Useful Life
The soft spot in the AI capex cycle is not the power line. It is the assumption underneath the earnings.
Read the essayFor the people who actually have to decide.
One essay a week. Long enough to matter, short enough to read. Always sourced. Always disclosed.
The soft spot in the AI capex cycle is not the power line. It is the assumption underneath the earnings.
Read the essayThree essays that each show a different strand of the argument the rest of the archive is building on.
After chips and power, permission is the third constraint. How AI infrastructure became a regulatory cost-allocation story.
Every Fortune 100 board is about to hire one. The reporting line determines whether the role works — and it gets decided before the job description is even posted.
The shareholder suit that turned AI oversight from a governance hobby into personal director liability.
Pick the question you're actually trying to answer.
Decisions playing out over 18–36 months, not news-cycle takes.
Tracing cascading effects, not just first-order outcomes.
Connecting lessons across banking, manufacturing, healthcare, EdTech, governance.
Written for board directors, PE operating partners, CTOs, GCs, and founders.
The process industries spent 70 years learning to govern autonomous control of complex physical environments. Physical AI has not imported the discipline yet.
AI has a body now, and the discipline to run it is not yet in place.
The week Claude Fable and Mythos went dark showed that frontier AI has moved from product to permission. The architecture implications will take years to work through.
Long enough to chew on Monday. Short enough that you actually read it. Read by board directors, operating partners, and founders working out what to do about AI.

Ritesh Vajariya advises boards, PE operators, and founders on AI strategy through AI Guru, the firm he founded. Previously eighteen months at Cerebras as Global Head of Generative AI Strategy & Business Development. Before Cerebras, at Amazon Web Services: led the team behind Bloomberg’s BloombergGPT, ran GTM for Amazon SageMaker, and launched Amazon Bedrock, growing AWS generative-AI revenue from $40M to $700M. Has trained 100,000+ professionals on enterprise AI adoption and risk.