The AI Investment Paradox

Listen to Steve read this post below (5 min audio)

It seems like forever, but functional AIs have been around for less than 3 years.

In that time, we’ve seen an inordinate amount of new versions and features arrive. I actually added them up. For OpenAI alone, it’s well over 400 updates, with 150 of these being substantive changes. Not just new versions, but radical feature upgrades including voice, live video, live web search, a web browser, data analysis, image analysis, file analysis, image generation, Sora video generation, persistent memory, custom instructions, agent mode, and custom GPTs/apps, to name a few…

That’s crazy.

Of course, we can add to this the arrival of Gemini, Claude, Perplexity, Grok, Mistral, DeepSeek, and all their features.

The AI Intelligence Explosion

We are in the middle of an intelligence explosion. Historically, we expected computing power (capability) to double every 18 months or so. This has changed for one key reason: the new AI tools can create better versions of themselves. With some human guidance, sure, but the new models are literally used to create new models, giving birth to their own progeny. This is what is creating a sudden acceleration in capability.

This is what happens when intelligence improves the tools that improve intelligence. This is why we have this compounding effect of recursive self-improvement.

I don’t really have a better name for it, but it goes something like this:
If you invest now building out some AI tools, often on top of the available technology to solve a bespoke problem, the technology may, and probably will, arrive with an off-the-shelf version of this as a feature or tool while you build, or shortly after, hence making the investment redundant.

The paradox? If you wait, you may miss the opportunity. If you invest, you end up with sunk costs. I’ve seen this in every realm of AI innovation… the net result is inertia and subservience to platform operators.

If anything, it again shows the power of platforms and reminds us the best place to invest is in unregulated technology monopolies.

The second best time is now

So do we wait to avoid the sunk costs? I say we do it anyway. And yes, I’m sure economics or marketing textbooks will tell me I’m wrong. But sometimes the behaviour is more important than the potential sunk costs. The undertaking of adopting emergent technology doesn’t just build a better solution, it solidifies a culture of innovation, change, and exploration — a business which is future-centric. The worst-case scenario is that you or your company move up the learning curve and find out how to use the tools better, once they arrive.

As far as strategy goes, the history of companies that succeeded by waiting for the next version of the technology is very short. Winners rarely wait.