Renting Intelligence by the Hour

[saw how new features over the last year are all either "Pro-first" or even "Pro-early", have had lingering thoughts about this since ...]


Think of the gap between two people tackling the same problem:

  • One has ChatGPT Free: rationed prompts, slow responses.
  • The other has an unlimited Pro plan, API credits, and an agent stack.

That gap isn’t talent. It’s tokens.

Variable Cognition

  • In the old model: software was a one-time (or subscription) cost. Once you had Photoshop or Excel, you had it.
  • With AI, cognition is now consumption-based.
  • The person with budget isn’t just “better equipped”: they have more cycles of reasoning, more iterations, more time that "some thing is working for them".

Compounding Advantage

  • If you can afford it, you can't just buy better tools, you can buy more cognition, which compounds.
  • Better model access, more API calls => more experiments in parallel => more building => more revenue
  • Potentially, a positive feedback loop of sorts.

The "Augmentation Gap"

  • This isn’t about efficiency: it’s about amplified "intelligence"
    • (okay, air quotes, call it something else: reasoning, cognition, whatever -- it's about getting more done, and getting it done faster)
  • These "machine minds" are tiered, and there's a real aspect of "human augmentation", which is not linear in its benefits

Is this new?

  • Nothing new about haves & have-nots ... but the 'having' was about tools, resources, and not ... "access to thinking"
  • It's "reasoning/cognition" that's 'on tap', priced by the token
  • Whatever else it may be, it's definitely not "democratizing"

So yeah, I'm really enjoying what I can do with all this, I can make the case for getting a paid subscription to each of Claude/chatGPT/Gemini ... but at the same time, I do wonder where this is going ...