The Hidden AI Tax - Xist4

June 25, 2026

The Hidden AI Tax

The AI Tax You Didn’t See Coming

Last week I spoke to a CTO who thought his AI spend was under control. Then his finance director tapped him on the shoulder and asked a simple question: 'Do we know how much we spent on AI coding this quarter?'

Silence. Awkward laughter. Then a look of pure fear.

According to TechRadar’s reporting on senior industry analysts, AI coding costs are growing so quickly that they’re expected to exceed developer salaries by 2028. That’s right. Your AI agent may soon be the most expensive member of your engineering team.

And the scariest part? Most companies have no clue how much they’re actually spending. Token usage is the new shadow IT.

The Invisible Money Pit

AI feels cheap. It feels instant. It feels like magic. But under the hood, every request, refactor, code review, and speculative prompt is quietly ringing up a bill.

The TechRadar piece highlighted a simple truth. Token discipline will not emerge from developer choice alone. And they’re right. Developers aren’t wired to optimise cost; they’re wired to ship.

AI tools make them faster, so they lean in even harder. That means:

  • More prompts.
  • Bigger prompts.
  • Longer model contexts.
  • Infinite iterations.

It’s like giving a teenager limitless Deliveroo credit and then being shocked when you see the bill.

The New Economics of Engineering

For years, the biggest cost in engineering teams was simple. People. Salaries. Benefits. Laptops. That era is quickly ending.

AI code generation tools are accelerating output exponentially, but they’re also creating a new financial dynamic. Instead of:

More developers = more cost

We’re heading toward:

More AI usage = unpredictable cost

This shift has consequences for hiring, budgeting and leadership. If AI model usage becomes the dominant cost centre, then technical leaders need to learn a new skill: cost-aware architecture for AI workflows.

It’s not just engineering anymore. It’s engineering economics.

Why This Matters for Hiring and Team Design

As a recruiter, I see the ripple effects already. Companies are starting to ask for developers who understand not just how to build with AI, but how to build intelligently with AI.

They want people who can answer questions like:

  • How do we reduce wasted tokens?
  • How can we batch requests instead of sending thousands of small ones?
  • Can we use smaller models without losing quality?
  • How do we track AI costs at a feature level?

In other words, 'Can you make our AI bills stop inflating like a badly packed lilo?'

Expect job specs to shift. Expect AI usage governance to become as standard as Git hygiene. Expect CTOs to be quizzed on token burn rate the same way they’re grilled about cloud spending today.

What Smart Leaders Should Do Now

If AI coding spend is set to surpass developer salaries by 2028, then leaders need a plan before the CFO knocks again.

Start with these moves:

  • Create visibility. Implement tools that monitor AI usage by team, project and model.
  • Set guardrails. Define who can use what models and when.
  • Review prompts. Long prompts are the new memory leaks.
  • Educate teams. Make token awareness as normal as version control.
  • Hire with intent. Recruit developers who understand AI cost optimisation.

And most importantly, don’t treat AI spend as background noise. It’s becoming a primary budget line.

The Bottom Line

AI is the most powerful lever engineering teams have ever had, but it isn’t free. The companies that win won’t just be the ones who adopt AI fastest. They’ll be the ones who manage it smartest.

If AI coding costs really do overtake developer salaries by 2028, then the question isn’t how you build with AI. It’s how you build sustainably with AI.

Get ahead of it now or your CFO will get ahead of you later.

And trust me, you’d rather choose the timing.



Back to news