History is repeating itself — AI is the new cloud, and the same mistakes are back.

The AI wave is far bigger than the cloud wave ever was — yet I see many of us in tech making the same mistakes all over again. 

I want to share three thoughts that could save companies millions down the line. 

1. Token Cost Optimisation 

Everyone’s eager to jump on the AI bandwagon, but have they stopped to think about what happens when a large chunk of their workflows runs on AI? The cost of AI tokens will matter—a lot. 

We’ve already lived through a similar story with cloud adoption. Many companies moved to the cloud without proper planning, and years later, cloud cost optimisation is still hurting them. Token cost optimisation may sound like a boring problem, but it’s one with huge long-term value. 

2. Modernising the Existing Application Stack 

I see a common approach: pick one or two problematic applications and “modernise” them using AI. 

But AI isn’t just a quick-fix patch—it’s a brilliant opportunity to eliminate tech debt. The thinking needs to start from the ground up. 

Instead of allocating a small budget to tweak a couple of apps, invest in analysing what the right stack for 2025 and beyond should look like. Ask: 

During the cloud boom, many companies spent millions just maintaining integrations between cloud and on-prem systems. Let’s not repeat that mistake. 

3. Waiting for the Right AI Ecosystem 

Everyone wants to adopt AI now. But remember—success depends on the underlying methodology and the ecosystem you choose. 

In the early cloud days, companies jumped to a single vendor, only to realise later that a hybrid cloud setup was the best approach. The pivot cost them dearly. 

The AI world will be no different—today’s “best” model (say, Model A) could be replaced by Model B tomorrow, and suddenly everyone’s claiming B is far better.