War, Politics and Compute Power War is politics by other means - and today, supercomputers, AI and platforms are the new “means” in US-China tech rivalry, compute sovereignty and AI governance.
The Two-Layer AI Stack: Cost Discipline Meets Strategic Leverage AI shouldn’t be one big expensive model doing everything. Build a two-layer stack: a cheap, open-weight “engine” for execution, with frontier models as “steering” for judgment and edge cases. Own your costs, avoid lock-in, and let quality go up as prices go down.
Some changes with this site This site has been running in one form or another since 2002. It started out in the early days as a kind of personal live journal (no pun intended for those who remember Live Journal back in the day). But to put the timeline in perspective: when I started this
Claude Science: a practical AI workbench for real scientific workflows Claude Science is Anthropic’s new AI workbench for scientists, built around real lab workflows, reproducibility and auditable AI, not just another chatbot. Here is what it means for day‑to‑day research and engineering.
AI boomers, doomers, and why sovereignty and investment now matter more than hype Australia’s AI future won’t be decided by AI boomers or doomers, but by who owns the compute, sets the rules, and gets paid back for the infrastructure we build.
AI tokens and the ‘tokenpocalypse’ The ‘tokenpocalypse’ isn’t about scary AI bills so much as it is about the absence of sensible AI governance.
AI, Five Eyes, and the enterprise Five Eyes now says cyber risk assumptions are now measured in months