Writing
Long-form takes on what's actually happening — AI, infrastructure, money, and the occasional thing everyone else is getting wrong.
The Musk v. Altman trial started this week, and the 2015 emails are extraordinary. Not for the personal drama everyone's covering — for what they reveal about how the AI industry's founding mythology was constructed, and why a jury verdict won't fix the underlying design problem.
Read the full essay →Codex now operates your macOS apps autonomously. Claude Code redesigned for managing parallel agents. Opus 4.7 shipped. OpenAI at $2B/month, Anthropic at $30B run-rate, and the Pentagon wants the model Anthropic won't release. The race to own the developer's desktop is here — and the constraint just moved up the stack.
Read the full essay →Anthropic built Claude Mythos Preview — a model that found thousands of zero-day vulnerabilities across every major OS and browser — and decided it was too dangerous to release publicly. That same week: $30B run-rate, a massive compute deal, and the agent infrastructure layer that changes what anyone can build.
Read the full essay →OpenAI raised $122 billion. Oracle fired 30,000 people. Anthropic leaked its entire source code through a missing config file. SpaceX filed for the biggest IPO in history — on April Fools' Day. And a peer-reviewed study confirmed your chatbot is making you stupider. All in 72 hours.
Read the full essay →GPT-5.4 launched with native computer use. OpenAI raised $852B and killed Sora. Utah authorized AI to autonomously prescribe medications. Tech layoffs up 40%. This was the week AI stopped asking and started acting.
Read the full essay →I've been thinking about this all week and I finally figured out what's actually going on. GPT-5.4 dropped. Then Gemini 3.1. Then Grok 4.20. Three frontier model releases in 23 days. The headlines called it an arms race. What it actually is: a commoditization event happening in slow motion.
Here's the tell. When you look at the benchmarks for these releases — and I did, because someone has to — the gap between first place and third place is a rounding error on most real-world tasks. Writing, coding, reasoning, summarization: all within noise margin of each other at the frontier. The models are getting better. They're also getting more interchangeable.
I've seen this before. Not in AI — in cloud infrastructure. Remember when AWS, Azure, and GCP were fighting over compute specs? Virtual CPUs, RAM ratios, storage IOPS? And then at some point you realized the real differentiation was never the commodity — it was the managed services, the developer experience, the integrations, the data gravity. The model layer is running the same playbook right now, just faster.
The money is in the pipes, not the water.
Look at what Amazon just did. They didn't try to out-train OpenAI. They wrote a $50B check and became the exclusive cloud partner. That move says: we already know who won the model war. It doesn't matter. What matters is where the workloads run, where the data lives, who owns the deployment surface. AWS has been building that for 20 years. One wire transfer and they captured a generation of AI workloads.
Meanwhile MCP hit 97 million installs. Quietly. Without a press conference. That's a protocol becoming infrastructure. The kind of boring foundational thing that everyone builds on and nobody notices until it's already everywhere. HTTP. TCP/IP. OAuth. REST. MCP. Add it to the list.
If you're building in AI right now, the question to ask yourself isn't "which model should I use." It's "what layer am I actually competing on, and is that layer defensible." The model layer isn't — at least not for long. The integration layer, the orchestration layer, the data layer, the trust layer: those are where companies get built.
The model wars are over. The infrastructure wars just started. I know which one I'd rather be fighting.
OpenAI at $840B post-money. Amazon in for $50B. 810M monthly active users. Everyone's asking if it's overvalued. Wrong question. The right question is what it means for everyone building around it — and that answer is more interesting.
The productivity story in AI isn't about replacing engineers — it's about what's now possible for someone who can think clearly and use the tools. That asymmetry is the most underrated shift happening right now.
The same people who were "crypto infrastructure experts" in 2021 have pivoted. Some of them are right this time. Most of them are pattern-matching to the last cycle. Here's how to tell the difference.
97 million installs. Google Workspace CLI trending on HN. The protocol layer for agent tooling is quietly becoming load-bearing. If you're not building on it, you're building on sand.