For most of my career, the enterprise software conversation followed a predictable script. A CRM, an ITSM platform, an ERP system, these were not things you built. They were things you bought, implemented and lived with. The reasoning was sound: Building them required enormous capital, long engineering tenures and specialized domain depth that only a handful of vendors could concentrate. So, enterprises paid the toll. And if you have ever read a SaaS earnings report, you know what that toll looks like on the margin line.

That calculus is breaking down. Not collapsing overnight but breaking down in a way that every CIO I speak with is starting to quietly acknowledge.

The reason is simple. AI has fundamentally changed the economics of writing software, and with it, the central premise of the enterprise SaaS model.

The moat was never the code

Here is the uncomfortable truth the last 24 months have surfaced: Most enterprise platforms are not technological marvels. An ERP, stripped to its essence, is a workflow engine with a data layer a system that captures events, routes them through rules and centralizes information so leadership can see what the business is actually doing. A CRM is a contact database with stages and notifications. An ITSM platform is a ticket system with SLAs and escalation trees.

None of this is autonomous driving. None of it is protein folding. It is, for the most part, business logic wrapped in a user interface and stitched together with integrations, exactly the category of software AI assisted development has become remarkably good at producing.

The moat was never the code. The moat was the cost of writing the code. That moat is draining faster than most vendor roadmaps are willing to admit.

The shift is no longer hypothetical. Retool’s 2026 Build vs. Buy Shift Report, drawn from a survey of 817 enterprise builders, found that 35% of teams have already replaced at least one SaaS tool with a custom internal build, and 78% plan to build more custom tooling in the year ahead. The categories leading this displacement, workflow automations and internal administrative tools, are exactly the workflow engine layer most enterprise platforms occupy.

Integration, historically the second pillar of SaaS defensibility is going the same way. Building and maintaining API connections, parsing third-party schemas, handling auth flows and data mapping: These were once multi-quarter projects requiring senior engineering effort. They are now, increasingly, prompts and pull requests.

The real reason enterprises haven’t moved yet

If the economics have shifted so dramatically, why hasn’t every Fortune 500 CIO kicked off a build-in-house initiative? The answer has almost nothing to do with technology.

Replacing a CRM or an ERP is not a software problem. It is a change management problem at a scale most outside observers drastically underestimate. You are not swapping out a tool; you are disrupting the daily workflows of tens of thousands of employees, retraining entire functions, reworking reporting structures and absorbing a period of real operational risk. The software itself is perhaps 30 percent of the effort. The other seventy percent is people, process, data migration and the organizational gravity every incumbent system accumulates over a decade of use.

This is why current vendor pricing has, so far, held. Not because the product is irreplaceable, but because the switching cost is measured in quarters of lost productivity, not in license fees.

The pricing conversation vendors are avoiding

Here is the dynamic. Enterprises will not replace incumbent systems out of enthusiasm for their own engineering. They will replace them when the spread between vendor cost and internal build cost becomes large enough to justify the organizational pain.

The pricing pressure on the customer side is already showing up in the data. Zylo’s 2026 SaaS Management Index shows enterprise SaaS spend now averaging $55.7 million annually, up 8% year-over-year, while application portfolios have remained essentially flat at around 305 applications. The increase is coming from pricing inflation, AI tiers, consumption charges and contract expansion, not from adding new tools. Customers are paying more for the same number of platforms, and they have noticed.

That threshold is not fixed. It moves with every improvement in AI-assisted development tooling, with every successful internal proof of concept, with every reference case from a peer who has done it and survived. If SaaS pricing does not meaningfully adjust, enterprises will eventually take the path of building their own, not because it is glamorous, but because at some point the math stops being close.

Vendors who respond early, by genuinely lowering total cost of ownership, by building platforms that are cheaper to customize than to rebuild, by shifting from seat-based rent extraction to something that reflects AI-era economics, will survive this transition in strong shape. Vendors who defend legacy margin structures will, in the medium term, find themselves defending empty castles.

Repricing is the easy part. Rewiring the company is the hard one.

