Here’s the most expensive contradiction in enterprise technology: 88% of organisations now use AI in at least one business function. But only 6% are seeing meaningful bottom-line impact.

That’s not a rounding error. That’s a structural failure.

McKinsey’s 2025 State of AI survey surveyed nearly 2,000 participants across 105 countries. The findings are stark: despite near-universal adoption, just 39% report any EBIT impact at the enterprise level. Two-thirds of companies remain stuck in piloting or experimenting phases. And the gap between “using AI” and “transforming with AI” is growing wider, not narrower.

Everyone bought the tools. Almost no one changed the work.

The Pilot Purgatory Epidemic

The pattern is predictable: a promising pilot impresses leadership, gets budget, then slowly dies in staging. MIT’s research found that roughly 95% of enterprise generative AI pilots fail to deliver measurable financial returns. S&P Global data shows 42% of companies scrapped most of their AI initiatives in 2025—more than double the rate from just a year earlier.

The average enterprise abandoned 46% of AI proofs-of-concept before they ever reached production.

Why? Because pilots are designed to prove technology works. They’re not designed to prove the organisation can work differently.

What Actually Separates the 6%

Organisations seeing the most value set growth or innovation as additional objectives. Cost-cutting alone doesn’t create transformation.

McKinsey’s data reveals something crucial: the high performers aren’t just buying better tools. They’re fundamentally rewiring how work gets done.

The differentiators:

Workflow redesign, not tool deployment. High performers are nearly three times as likely to have fundamentally redesigned individual workflows. This factor has one of the strongest contributions to achieving meaningful business impact of all the variables McKinsey tested.

55% of high performers redesigned workflows around AI. Only 20% of everyone else did.

Growth and innovation, not just efficiency. 80% of companies set efficiency as an AI objective. But organisations seeing the most value set growth or innovation as additional objectives. Cost-cutting alone doesn’t create transformation.

Executive ownership, not delegation. High performers’ AI initiatives are 3x more likely to be championed by senior leadership. When AI stays in the hands of specialists and innovation labs, it stays in pilot purgatory.

Scale across functions, not isolated wins. High performers regularly use AI in more business functions than their peers—marketing, strategy, product development, not just IT.

The Commoditisation Trap

You can’t afford to hand that to a black-box SaaS vendor. Your code, your architecture decisions, your business logic, your customer data—all flowing through someone else’s infrastructure, training someone else’s models, locked into someone else’s ecosystem.

Here’s where most enterprises go wrong: they treat all AI adoption the same way.

For general productivity—summarising documents, drafting emails, answering questions—commoditised tools like ChatGPT and Copilot work fine. Buy the subscription. Roll it out. Move on.

But workflow-critical systems are a different beast entirely.

When AI touches your core development processes, your proprietary data, your competitive advantage—you can’t afford to hand that to a black-box SaaS vendor. Your code, your architecture decisions, your business logic, your customer data—all flowing through someone else’s infrastructure, training someone else’s models, locked into someone else’s ecosystem.

The 94% stuck in pilot purgatory made a category error. They treated transformational AI like commodity AI: buy the tool, plug it in, hope for results.

The 6% understood the distinction. For commodity tasks, they bought. For workflow-critical systems, they invested in platforms that keep control where it belongs—inside the enterprise.

Why Control Matters More Than Convenience

The vendors selling “enterprise AI” want you to believe transformation is a subscription away. Upload your data. Trust our models. Pay monthly.

But here’s what that bargain actually costs:

Your data becomes their asset. Every prompt, every codebase, every architectural decision flows through infrastructure you don’t control. Today it’s “just” training their models. Tomorrow it’s competitive intelligence for their next customer—who might be your competitor.

Your workflows become their moat. The deeper you integrate, the harder it is to leave. That’s not a bug; it’s their business model. Switching costs are features, not accidents.

Your differentiation becomes their commodity. When every company in your industry uses the same AI tools the same way, where’s the advantage? You’ve optimised yourself into parity.

Your visibility becomes their black box. When something goes wrong—and it will—can you trace exactly what happened? Can you audit the decisions? Can you explain to regulators why your AI did what it did? Not if the logic lives in someone else’s cloud.

The enterprises capturing real value from AI aren’t the ones outsourcing transformation to vendors. They’re the ones investing in platforms that enable transformation while preserving what matters: control over data, visibility into decisions, ownership of code, and the ability to build genuine competitive advantage.

The Platform Imperative

This is why the conversation is shifting from “which AI tools should we buy?” to “what AI platform should we build on?”

The distinction matters:

Tools solve point problems. Code completion here. Test generation there. PR review somewhere else. Each tool is a separate vendor, a separate integration, a separate data silo, a separate point of lock-in.

Platforms solve system problems. How does AI flow through the entire development lifecycle? How do decisions in requirements inform architecture inform code inform testing inform deployment? How does context persist across the full journey from intent to production?

The 6% aren’t assembling Frankenstein stacks of disconnected AI tools. They’re investing in platforms that redesign workflows end-to-end—while keeping code visible, data controlled, and lock-in minimal.

That’s a harder path than buying subscriptions. It requires genuine investment. But it’s the only path that creates durable advantage rather than temporary efficiency.

What This Means for 2026

The hype is fading. The disillusionment is setting in. And that’s actually good news.

Because now the conversation can shift from “are you using AI?” to “are you building AI capabilities you actually own?”

The enterprises that will dominate the next decade aren’t the ones with the most AI subscriptions. They’re the ones building proprietary AI-native workflows on platforms they control—where the data stays theirs, the code stays visible, and the competitive advantage stays internal.

Everyone else is funding someone else’s moat.

The Bottom Line

The 6% problem isn’t a technology problem. It’s a control problem disguised as a technology problem.

Commoditised AI tools work for commodity tasks. But transforming how your organisation builds software? That requires platforms that give you the transformation and the ownership—full visibility into every decision, every line of code, every architectural choice.

The path forward isn’t more SaaS subscriptions. It’s investment in platforms that redesign workflows from the ground up while keeping control where it belongs: with you.

That’s what we’re building.

Ready to transform your development workflow without surrendering control?

Sign up to see how Ardor delivers full-lifecycle AI with complete code ownership, zero lock-in, and your data staying yours.