Technical essays, investment research, and notes on intelligence.
Analysis of supply chain vulnerabilities, memory bottlenecks, coherence limitations, and strategic positioning in AI infrastructure. Deep dive into Google's Broadcom dependency, GB300 drop-in compatibility, Meta's checkpoint crisis, 12-Hi HBM3E constraints, SaaS margin compression, GPU coherence ceilings, edge AI threats, and ROIC dynamics.
Comprehensive deep dive into the $652 billion data center market, examining AI infrastructure revolution, liquid cooling imperatives, power grid challenges, and investment opportunities in hyperscale platforms. Analysis covers the transition to gigawatt-scale training clusters, GB200 NVL72 architecture, and private equity strategies in this generational infrastructure opportunity.
A field guide to agent-native architecture: shared actions, tool contracts, interface standards, and the structural security model required to let agents act without giving them an open path to exfiltrate data.
GPU derivatives could become a market-implied model of AI's hidden state. This essay explains what the curve may reveal, why its signals will remain ambiguous, and how to read prices, spreads, regional basis, and volatility without mistaking them for a crystal ball.
GPT-5.6 Sol on Cerebras turns inference speed into a systems question. A technical analysis of decode latency, wafer-scale hardware, tail reliability, and why time to verified action is the metric that matters for agents.
A research-grounded look at the claim that AI flips ADHD from liability to advantage. What the evidence actually says about divergent thinking, executive function offloading, and the structural timing — plus where the thesis breaks down and the version of the playbook that accounts for the tradeoffs.
A practitioner's take on the tool-interface problem, and why I think the industry picked the wrong default. Typed tool catalogs over MCP are broken in ways that aren't about any particular server — they're properties of the pattern. The fix is to hand the model Unix-style CLI conventions it already knows.
How diffusion, instability, and crystallography conspire to create ephemeral beauty. A deep dive into the physics of pattern formation at the edge of equilibrium—from the Mullins-Sekerka instability to phase field dynamics.
What begins as a poetic observation about decoration becomes a journey into the nature of consciousness, pattern, and intelligence itself. A meditation on perturbation and the dissolving boundaries between natural and artificial.
The ancient Greeks had a word for the art of midwifery: maieutikos. Socrates used it as a metaphor for his philosophical method. Today, something remarkably similar is unfolding in the quiet hum of algorithms.