Is AI Ending Enterprise Partner Careers? Salesforce May Be Just the Start

Ben McCarthy, a top Salesforce analyst, just declared the end of Salesforce careers as we know them. He makes a strong case. The question worth asking is how far it extends. Is Salesforce the leading edge of something that will hit every enterprise ecosystem, or a special case being read as a trend?

The Salesforce story McCarthy tells is well documented: a platform that took a few weeks to learn, a market that paid well to learn it, a flood of certified professionals, and a saturation rate his sources put at 330%, which works out to more than three qualified people for every open role. His real point is broader than Salesforce. It comes down to one word he uses halfway through the piece: arbitrage, the temporary gap between what a skill is worth and what it costs to acquire. That kind of gap always closes eventually.

The same arbitrage underwrote the entire enterprise partner economy. Salesforce just happens to be where it is being priced out first and most visibly.

The arbitrage was never Salesforce-specific

Consider what the last fifteen years rewarded. Microsoft built a partner economy on Dynamics and the Power Platform, where a certification and some willingness to learn could turn into a consulting day rate. ServiceNow did the same with ITSM and workflow. SAP and Workday did it at the high end, where implementation specialists earned premium rates because the platforms were complex enough to gatekeep. The system integrators (Accenture, Deloitte, IBM, Capgemini, Cognizant, and the Indian majors TCS, Infosys and Wipro) industrialized it: take complex work, break it into repeatable processes, train graduates by the thousand, and bill clients a margin on the gap between cost and value.

That model created campuses, vendor ecosystems, conference circuits, and millions of careers. It was a good model. It was also an arbitrage, and like every arbitrage, it depended on the gap staying open.

McCarthy shows the gap closing through human oversupply: demand for Salesforce professionals fell while certified supply kept climbing, year after year, until saturation hit triple digits. That is the slow version. AI is the fast version, because AI does not only add more people to the supply side. It manufactures supply of the work itself: the configuration, the integration scaffolding, the boilerplate, the first-draft delivery. Those are the exact tasks the certification pipeline was built to fill.

This is not 2023

The current round of layoffs is not a repeat of the post-pandemic correction. The 2023 cuts were companies unwinding pandemic overhiring. The 2026 cuts are structural, and the people making them are saying so.

By mid-2026, tracking site Layoffs.fyi had logged well over 100,000 tech job losses for the year, running roughly a third ahead of 2025's pace. The stated cause is what changed. Multiple trackers now find that close to half of 2026's layoff events explicitly cite AI or automation as a driver, a category that barely registered two years ago. Outplacement firm Challenger, Gray & Christmas tied tens of thousands of planned cuts directly to AI initiatives in the first months of the year.

The clearest signal for the partner ecosystem came from inside it. Accenture, the firm that advises the world on AI transformation, cut close to 22,000 roles across its 2025 fiscal year under an $865 million "business optimization" program. CEO Julie Sweet said the company was exiting people on a compressed timeline where reskilling was not a viable path. TCS ended its fiscal year down more than 23,000 staff on a net basis, with its chairman suggesting that a company of half a million employees might one day run half a million AI agents alongside them. When the integrators that built the arbitrage start automating their own delivery, that tells you where this is heading.

The bifurcation: same split, every ecosystem

McCarthy is right that "your job is safe" is the wrong question, because the role title tells you very little. What matters is which layer of the work you sit on.

His Salesforce role data is the clearest example. Demand for admins rose 14%, but supply rose 47%, so the floor keeps dropping. Developer demand actually fell 12% while supply grew 20%, the strongest sign in the dataset that AI is eating configuration and code work. Technical Architect demand jumped 27% against only 4% more supply, and Solution Architect demand rose 21%. Salaries follow the scarcity, with architects clearing $165,000 to over $190,000 while the entry tier compresses.

That pattern holds across every other ecosystem:

• The "configure and deliver" layer (entry-level admins, junior consultants, anyone whose value was knowing how to click through a platform quickly) is where AI bites hardest and supply is most saturated. It is also the layer the integrators are automating first.

• The "connect it to the business" layer (architects, integration leads, the people who can make several systems and an AI agent behave as one coherent thing) is where demand is rising against thin supply. The complexity did not disappear. It moved up the stack.

The platform stopped being the moat. Knowing Salesforce, or Dynamics, or ServiceNow, was the asset for fifteen years. It is now table stakes. The skill that holds value is orchestration: connecting AI capability to a business outcome across systems that were never designed to talk to each other.

Follow the budget, not the platform

McCarthy's strongest advice is to attach yourself to the AI budget, and once you widen the lens, it is clear why that matters more than anything else.

The partner economy was funded by the enterprise software and digital-transformation budget: cloud migrations, CRM rollouts, SaaS sprawl. That budget is now being repriced and rerouted. Andreessen Horowitz's survey of around 100 enterprise CIOs found LLM budgets expected to grow roughly 75% year over year, with the share treated as experimental "innovation" money falling from 25% to just 7%. Gartner projects global AI spending approaching $2 trillion in 2026. A meaningful chunk of that is not new money. It is being carved straight out of existing software line items.

