35% of enterprises quietly ditched their SaaS tools for custom AI

Somewhere inside a Fortune 500 company right now, a team is canceling a six-figure SaaS contract and replacing it with an AI agent that costs less than a team lunch. They are not early adopters. They are not experimenting. According to a late-2025 survey of over 800 enterprise builders, 35% of companies have already replaced at least one paid SaaS tool with a custom AI build, and 78% plan to build more in 2026.

The number that should worry every software CEO sits in Gartner's 2026 Strategic Predictions: AI agents will trigger a $58 billion shake-up in the productivity software market by 2027. Not a gradual evolution. A structural collapse of a category that has dominated enterprise spending for 35 years.

Why the per-seat model is breaking

The economics are brutal. Traditional SaaS charges per user, per month. But when one employee armed with AI agents can do the work that previously required five people (each with their own software licenses), the math stops making sense. Bain & Company found that AI model costs are collapsing so fast that OpenAI's frontier reasoning model dropped 80% in price within two months. Every price drop widens the gap between what a SaaS subscription costs and what a custom AI replacement costs.

The companies moving fastest are not just saving money. Harvard Business Review reports that Hitachi Global achieved 70% operational efficiency gains by deploying AI across 120,000 employees in just eight weeks, replacing traditional HR shared services. Klarna, the Swedish fintech serving 85 million customers, announced plans to replace both Salesforce and Workday with internal AI systems.

The spending shift is already measurable

If you think this is theoretical, the financial data says otherwise. YipitData tracked spending patterns across dozens of companies and found that mid-market AI early adopters slashed their project management software allocation by roughly 50% year-over-year. Where did that money go? Core AI platform spending jumped more than 300%.

The pattern is clear: companies are not adding AI on top of their existing software stack. They are ripping out entire categories and rebuilding them. Project management tools, CRM dashboards, HR platforms: anything that follows predictable rules and processes is vulnerable.

The $913 billion question nobody is asking

Global enterprise software spending hit $913 billion in 2023, growing 12.4% year over year. That number is still climbing. But the composition is shifting beneath the surface. Gartner predicts 40% of enterprise applications will include task-specific AI agents by the end of 2026, up from less than 5% in 2025. That is not a trend. That is a phase transition.

Here is what most coverage misses: the disruption is not uniform. Enterprise companies that adopted AI early actually increased their project management spending, because AI amplified operational complexity at scale. The real casualties are mid-market SaaS tools: the project trackers, the form builders, the scheduling apps that charge per seat for functionality an AI agent can now replicate in an afternoon.

What happens to the remaining 65%

The companies still paying full price for traditional SaaS are not just behind. They are funding their competitors' advantage. Every month of delay widens the cost gap. The 35% who have already switched are reinvesting those savings into capabilities their competitors cannot match.

Bain's analysis maps the threat across two dimensions: how easily AI can automate user tasks, and how deeply AI can penetrate existing SaaS workflows. Where both scores are high, they use a stark label: "AI cannibalizes SaaS." Project management, expense reporting, applicant tracking, and basic CRM all fall into that quadrant.

The shift will not happen overnight. But Gartner's timeline is 2027, not 2035. That gives the average enterprise roughly 18 months to decide whether they are building the replacement or becoming the thing that gets replaced.

For a look at which companies are actually profiting from AI adoption (and the 94% that are not), see our breakdown of why only 6% of AI-using companies see real returns.