started questioning the value of those investments? Salesforce would have to manage customer confidence just as much as the financial consequences.
Would customers still buy into the next big vision from Salesforce as readily? Would Benioff’s AI push look ambitious in hindsight, or excessive? And would an already complicated relationship between Salesforce and parts of its ecosystem become even more strained?
Of course, this is all speculative. Right now, Salesforce is still going full steam ahead with AI. Claudeforce is here, Agentforce remains central to the company’s direction, and enterprise AI is not going away. But technology cycles have a habit of looking obvious right up until they don’t.
Feel free to reach out to me on LinkedIn to continue the conversation. And if you have your own take on where Salesforce careers would sit in an AI downturn, I’d be interested to hear it for a future SF Ben article.
