The SaaS Reckoning
A healthcare company called Curative just canceled a $600,000-per-year Salesforce contract. They built their own CRM in about two months. That's the headline. The story underneath is bigger.
Curative's CEO, Fred Turner, says the company plans to reduce overall SaaS spending by as much as 80% during 2026 and redirect some of that capital toward AI. They've also built an internal contract-negotiation agent that reduced the processing cost of certain contracts from roughly $1,500–$2,000 down to about $70. If you've been watching enterprise software economics, this isn't surprising. It's the first clean case study of something that's about to become a pattern.
Here's the obvious interpretation: AI gave Curative a new build-versus-buy option. They could build instead of buy, so they did. That's not wrong, but it's too narrow. Companies have always been able to build software. The historical problem was that custom software required a development team, a long requirements process, months or years of implementation, and permanent technical overhead. Salesforce could charge $600,000 because the alternative might've cost $3 million and taken eighteen months.
AI doesn't just give you a new option. It changes the denominator.
The comparison used to be: pay $600,000 for Salesforce, or spend $3 million and eighteen months building something that does the same thing. That wasn't a real choice. Now it's becoming: renew the $600,000 contract, accept its licensing restrictions, and keep adapting your business to its architecture — or spend $150,000 building exactly the workflow you need and maybe $100,000 annually maintaining it.
That's an entirely different procurement decision. And it doesn't just apply to Salesforce.
The deeper change is that enterprise software is being decomposed from products back into workflows. Executives are going to stop asking, "Which CRM should we buy?" and start asking: What work does our CRM actually perform, which portions of that work remain valuable, and which can now be reproduced more cheaply?
That question applies to CRMs. It applies to ERPs. HR systems. Project management platforms. Contract tools. Compliance systems. Analytics packages. Support desks. Document management. A thousand specialized vertical SaaS products that exist because building a custom version used to be prohibitively expensive.
The math is shifting. When the cost of building a custom workflow drops by an order of magnitude, every SaaS product becomes a decomposition target. Not because the software is bad. Because the premium you're paying for "someone else built and maintains this" is collapsing toward the cost of building and maintaining it yourself.
That's the reckoning.
Now the caveat, because caveats matter — especially if you've spent years inside enterprise integrations. The visible application is rarely the whole system. The dangerous parts of enterprise software aren't the dashboards and forms. They're identity, permissions, auditability, data migration, regulatory controls, reporting, integrations with other systems, institutional knowledge encoded in configuration, and the ongoing maintenance of all of it.
Current research benchmarks still show AI agents struggling with complex professional CRM tasks, particularly when they must follow rules reliably and manipulate interconnected records. A model that can build a beautiful contact management interface in an afternoon is not the same thing as a system that can safely manage permissions across forty thousand records, maintain audit trails for compliance, integrate with your billing platform, and not lose data when someone updates a workflow at 2 AM.
So Fortune 500 companies probably won't "vibe-code Salesforce" out of existence. Not this year. Maybe not next year. But they'll start hollowing it out.
Keep Salesforce temporarily as the system of record. Move the user experience and workflow automation into custom AI applications. Reduce licenses. Eliminate expensive modules nobody actually uses. Gradually migrate the underlying data and business logic. Retire the platform once the dependency graph is manageable.
That's where the actual disruption occurs. Salesforce doesn't disappear suddenly. It becomes an increasingly expensive database sitting underneath interfaces and agents that companies own themselves. The license doesn't get renewed because nobody's logging in anymore — they're interacting with the data through their own layer. And once that pattern starts, it accelerates. Every successful hollow-out becomes proof of concept for the next one. The CIO who saved $600,000 on Salesforce starts looking at Workday the same way.
Here's the second-order effect that most people aren't talking about yet.
The SaaS industry has spent fifteen years building a specific business model: identify a workflow, productize it, sell it as a subscription, and lock in customers through switching costs and data lock-in. That model depends on the build-vs-buy gap being wide enough that buying is obviously cheaper than building.
AI narrows that gap. Not to zero. But enough that the question changes.
When the question changes, the pricing power changes. And when the pricing power changes, the valuation multiples change. SaaS companies trade at premium multiples because investors believe recurring revenue is sticky. If recurring revenue becomes negotiable — because customers have a credible alternative — the entire valuation framework gets re-priced.
That doesn't mean SaaS collapses. The best products will adapt. They'll absorb AI into their platforms, reduce pricing, move to usage-based models, or build such deep integration that replacing them still isn't worth the hassle. Salesforce isn't stupid. They'll fight back. But the negotiating position shifts. For the first time in a decade, the buyer has leverage. Not because of a better competitor. Because the alternative to buying isn't what it used to be.
There's a third order too, and it's the one I keep coming back to. The entire category of "software implementation consulting" — the firms that help you configure Salesforce, integrate Workday, customize ServiceNow — that category exists because enterprise software products are general-purpose platforms that need to be shaped to fit specific businesses. The implementation partner is the translation layer between what the product does and what the company actually needs. If AI lets you build exactly the workflow you need, you don't need someone to configure a general-purpose platform to approximate it. You need someone who understands the workflow, understands AI, and can build the specific thing.
The implementation partner doesn't disappear. It evolves. The ones who survive will be the ones who stop selling "we'll configure your SaaS" and start selling "we'll examine your software spend, identify what you don't need anymore, and build replacements that cost less and do more." That's not a services business. That's a displacement business. And the first sale is the easiest: walk into a company with a software-spend inventory, identify five contracts totaling $2 million annually, determine which workflows are realistically recoverable, and propose a staged replacement program with measurable savings.
The pitch becomes: don't begin your AI transformation by purchasing more software. Begin by examining the software you may no longer need. That's a boardroom-level offering. Not a "we build websites and bots" offering. A "we reduce your annual software spend by 40-80% and you own the result" offering.
Curative canceled one contract and built something in two months. That's a data point, not a trend. But it's the kind of data point that starts things. Every CIO who reads that story will think the same thing: if they can do it, why can't we? And for the first time, the answer isn't "because you can't afford to build it." The answer is "you probably can."
The question isn't whether this happens. It's how fast, and who captures the value on the way down.
The SaaS industry has been building moats for fifteen years. AI is filling them in. Not all at once. But fast enough that the companies standing behind those moats should probably start thinking about what they're worth without them.
The real beginning of the SaaS reckoning isn't when one company cancels a contract. It's when every company with a $500K+ software bill starts running the same math and reaches the same answer. We're there.
Rich Washburn is a technologist and strategist working at the intersection of AI, infrastructure, and capital. He is Managing Partner and Chief AI Officer at Eliakim Capital.
Comments