The $55.7 Million Threshold: When AI Costs Drive Enterprise Build Decisions

The New Economics of Software Acquisition

Annual SaaS spend now averages $55.7 million while application portfolios have remained essentially flat at around 305 applications. The increase stems not from adding new tools, but from pricing inflation, AI tiers, consumption charges and contract expansion: enterprises pay more for the same number of platforms.

Thirty-five percent 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) occupy exactly the workflow engine layer most enterprise platforms depend upon.

What Clayton Christensen called the "performance trajectory" has inverted. Two years ago, a custom internal tool might have taken an engineering team weeks or months and cost six figures. Today, an operations lead with the right platform can have a working prototype in a day or two.

The Mathematical Shift

The threshold is not fixed: it moves with every improvement in AI-assisted development tooling. Enterprises will replace incumbent systems when the spread between vendor cost and internal build cost becomes large enough to justify the organisational pain.

SaaS pricing hasn't adjusted, still charging per-seat for generic software that requires customisation and integration costs on top. OpenAI's CEO predicted "AI prices will drop 10x annually," but enterprise vendors are doing the opposite: Microsoft added Copilot to Microsoft 365 and raised subscription prices by $3 per month.

Consider the unit economics. Microsoft Copilot is priced at $30 per user per month, but only if you already have a Microsoft 365 licence. That makes the actual cost significantly higher.

Organisations spent an average of $1.2 million on AI-native apps, a 108% year-over-year increase. This is not sustainable mathematics.

The Shadow Infrastructure

Sixty percent of builders have created tools, workflows, or automations outside of IT oversight in the past year, and 25% report doing so frequently. Shadow IT at this scale is a demand signal.

The people closest to the problems are circumventing procurement. Thirty-one percent of those going around IT do so simply because they can build faster than IT can provision tools.

But ungoverned building invites security risks. Data privacy and security ranked as the top AI concern at 73%, with governance capabilities close behind at 46%.

When Buying Still Makes Sense

Recorded Future's research indicates that 91% of organisations plan to increase threat intelligence spending in 2026, even as broader software budgets come under pressure.

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.

Cybersecurity represents the category where specialised knowledge compounds faster than internal capabilities can develop.

The Hybrid Architecture

The future isn't all incumbents or all AI-native startups, it's a blended ecosystem of agent-driven platforms, new pricing structures, and AI governance layers.

The enterprise market has shifted toward buying foundational AI capabilities while reserving custom development for systems that differentiate the business. In practice, many organisations combine these approaches, purchasing commodity AI services while building custom software that integrates those capabilities.

This represents what Ben Thompson would call the "integration point shift." Value creation moves from the application layer to the orchestration layer.

The Strategic Framework

The correct question: Is this AI capability part of how you compete, or just how you operate? If it's "how we compete," you need to own it.

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.

Record labels in 2003 also believed their distribution model was defensible. The companies that recognise this structural shift (on both sides of the table) are the ones that will define what comes next.