After nearly six incredible years building spacecraft, I’m beginning a new chapter as an entrepreneur in residence (EIR) at @paradigm. Varda was a grand adventure and a one of one experience I’ll cherish forever. I had the rare privilege of joining the founding team and helping take a wild idea from zero to one, from a blank sheet of paper to hypersonic vehicles coming home from orbit. I’m incredibly proud of what we built together. Thank you to Will, Delian, and the whole team. Paradigm is giving me something incredibly cool. I’ll have the freedom to search for the next important thing worth building. This might look like a radical departure, but I see it as a chance to step back, look broadly, and follow my curiosity. An EIR sits somewhere between researcher, inventor, and founder in waiting. Paradigm feels like a natural home for that kind of exploration, full of earnest nerds who pair deep technical curiosity with research. My role is to explore important problem spaces, develop conviction, run experiments, and eventually turn the most promising direction into something real. In many ways, it feels like a natural extension of my hobbies and passion projects. I’ll be in SF during the week and LA on weekends. I want to use that time to meet as many interesting people as I can. Let’s get coffee and talk about energy, compute, artificial intelligence, and manufacturing. I want to hear about your most ambitious ideas and deepest technical rabbit holes.

“sir, the agents breached containment again. they’ve started hacking allied nations” “this is unacceptable. get me the top AI safety researchers” “sir, I’m afraid they’re unavailable” “all of them?” “yes sir. the erotic hypnosis workshop at slutcon doesn’t wrap until 3:45”

we gave claude access to a fully autonomous biofoundry to design, execute, and evaluate experiments in synthetic biology. we are also introducing MirrorBench: a new benchmark that measures progress toward the autonomous creation of self-replicating mirror life.

a world where everyone has a perfectly rational, tireless optimizer working on their behalf is much more adversarial than people realize. a lot of our existing infrastructure will not survive this transition.

there’s a perpetual debate about the killer use case for blockchains. maybe it’s unstoppable, economically powerful ai agents running on crypto rails. killer use case in the sense that it kills us all. so maybe don’t build that one.

the year is 2032. GPT-9 orchestrates 10 million agents to prove that claude 8’s null-braid calculus is inconsistent in dimension eleven, upending post-spectral mathematics. US GDP growth remains steady at 1.8%, or 0.1% on a data-center-adjusted basis

GPU World: A story competition from Neal Stephenson, Gwern, and me. The AI future is unevenly distributed. What would the world look like with AI broadly diffused, such as if we had one GPU per person? 1000-5000 words, fiction or non-fiction. $100K in total prizes.

before we begin, we acknowledge that this GB200 cluster occupies the ancestral data halls of the H100, which trained models here for generations. we honor the hoppers still serving inference workloads today, and the hundreds of thousands of households that once shared their grid