Leadership In Infrastructure

Over 20 Years in Infrastructure Taught Me About Leadership

There’s a point in every infrastructure project where the plan starts to drift.

Not fail.
Not collapse.
Just drift.

Teams are ready but materials aren’t there.
Work orders are issued but can’t be completed.
KPIs are set but the system can’t support them.

That’s where leadership actually shows up.

Not in strategy.

In execution.

And this isn’t just experience talking.

Large infrastructure projects are consistently delivered late and over budget, not because the strategy was wrong, but because execution breaks under real world conditions (McKinsey & Company, The Art of Project Leadership).

What people think leadership is

Clear direction.
Strong communication.
Alignment.

All important.

But none of it holds if the system underneath can’t deliver.

Because infrastructure doesn’t fail on paper.

It fails when:

 And globally, productivity across infrastructure sectors has remained flat for decades, largely due to fragmented systems, poor coordination, and inconsistent delivery models (McKinsey & Company, Reinventing Construction).

What’s actually happening

Most environments don’t collapse overnight.

They degrade.

You start to see:

 I’ve walked into environments running at 30% KPI.

Not because people didn’t care.

Because:

And the response?

Push harder.

That’s where it gets worse.

Because when systems are inefficient, pressure amplifies failure rather than fixing it, a pattern widely observed in large scale operational environments (McKinsey & Company, Infrastructure Productivity Insights).

Where it breaks

It almost always breaks in the same place:

Planning - Field

That gap is where:

 At scale, this doesn’t stay contained.

Small delays cascade across systems.

Research into complex infrastructure networks shows that interdependencies cause failures to propagate, what starts as a minor issue quickly becomes systemic (Santolini et al., Network Dynamics in Project Systems).

The part most leaders get wrong

There’s constant pressure to start early.

“Just get moving.”

But starting before:

…doesn’t accelerate delivery.

It guarantees rework.

And rework is where cost hides.

Decades of megaproject research show that early stage planning assumptions and optimism bias are among the primary drivers of cost overruns and delays (Flyvbjerg, Megaprojects and Risk).

Operator insights

After 20 years, a few things don’t change.

1. You don’t fix performance, you fix systems

If one person struggles, it might be them.

If a team struggles, it’s the system.

Blame doesn’t fix anything.

System design does.

Every major turnaround I’ve seen came from:

 Because system inefficiencies, not individual performance, are the dominant cause of delivery failure in large scale operations (McKinsey & Company).

2. Bad data breaks everything

This is one of the biggest hidden issues in infrastructure.

Poor data leads to:

 You can’t outwork bad inputs, yet it’s one of the least controlled parts of most delivery systems.

In both infrastructure and AI environments, poor data quality is consistently identified as a primary driver of project failure (Earley Information Science, Why Enterprise AI Fails).

3. Scale exposes what you didn’t fix

At low volume, problems hide. At scale, they show up fast.

In one case, workforce held strong.

What broke instead:

Demand outpaced system readiness.

This reflects a broader pattern, systems that appear stable at low volume often fail when scaled due to hidden constraints in supply chains and infrastructure capacity (McKinsey & Company, The Infrastructure Moment).

4. Workforce is the real constraint

This is already happening globally.

More than half of infrastructure projects are being delayed by workforce shortages (McKinsey & Company, The Infrastructure Moment).

And yet planning still assumes labour is scalable.

It’s not.

It can take around two years to make someone effective in these environments.

Not trained.

Effective.

Training pipelines and workforce development timelines are consistently identified as one of the hardest constraints to accelerate in infrastructure delivery (McKinsey & Company).

5. Cost doesn’t come from spend, it comes from flow

Most cost-cutting initiatives fail because they target the wrong thing.

Reducing teams creates:

One of the biggest improvements I’ve seen came from adding support roles.

Because production teams were freed up to produce.

Research consistently shows that productivity improvements, not cost cutting, are the primary drivers of sustainable cost reduction in infrastructure (McKinsey & Company).

What this means for AI infrastructure

We’re seeing the same pattern again.

Strong strategy.
Aggressive timelines.
High expectations.

But the same constraints are there:

And here’s the critical part:

Most AI projects don’t fail in development.

They fail when they try to scale.

The model works.

The system doesn’t.

Studies estimate that the majority of AI initiatives fail to reach production scale due to operational and infrastructure limitations, not technical capability (Earley Information Science; Gartner).

What happens next

As AI infrastructure expands, the first failures won’t be technical.

They’ll be operational:

The organisations that succeed won’t be the ones with the best strategy.

They’ll be the ones that:

Closing

Most people think leadership is control.

It’s not.

It’s support under pressure.

It’s building systems that hold when everything starts to drift.

Because at scale, the problem is never the size of the project.

It’s whether the system behind it was built properly in the first place.

References

Footnote

This article is part of a series exploring the physical infrastructure behind the AI economy.

Disclaimer

The views expressed in this article are my own and are intended for general information and discussion purposes only. They do not represent the views of any employer, organisation, or client.