Why Most AI Projects Will Fail — and How to Spot the Companies That Won't
For the past two years, saying "we're investing in AI" has been enough.
Boards applauded. Investors rewarded it. Press releases multiplied. Every software vendor suddenly became an AI company.
Now the mood is changing.
The question is no longer:
"What's your AI strategy?"
It's becoming:
"What did it actually deliver?"
That shift changes everything.
Because the organisations that survive the coming AI reckoning won't necessarily have the smartest models or the largest GPU clusters.
They'll simply be the ones that can prove a return on investment.
And that distinction matters.
AI has entered its ROI phase
Every major technology follows a familiar pattern.
First comes excitement.
Then experimentation.
Eventually comes accountability.
AI is now entering that third phase.
Across boardrooms, CIOs are no longer being asked whether they have an AI strategy. They're being asked whether the millions invested have actually produced measurable business outcomes.
That's a very different conversation.
For the last two years, almost any AI initiative could be justified under the banner of innovation.
Today, that same project has to justify its existence.
That's healthy.
Technology has never been the goal.
Business outcomes always were.
The easiest projects to start are often the easiest to abandon
One of AI's greatest strengths is also becoming one of its biggest weaknesses.
It's never been easier to build something.
A few prompts.
An API.
Some company data.
Within an afternoon you've got a working prototype.
The danger is that prototypes create momentum without creating value.
Teams become fascinated by what AI can do before asking whether anyone actually needs it.
Business requirements become optional.
Success criteria become vague.
ROI becomes something you'll "measure later."
Eventually, later arrives.
And nobody can explain why the project exists anymore.
That's why I believe many AI projects won't fail because the technology failed.
They'll fail because nobody defined success before writing the first prompt.
AI doesn't replace strategy
Good architecture still matters.
Good governance still matters.
Good programme management still matters.
Good data matters more than ever.
AI hasn't replaced any of those disciplines.
If anything, it's made them more important.
Organisations rushing headlong into AI often assume the technology compensates for poor foundations.
It doesn't.
It amplifies them.
Poor processes become automated poor processes.
Poor data becomes automated poor decisions.
Poor governance becomes automated risk.
AI simply accelerates whatever already exists inside your organisation.
Trust remains the hidden dependency
I've seen enough business intelligence projects over the years to know one universal truth.
People don't use systems they don't trust.
It only takes one meeting.
Someone points at a dashboard.
"Those numbers are wrong."
Immediately confidence evaporates.
It doesn't matter whether the numbers were actually wrong.
Trust has already been lost.
AI raises the stakes considerably.
Now we're no longer asking employees to trust reports.
We're asking them to trust recommendations.
Decisions.
Automation.
Perhaps even work that was previously performed by another human being.
Without trust, adoption stops.
Without adoption, ROI disappears.
Technology doesn't fail.
Confidence does.
Data is becoming the competitive advantage
Many organisations still think AI is the product.
It isn't.
Data is.
The AI models available today are increasingly becoming commodities.
Your competitors can access many of the same models you can.
What they can't easily copy is your organisational knowledge.
Your customer history.
Your operational experience.
Your engineering expertise.
Your manufacturing processes.
Your service records.
Your intellectual property.
That's where the real differentiation lives.
The organisations quietly investing in data quality, integration, governance and accessibility are building something far more valuable than another chatbot.
They're building an AI-ready enterprise.
Infrastructure is suddenly interesting again
For years, infrastructure has been viewed as the boring part of technology.
Nobody writes headlines about integration platforms.
Or identity.
Or governance.
Or metadata.
Or master data management.
But AI changes that.
The organisations enabling trusted, secure, governed data movement may end up creating significantly more value than those producing flashy demonstrations.
It's difficult to build intelligent systems on chaotic foundations.
Eventually, every successful AI initiative becomes an infrastructure story.
Don't measure AI by the demo
One habit worries me.
We're becoming impressed by demonstrations.
A polished chatbot.
A document summariser.
An AI-generated presentation.
They're fascinating.
But demonstrations aren't businesses.
The questions I would ask are much simpler.
- Does this solve an expensive business problem?
- Is someone using it every day?
- Can you measure time saved?
- Can you measure revenue created?
- Can you measure risk reduced?
- Would anyone notice if you turned it off tomorrow?
If the answer to that last question is "not really", then it probably isn't creating much value.
The companies that will win
The AI winners won't necessarily be those shouting the loudest.
They'll probably look surprisingly ordinary.
They'll have cleaner data.
Better governance.
Stronger integration.
Clear ownership.
Defined business outcomes.
And relentless measurement.
Their AI projects won't begin with:
"Let's build an AI assistant."
They'll begin with:
"Here's an expensive problem. Can AI solve it?"
That's a subtle difference.
But it's the difference between innovation theatre and business transformation.
A reckoning is coming
The grace period won't last forever.
Boards will increasingly ask uncomfortable questions.
Budgets will tighten.
Projects without measurable outcomes will quietly disappear.
That's not evidence AI has failed.
It's evidence organisations are finally treating AI like every other business investment.
Which is exactly where it belongs.
The irony is that the companies most likely to succeed won't necessarily be the ones talking about AI the most.
They'll be the ones still talking about data.
About architecture.
About governance.
About trust.
Those topics may not generate headlines.
But they generate something far more valuable.
Return on investment.
Final thought
The first wave of AI rewarded organisations simply for showing up.
The next wave will reward organisations that can prove value.
As CIOs, we shouldn't ask which company has the most AI.
We should ask a far better question:
Which organisations have built the foundations that allow AI to deliver measurable business outcomes, year after year?
Because in five years' time, nobody will care how many AI pilots a company launched.
They'll care how many are still running.
