35% of Companies Already Replaced a SaaS Tool with Custom Software -- Should You Be Next?

One in three companies has already pulled the plug on a SaaS subscription and replaced it with something they built themselves. That number is not a projection. It comes from Retool's 2026 Build vs. Buy report, which surveyed 817 software builders and business teams in late 2025. The question for most businesses is no longer whether to consider custom software. It is whether they are already falling behind.

The Numbers Behind the Shift to Custom AI Development

The Retool data tells a clear story. 35% of teams have already replaced at least one SaaS tool with a custom-built alternative. 78% plan to build more internal tools in 2026. And 51% of those builders are using AI-assisted development — what some are calling "vibe coding" — to get there faster than anyone expected.

The speed gains are real. GitHub Copilot data shows developers complete tasks 55% faster with AI assistance. Pull request cycle times dropped from 9.6 days to 2.4 days at high-adoption organizations. JPMorgan Chase reported a 30% improvement in developer velocity after deploying AI coding tools broadly.

What this means practically: software that would have taken months and cost six figures now takes weeks and costs a fraction of that. The barrier to building has collapsed.

Why SaaS Vendors Are Quietly Panicking

On February 3, 2026, Anthropic launched Claude Cowork. By end of day, the software industry had lost approximately $285 billion in market capitalization. By mid-March, total losses in enterprise software stocks reached roughly $2 trillion. The iShares Software ETF dropped more than 21% year-to-date.

Atlassian cut 1,600 jobs — 10% of its entire workforce. Workday slashed 8.5% of its staff. Salesforce, HubSpot, and Adobe all saw significant declines. Goldman Sachs analysts warned of a "structural decline comparable to the newspaper industry collapse."

This is not a market correction. It is a structural shift. Analysts are calling it the SaaSpocalypse. The core problem for legacy vendors is a concept called "seat compression": if an AI agent can do the work of five employees, a company that once needed 50 licenses now needs 10. Subscription revenue assumptions built over two decades are collapsing.

40% of IT budgets are being reallocated from traditional SaaS platforms to AI-native or owned solutions. That reallocation is accelerating.

The Real Cost of Staying on Subscriptions

Average SaaS spend per employee reached $4,830 in 2025 — up 21.9% year-over-year. Mid-market companies saw SaaS costs climb roughly 40%. Most of that spend goes toward tools that do exactly what AI automation handles best: project management, CRM workflows, simple automations, reporting.

The problem is not just the monthly line item. It is what you cannot do.

Deloitte found that 41% of companies cited lack of flexibility or customization as the primary reason they left a vendor AI solution. SaaS tools are built for the median customer. Your workflows, your data structures, your edge cases — they are afterthoughts. You get what the product roadmap gives you, on the vendor's timeline, at the vendor's price.

And then there is lock-in. As organizations accumulate data in proprietary formats, migration costs grow exponentially. The EU Data Act, effective September 2025, introduced new data portability requirements precisely because this problem had become so acute. High switching costs — from training, customization, and integration investments — keep companies paying even when a tool has stopped delivering value.

What "Build vs. Buy" Actually Means in 2026

McKinsey frames the decision across three dimensions: strategic importance, capability availability, and economic efficiency. That framework still holds. But the economic efficiency calculation has changed dramatically.

In 2023, building a custom CRM integration might have cost $80,000 and taken three months. Today, the same functional outcome is achievable in two to four weeks at a fraction of the cost. The math that once made buying the obvious answer no longer applies uniformly.

TechCrunch put it directly in March 2026: barriers to entry for creating software are now so low that build-versus-buy is shifting structurally toward build.

The most vulnerable SaaS categories are also the most common ones businesses pay for:

• Project management tools — core workflows that AI handles precisely
• CRM platforms — relationship and pipeline logic that custom builds replicate well
• Single-purpose automation tools — the easiest to replace with purpose-built alternatives
• Internal reporting and dashboard software — high cost, low differentiation

More resilient: regulatory and compliance software, and platforms with deep proprietary data moats. But even in those categories, the analysis is worth running.

Ownership Changes How You Build and What You Keep

Companies that have moved to owned or composable solutions report 80% faster feature deployment. That is not surprising. When you own the code, you change it when you need to. You do not file a support ticket and wait for a roadmap update.

Ownership also means your software reflects your actual business logic — not a generalized approximation of it. Your customer segmentation, your approval workflows, your pricing rules. Built once, working exactly as you need, without a monthly fee for the privilege.

Data sensitivity is another factor that is pushing more businesses toward building, particularly in healthcare, finance, and professional services. When sensitive data runs through a third-party SaaS platform, you are dependent on their security posture, their compliance certifications, and their continued viability as a business. Owning the software means owning the data environment.

What This Means for Your Business Right Now

The companies moving first are not all large enterprises with dedicated engineering teams. 58% of small businesses are now using generative AI, up from 40% in 2024. 91% of SMBs using AI report revenue increases. The tools and the economics are available to businesses of every size.

The practical question is not whether you could replace a SaaS tool. It is which one to start with, and what approach makes sense given your timeline and budget.

A few honest starting points:

• List every SaaS subscription over $200/month
• Flag any where your team regularly works around limitations
• Identify tools where your data is locked into a proprietary format
• Estimate what two years of subscription fees adds up to

That exercise usually surfaces one or two tools where the build case is obvious.

Build vs. Buy vs. Own

The frame has shifted. It is no longer just build versus buy. The real question is whether you want to own your software or rent it indefinitely.

Custom AI applications now start at $5,000 with two-to-four week delivery timelines. Pre-built AI apps purpose-built for common business workflows are available from $500, with full source code included. No subscriptions. No vendor roadmaps. No seat compression working against you.

The 35% of companies that have already replaced a SaaS tool are not early adopters chasing novelty. They ran the numbers. The numbers worked. Most of them wish they had done it sooner.

If you want to see what is available off the shelf, or talk through a custom build for your specific workflow, Intraverse AI builds owned AI solutions for businesses that are done paying for software they do not control.