The displacement is already concrete. Retool's 2026 Build vs. Buy report found that roughly a third of enterprise teams had already replaced at least one SaaS tool with a custom internal build, and nearly four in five planned to build more in the year ahead, concentrated in exactly the workflow-automation and internal-tooling layers that enterprise platforms occupy. Integration work that used to be a multi-quarter engagement for a billable team is becoming, in the words of one CIO analysis, a matter of prompts and pull requests.

So the budget that paid for your certification is being routed to AI agents, AI infrastructure, and the people who can connect those agents to business outcomes. The professionals best positioned for the next five years are not the ones defending their platform. They are the ones standing on the line item that is growing.

Where the careers actually go

It would be easy to stop here and call it bleak. It mostly is, if you do nothing. But the same data shows where the value moved, which is the part worth acting on.

The companies cutting hardest are signaling the new skill by where they spend. Accenture has trained more than 550,000 staff on generative AI and tens of thousands more on agentic AI, and it is hiring again, selectively, for AI, data and orchestration roles even as it exits the layer it is automating. Infosys raised its outlook on the back of AI-led deals while leaner rivals shrank. OpenAI is building a multi-billion-dollar enterprise deployment arm and hiring forward-deployed engineers by the hundred, and Anthropic keeps hiring aggressively. The pattern is consistent: organizations building and deploying AI are growing, and organizations whose value was delivering the old work are shrinking.

For anyone with an ecosystem career, that points to one move, the same one across Salesforce, SAP, ServiceNow and the SI floor.

Stop being the person who configures the platform. Become the person who makes AI produce a business outcome on top of it. In practice that means the integration and architecture layer (the systems thinking AI is worst at and enterprises need most), the orchestration of agents across tools, and the governance, observability and security work that grows because everything else is being automated and nobody wants to be accountable for an agent that goes wrong. Those are the roles sitting on the AI budget rather than the SaaS budget being cut from under them.

But is this actually happening?

It is worth holding this argument up to its strongest objections, because the case is not airtight.

Start with attribution. When close to half of 2026's layoff events name AI as a driver, that is what companies say, not necessarily what is true. Outplacement firm Challenger, Gray & Christmas tied a much smaller share of actual cuts, closer to one in seven, directly to AI. Deutsche Bank analysts have described a pattern of "AI redundancy washing," where firms use AI as cover for reductions that are really about overhiring, weak revenue or investor pressure. Even OpenAI's leadership has conceded that some companies overstate AI's role. If that is what is happening, then part of the decline is the ordinary business cycle wearing a more fashionable label.

The integrator story is also more mixed than the layoff headlines suggest. The same Accenture that cut 22,000 roles added more than 4,000 in the first quarter of its 2026 fiscal year, reported roughly $20 billion in new bookings, and told investors it expects to keep hiring for AI and data work. Infosys raised its revenue forecast on the back of AI-led deals and a record pipeline of large contracts. TCS flagged strong 2026 demand even as its headcount fell. Read generously, this looks less like an industry collapsing and more like one rotating its talent, which is roughly what Accenture's own leadership calls it.

There is history to reckon with too. McCarthy wrote a version of this warning back in 2024, and the ecosystem did not disappear. Predictions that AI will end developer careers have a long record of arriving early and overshooting. And the advice to "attach yourself to the AI budget" is itself a narrative that happens to serve the vendors selling AI. If the returns on all this spending disappoint, those budgets can contract as fast as they expanded, and the orchestration roles everyone is told to chase could end up just as crowded as the admin roles they were supposed to replace.

None of this proves the thesis wrong. The saturation data is real, the budget shift is real, and the split between commodity delivery and high-complexity work shows up in too many places to dismiss. But "the arbitrage is closing" and "AI is ending these careers" are not the same claim, and the gap between them is exactly where reasonable people disagree.

So where does this leave us?

The honest answer is that nobody knows yet, and that is the point. The platforms are not going away. Salesforce will not be vibe-coded out of existence, and neither will SAP or ServiceNow. What is genuinely in question is whether the work that funded a generation of careers on top of them is being permanently repriced and pushed up the stack, or simply rotated through another cycle that will look ordinary in hindsight.

If you believe the first reading, the move is clear: stop identifying as a platform expert and start identifying as the person who turns AI capability into outcomes a business will pay for, wherever the budget sits. If you believe the second, the smarter play might be patience, real depth on a platform that is not going anywhere, and a healthy skepticism toward anyone telling you to chase the AI line item.

We lean toward the first reading. But we would rather start that argument than settle it. Is Salesforce the leading edge of something that hits every ecosystem, or a special case being mistaken for a trend? Where do you think the careers actually go? Tell us where this read is wrong.