I want to be precise here, because most vendor responses to this moment will get it wrong. The instinct is to run the discount lever, enterprise-wide deals, bundled SKUs, harder renewal negotiations. That is the obvious move. It is also insufficient.

The deeper shift is about the operating model of the SaaS company itself. For two decades, vendors were built around specific assumptions: Customers were captive, renewal was largely automatic, roadmaps could be dictated from the vendor side and the sales motion was about expansion rather than earned trust. Those assumptions produced the margins, and the cultural habits that defined the category.

Industry analysts are starting to document the early stages of this shift. L.E.K. Consulting’s analysis of how AI is changing SaaS pricing tracks the structural move away from seat-based and flat-fee models toward usage-based and hybrid structures, with specific examples like Atlassian’s recent cloud price changes and Microsoft phasing out volume-based enterprise discounts. The pricing models themselves are in flux, but pricing change alone is not the answer.

Rewiring means rebuilding the company around the assumption that the customer now has a credible alternative. It means product teams that treat customer workflow fluency as an engineering skill. It means pricing structures that share value rather than extract it. It means professional services organizations measured on reducing customer dependency rather than deepening it. And it means sales cultures that stop treating the CIO office as a target and start treating it as a partner whose internal build capability is now a legitimate competitive force.

A discount is a transaction. A win-win relationship is a rewiring. The vendors who survive this will be the ones who understand the difference.

A note to my fellow CEOs: If you have not yet thought of your CIO as a strategic voice in your leadership team, now is the time to reconsider. A good CIO today is not a technology manager. They are a business strategist with a technical fluency the rest of the C-suite, in most cases, does not have.

The exception: cybersecurity

I want to be careful not to overstate the thesis. One category of enterprise software does not follow this logic, and I believe the incumbent model will not only survive but strengthen: cybersecurity.

Palo Alto Networks, Zscaler, CrowdStrike and their peers sit in a fundamentally different position. The product is not really the software. The product is the threat intelligence, the accumulated, continuously updated understanding of adversaries, attack patterns and zero-day exposures, drawn from telemetry across thousands of customer environments that no single enterprise can replicate.

The investment patterns reflect this. Recorded Future’s research on enterprise threat intelligence indicates that 91% of organizations plan to increase threat intelligence spending in 2026, even as broader software budgets come under pressure. No enterprise wants to stand up its own cyber research function. No CIO wants the accountability of keeping pace with nation-state actors. The expertise is genuinely scarce, the stakes are existential and the failure mode, a breach, is career-ending in a way that a clunky CRM never is. That’s a real moat, and AI does not dissolve it. If anything, AI on the adversarial side only deepens the case for specialist providers.

But the exemption is not a license to stand still. If the rest of the enterprise stack moves toward internally built, AI-generated platforms, cybersecurity vendors must fundamentally rethink how and where they show up. The old model assumed a well-defined perimeter of vendor systems to protect. That perimeter is dissolving. What replaces it is a sprawl of internally built applications, AI agents and custom workflows, each one a new attack surface that wasn’t on any vendor’s roadmap six months before it went live.

The winning cybersecurity companies of the next cycle will not sell a product alongside the enterprise stack. They will embed inside it as security primitives in AI development tooling, as guardrails in agent frameworks, as policy layers in the internal platforms enterprises are building. Security must become a capability present at the moment of construction, not a control applied after deployment.

What this means for the next five years

I don’t believe the SaaS model is dying. I believe it is being repriced, and the repricing will be uneven, painful in some categories, barely felt in others. The vendors who recognize this early and move first will be the winners of the next cycle.

For CIOs, the honest advice is not to rush into replacing your ERP next quarter. The change management reality is still real. But it is worth building internal AI-assisted development capability now, starting with the edges, small workflow apps, agents, integrations, internal tooling where the risk is contained and the learning is transferable. Over time, that capability becomes optionality. And optionality, in a negotiation with a SaaS vendor, is worth more than any discount they will offer you unprompted.

The industry has spent a long time operating on the assumption that enterprise software is something you must rent. That assumption is now, for the first time in a generation, genuinely up for renegotiation. The companies that recognize this on both sides of the table are the ones that will define what comes next.

The rest will be renegotiated around.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?