AIWA "The One Thing" #04: Clawdbot just gave most people their first glimpse of a personal AI. AI infrastructure bear case is playing out in real-time.

Clawdbot is fascinating for two reasons. 

1: AI doesn’t automate jobs, AI automates tasks. Augmentation is where we get 5-10x superpowers.

As Richard Socher said, Clawdbot “is the first glimpse of a personal AI for many people” and as a result, it’s becoming abundantly clear how they can “automate themselves and scale their output”. 

Doers and those who get “paid for output will love AI” but “you will need to have the agency and intelligence to pick the right problems” aka taste / subject matter expertise is the moat.

Equally as important is to “run at it”, as Bill Gurley would say. Always.

2: How long will it take the market to digest this?

Gavin Baker was interviewed on Invest Like the Best with Patrick O’Shaughnessy on 12.09.25 and shared his warning about the “scariest bear case” for AI infrastructure (32:30). 

He also endorsed launching data centers into space in a major way (52:50). A big s/o to my man Philip Johnston, an AI with Alec podcast guest back in April 2025.

If you combine Gavin’s “scariest bear case” with the most recent E259 of All-In Podcast, Jason Calacanis’ demo of his team’s Clawdbot build and Chamath’s comments about Kimi K2.5 (45:54), this starts to feel like DeepSeek 2.0.

(*AIWA Newsletter #34 covering DeepSeek 1.0)

Reinforced by an outstanding Twitter thread, “The Edge AI Bear Case is Playing Out in Real-Time” from Jordi Visser who does an excellent job connecting the dots:

“His timeline? 3 years. This week’s All-In showed it’s not 3 years away. It’s starting now.”

“Jason cancelled $25K/year in OpenAI subscriptions.”

Something tells me that thread is moving around the street, also in “real-time.”

$1 bet it is going to help the market process this info a bit faster this time around…