Hey Income Hunters,

Every few generations, a technology arrives that doesn’t just change how we work — it redefines what “work” even means.

In the 1800s, the railroads transformed America from a local economy to a continental one. Time, distance, and value itself were reinvented.

Today, Artificial Intelligence (AI) is doing the same — only faster.

As Jan Van Eck, the ETF creator put it during an interview last week:

AI is “the second biggest technological shock to ever hit America.” Bigger than the Internet. Bigger than mobile. And this time, the productivity surge is already showing up in earnings, energy demand, and global money flows.

This isn’t a tech bubble, my friends. It’s a compute revolution. Today, we’ll show why this revolution may have a lot of runway …

Token Demand and the cost of processing

Two exponential curves are crossing:

• Token demand (the fuel for AI models) is rising 38× per year.
  
  
• Compute costs (the cost of processing those tokens) are falling as much as 90% per year.

In AI (especially Large Language Models like ChatGPT or Claude), a “token” is a tiny chunk of text or data — usually a few characters, parts of a word, or symbols.

When someone (or something) uses an AI model — every word generated or processed by that model consumes tokens. So “token demand” refers to the total number of tokens processed by all AI models across all users and machines.

In short:

Token demand = the total computational work AI systems perform.

That creates a flywheel of innovation — cheaper computation → more use cases → more data → more compute demand → more chips.

Token demand IS the fuel that drives the AI economy.

The more tokens processed:

• the more electricity used
• the more GPUs (like NVIDIA’s) are needed
• and the more revenue flows to AI infrastructure companies (cloud, chips, and data centers).

When analysts say token demand is growing 38× per year, they mean the total number of tokens processed by AI systems is exploding exponentially …

Meaning more people are using AI tools, those tools are embedded in more products, and increasingly, machines are using AI to talk to other machines.

Think of tokens like miles driven by all the world’s cars:

• The cars are AI models (GPT, Claude, Gemini).
• The gasoline is electricity + computation.
• The miles driven = tokens processed.
  
  

As more cars hit the road (new apps, chatbots, copilots, robots), and each drives longer distances (more complex tasks), total “token miles” explode — even if the cost per mile keeps falling.