DeepSeek threw a monkey wrench into a powerful AI narrative last week.

It challenged the story that developing AI models requires capital that only Big Tech could afford. 

It went like this:

Big Tech’s massive balance sheets provided a competitive moat. That moat justified $100  million outlays to build proprietary AI models. There would only be a few winners because only they could afford to compete.

But DeepSeek changed the AI story with a plotline to drain those moats.

The team behind it claimed to have trained an AI model that could outperform ChatGPT for $5 million.

Now, it doesn’t really matter what the claims behind the model’s performance are or how much intellectual property was stolen. What matters is that the “Inexpensive AI” storyline has entered the narrative.

The minimum capital requirement to build an AI is no longer a difficult obstacle to clear. More capital sources can now compete in the AI model space, thereby undercutting the big balance sheet advantage.

Big Tech’s part in AI’s story changed this week.

But energy still has the most important role to play.

Intelligence Anarchy

No matter how the story is written, AI will work its way into every aspect of economic life.

A superior ability to work with and manipulate AI will make the difference between living hand-to-mouth or abundantly. Artificial intelligence becomes the means by which you gain a competitive edge. 

Prior to DeepSeek’s release, the assumption was that Big Tech and/or governments would own that edge. They could dole out access to varying levels of intelligence according to its value.

The more valuable the intelligence, the more those who could control it could charge for access.

Marginally valuable intelligence would be priced at a commonly accessible price (like a smartphone). More valuable intelligence would come at a higher price. And the most valuable intelligence would be monopolized.

But that hierarchy has been shattered. Without a hierarchy for competitive intelligence, we have, by definition, intelligence anarchy.

More accessible AI means the benefits of AI become more accessible, and growth will accelerate along with that increased accessibility.

The cost of building individualized AI will decline to the cost of the energy required to run it. The drive to remain competitive means that it will consume all available energy.

Currently, the most abundant source of energy is oil and gas. Nuclear energy could become the most abundant, with fusion ultimately taking over, but those resources are years away.

Until then, oil and gas will be the go-to source powering an individualized AI economy.

And demand for it will grow as quickly as it can be supplied.

Think Free. Be Free.

Don Yocham, CFA

Managing Editor of The Capital List