Watching AI reshape tech valuations, I keep thinking about 2008-2014. Not in software. In shale. The parallels are striking.
ChatGPT was released in late 2022 as the most widely visible and immediately accepted large language model (LLM).
It followed years of smaller scale development across various academic and industrial research labs.
Then we got Claude. Then came Gemini.
Now, three years after ChatGPT emerged, the market realizes the rules of the game have changed.
Software engineering isn’t nearly as scarce a resource as it was in the past.
The economic reverberations of this realization are profound.
Backtrack to the early days of shale.
The shale gas revolution kicked off in 2008, after years of smaller-scale experimentation by George Mitchell and others in shale plays.
Within five years, the full unconventionals unlock was upon us, with both oil and gas production booming, and big players entering the market.
US oil and gas resources were no longer as scarce as they had been for decades.
The economic reverberations were profound.
Supply ramped up. Oil and gas prices crashed.
US industrial activity that previously had to move offshore to compete could remain in the country.
The changing energy foundation reset the global economy and geopolitical order.
The tech disruption is following the same pattern.
The scarce resource shifted. It’s no longer human coders. Now it’s compute resources: chips and energy.
The companies that own the infrastructure (e.g. Nvidia and hyperscalers) become the new majors.
The legacy SaaS players that depended on human scarcity as a moat? They’re the ones trying to figure out what assets still matter.
In a world where you can build and deploy a custom enterprise-grade software solution in hours, paying a premium for an equity stake in yesterday’s SaaS behemoth becomes a lot more precarious.
We know history doesn’t repeat itself.
But personally, I hear a lot of rhyming between the AI emergence today and the unconventionals unlock 15 years ago.
(The AI image is of exhausted, overwhelmed coders that are now trying to work on a rig site. I had no idea what AI would come up with here. But I got a chuckle out of it. Hopefully you do too.)
