AI is behaving like ERP
AI is behaving like ERP now
You laid off people to save money on AI.
Now AI costs more than the people did.
So now what?
Can you go back?
Or you got to live with it?
People who have lived through ERP implementations will get a Deja Vu feeling here.
Let's set the context first.
I am sure you have already read about some very interesting recent news items regarding 'Cost of AI'
- One company accidentally spent 500 million dollars in a single month on Claude because no one set usage limits.
- Uber burned through its entire 2026 AI budget by April. Per engineer per month? Between 500 and 2,000 dollars.
- Microsoft canceled most of its internal Claude Code licenses because the justification couldn't survive the spreadsheet.
- One healthcare enterprise consumed 1 trillion tokens in six months. 6 million dollars in unplanned costs before anyone realized what was happening.
CFOs are now uncomfortable while CIOs are transfixed - whether to continue to charge ahead or hit the brakes.
In the beginning, the logic was very simple.
Teams get smaller as AI does a lot of work.
Automation equals manpower reduction
Manpower reduction is cost reduction.
Headcount drops. Budget shrinks.
Clean math. Except, it is not.
It's got super messy.
AI Bill has arrived.
Now the ERP Deja Vu ...
Do you remember #TCO framework from the ERP days?
The Total Cost Of Ownership
ERP wasn't just implementation.
It was Training.
Integration.
Customization.
Infrastructure.
Change Mgmt.
Maintenance.
Annual Licenses
And most often, these were not accounted in the beginning.
Organizations began to understand the full force of TCO in about five years
And then there was not turning back ...
AI is following a similar path. Except FASTER.
The question and conclusion at the end of the tunnel is still the same:
You can't go back to manual.
You've already let the people go.
You can't justify hiring them back.
And you can't shut off AI because your entire operating model now depends on it.
So what's the move?
Do you optimize by:
Building guardrails that slow down adoption and leave productivity on the table?
Running guardrails that never quite stop the bleed?
Or do you accept that AI adoption has a real cost of ownership that nobody bothered to calculate upfront, and now you're paying the premium for skipping the math?
In case you still missed the point:
This isn't an AI Problem.
This is a Leadership Problem.
This is a holistic planning Problem.
And ... Picture abhi baaki hain mere dost!
(The game has just begun ... long way to go)

Comments
Post a Comment