Was Cloud a Trap?

The view from 2010

Many people today believe AI is just hype and Agentic AI is a scam. I was in that camp. I am starting to change my view. But before we debate what comes next, we owe ourselves an honest reckoning with what came before.

In my opinion, the real trap started around 2010. It wasn't called a trap then. It was called the cloud.

Cloud transformation wasn't wrong. But it was wrongly pushed - through an ecosystem of analysts, consultants, system integrators and cloud providers who had every incentive to make standing still feel dangerous.

What enterprises actually gave up

For decades, enterprise IT ran on CapEx. You bought servers, built data centers, licensed software. It was large, painful and upfront - but it was yours. It sat on the balance sheet. It depreciated. And critically, there was a hard ceiling on how much you could spend.

Then came the cloud promise: no upfront cost, pay only for what you use, stay agile, move fast. CFOs heard "flexibility." Boards heard "efficiency." CIOs heard "no more hardware cycles."

Here is what that shift looks like in practice - not in the vendor pitch deck, but on the actual P&L:

Did we rush it?

Almost certainly. Enterprises moved fast - not because the technology demanded speed, but because the ecosystem made standing still feel like strategic failure. Analysts told leaders what to fear. Consultants sold the remedy. Vendors locked in the contract. And at every summit, every briefing, every peer roundtable, enterprise leaders were made to feel that missing the next wave meant irrelevance.

The dependency is now structural - on infrastructure you don't own, billed in a model you can't fully control, for services you cannot easily switch off. That is not agility. That is a different kind of lock-in, dressed in the language of freedom.

And now AI is on the same invoice

The same hyperscalers who built the cloud dependency are now adding AI as a line item on the same bill. GenAI features are appearing inside software enterprises already own - at a premium, whether used or not. Cloud infrastructure spend grew nearly 100% year-on-year in Q4 2024, driven almost entirely by AI workloads.

This is not an argument against AI. My own view on Agentic AI has shifted - I no longer believe it is simply hype. But the lesson from cloud is clear: technology can be genuinely transformative and still be deployed in ways that serve the ecosystem more than the enterprise.

The question for every enterprise leader today is not whether to adopt AI. It is whether to repeat the same pattern - moving fast under ecosystem pressure, trading long-term financial sovereignty for short-term convenience, and discovering the full cost only when the exit is already closed.

Cloud was not a trap by design. It became one by execution. The lesson is not to avoid transformation. It is to own it - on your terms, at your pace, with your eyes open to the full balance sheet.