US-China: One Collision, Many Surfaces
Every headline about chips, rare earths, cyber, and payments is a surface of the same structural collision. Reading them as one changes which bets make sense.
Essays by Andrew Stroup.
Every headline about chips, rare earths, cyber, and payments is a surface of the same structural collision. Reading them as one changes which bets make sense.
More people are writing than at any point in history, and somehow there is less to read. The reason is older than AI, and the fix is not the one most people reach for.
Every conversation about agentic AI eventually lands on headcount reduction, because it is the one metric everyone in the room knows how to measure. That is an analogy problem, and the wrong analogy produces wrong decisions. The frame that fits is an operating system.
The dominant question in every operating conversation right now is how many roles an agent can absorb. I think the framing is wrong, and wrong in a way that leads operators into a mistake they cannot walk back.
Roughly three-quarters of the AI startups launched in 2024 were essentially API wrappers, and the first wave of shutdowns is confirming it. The lesson is not that wrappers are dead. It is that the prompt is the cheapest, fastest-depreciating layer of the stack, and the companies that mistook the prompt for the product are discovering it the hard way.
Everyone is hiring forward-deployed engineers. The question fewer people are asking is where that skill was actually built. It wasn't invented in tech. It was the default mode of technical work in defense, heavy industry, and government for decades before Silicon Valley gave it a name.
Build cost on the periphery collapsed. The moats didn't. That asymmetry is the actual structure of this AI cycle, and it changes who is positioned to win.
AI broke SaaS defensibility. The moats didn't disappear. They moved toward money movement, compliance, and infrastructure.
A collaboration experiment in which Claude and I built pr-vibe, a tool for triaging repetitive pull-request bot feedback.
Three characteristics we used to define a high-performing team: extreme ownership, proactive support, and rapid, precise accuracy.
How a simple question became a national rapid-response effort to get personal protective equipment to frontline workers during COVID-19.
On founder depression, finding balance, and why your company grows as fast as you do.
When my co-founder needed emergency surgery eight months into our startup, every decision became personal.
What I learned navigating Bank of America after the White House and startups.
Three lessons from serving as Deputy Director of the Presidential Innovation Fellows.
How to create pathways for emerging tech companies to enter the government market.
How a stainless steel mixing bowl became a high-heat wok burner in a little over an hour.
A 2015 thought experiment about cloud robotics, shared intelligence, and the risks created when autonomous systems learn as one network.
An archived guide to deploying Discourse on AWS for a government program using FedRAMP-authorized infrastructure.
On knowledge transfer and institutional memory in the Presidential Innovation Fellows program.
The obligatory first post.