I initially grouped Newton Protocol into the same category as most AI-related crypto projects: an interesting narrative with a lot of future assumptions baked into it.
But after reading through the architecture a bit more carefully, I started wondering if the AI angle is actually distracting from what Newton is trying to solve.
At roughly a $15M market cap with around $5M in daily volume and less than a third of its 1B supply currently circulating, NEWT is still being priced like a thematic bet on autonomous agents. That makes sense on the surface. AI agents executing trades, managing strategies, and interacting with markets is an easy story to understand.
What seems less obvious is that autonomous execution itself may not be the scarce component.
If AI agents become increasingly capable, then the harder problem becomes defining what they're allowed to do, proving they followed those rules, and creating systems that other participants can trust without relying on the operator's reputation.
That's where Newton became more interesting to me.
The idea of turning policies, permissions, and constraints into something verifiable onchain feels much less marketable than "AI trading infrastructure," but potentially much more foundational. Most of the discussion around AI agents focuses on making them smarter. Very little focuses on making them accountable.
I'm still not convinced the demand for this layer exists at the scale the thesis implies. But I also can't shake the feeling that if autonomous systems do become a meaningful part of crypto markets, the protocols enforcing their boundaries may end up mattering more than the agents themselves.