72% of CIOs Are Losing Money on AI. Here's What Actually Works in IT.

In partnership with

Let's talk about the elephant in the server room.

Your CTO wants AI everywhere. Your vendor is slapping "AI-powered" on everything from ticketing to toilet paper procurement. And you — the person who actually keeps the lights on — are watching 72% of organizations lose money on their AI investments.

That's not a typo. A Gartner survey of 506 CIOs found that nearly three-quarters are breaking even or underwater on AI spending. And it gets worse from there.

The Numbers Nobody Wants to Talk About

Here's the reality check your leadership team isn't getting from their LinkedIn feed:

• 95% of enterprise AI pilots deliver zero measurable business impact (MIT)

• 70-85% of generative AI deployments are failing to meet desired ROI (NTT DATA)

• 30% of sysadmins report AI troubleshooting failures — nearly double from last year

• 42% of companies have already scrapped most of their AI projects

The most damning stat? AI troubleshooting — the use case vendors pitch hardest to IT teams — is where AI fails the most. 41% of sysadmins use AI for troubleshooting, but nearly a third say it makes things worse, not better. It hallucinates root causes. It confidently recommends the wrong fix. It creates new incidents and then hands them to you.

The Copilot Problem

Microsoft just paused its forced auto-deployment of Copilot after admin outcry. And the complaints aren't about the concept — they're about the execution.

Nearly 70% of security teams worry Copilot could expose sensitive data. Over 15% of business-critical files are at risk due to oversharing permissions that Copilot happily surfaces. The governance nightmare is real: deployments stall because identity protections are weak, licensing is a mess, and nobody addressed data governance before flipping the switch.

Meanwhile, each Copilot iteration lives on a separate platform with no interoperability. It's not one product — it's a dozen products wearing a trench coat pretending to be one product.

Agent Washing: The New AI Snake Oil

Remember "cloud washing" a decade ago? Meet its successor: agent washing. Vendors are marketing rule-based workflows and chatbots with basic prompt engineering as "autonomous AI agents." The SEC has already issued enforcement actions for AI misrepresentation. The FTC launched "Operation AI Comply" to crack down on overblown claims.

Gartner predicts 40% of "agentic AI" projects will be canceled by 2027 due to high costs, unclear value, and weak risk controls. If your vendor can't explain exactly what their AI does differently from a well-tuned automation script, you're probably looking at a rebrand, not a revolution.

88% resolved. 22% stayed loyal. What went wrong?

That's the AI paradox hiding in your CX stack. Tickets close. Customers leave. And most teams don't see it coming because they're measuring the wrong things.

Efficiency metrics look great on paper. Handle time down. Containment rate up. But customer loyalty? That's a different story — and it's one your current dashboards probably aren't telling you.

Gladly's 2026 Customer Expectations Report surveyed thousands of real consumers to find out exactly where AI-powered service breaks trust, and what separates the platforms that drive retention from the ones that quietly erode it.

If you're architecting the CX stack, this is the data you need to build it right. Not just fast. Not just cheap. Built to last.

See the data

OK, So What Actually Works?

Here's the thing — AI does work in IT. Just not the way the marketing slides promise. The wins are narrow, boring, and measurable. Which is exactly why they work.

Ticket routing and triage: Organizations using AI-driven ticket classification are resolving up to 80% of routine requests automatically and cutting mean resolution time by 70%. This isn't magic — it's pattern matching on historical data, and it's been quietly excellent for two years.

Alert deduplication and noise reduction: AI filtering that eliminates false positive alerts is reducing alert fatigue dramatically. When your monitoring tool fires 200 alerts and AI collapses them into the 3 that actually matter, that's real value.

Predictive maintenance: Tools like Siemens Senseye and C3 AI Reliability are cutting downtime by up to 50% using IoT sensor data. These work because they're narrowly scoped, data-driven, and don't try to be a chatbot.

The pattern is clear: AI that automates narrow, well-defined tasks with clean data succeeds. AI that tries to "think" like a sysadmin fails.

What to Tell Your Boss

Next time leadership asks why you're not "leveraging AI more," here's your framework:

• Buy, don't build. Organizations that purchase pre-built AI tools achieve 67% success rates vs. 22% for teams that build internally. A 3:1 advantage.

• Start with routing and triage. Not chatbots. Not "AI assistants." Ticket classification and alert deduplication have the highest proven ROI.

• Fix your data first. 60% of AI projects fail because the organization's data isn't AI-ready. Clean data beats fancy models every time.

• Demand specifics from vendors. "What exactly does the AI do? What data does it train on? What's the measured accuracy?" If they can't answer, walk.

The sysadmins who are winning with AI aren't the ones chasing the hype. They're the ones deploying boring automation on clean data with clear success metrics. It's not exciting. It works.

Don't let the AI hype cycle make you feel behind. You're not behind — most of the people ahead of you are lost.

Stay skeptical. Stay patched. See you Tomorrow.