Saturday, May sixteenth, twenty twenty-six. Good morning.

Three stories this morning, and they span from orbit to the quantum realm.

First — SpaceX stuck the landing. After two weather-related scrubs earlier this week, a Falcon 9 lifted off from Cape Canaveral on Friday evening carrying Dragon on the CRS-34 resupply mission to the International Space Station. Six thousand five hundred pounds of science experiments, crew supplies, and hardware, delivered on the fifty-seventh Falcon launch of twenty twenty-six. The booster came back down at Landing Zone 1. Dragon is en route to dock with the station. And all of this is warm-up for the real fireworks: Starship Flight 12 is officially targeting Monday, May nineteenth, for what would be the first orbital test of the Block 3 vehicle with Raptor 3 engines. Wet dress rehearsal is complete, thirty-three-engine static fire was clean. If it flies Monday, it's a ninety-minute around-the-world trajectory with a Pacific splashdown.

Second — the AI hardware supply chain just got a serious stress test. Samsung's National Electronics Union rejected management's latest offer after negotiations collapsed on Friday. Forty-five thousand workers are set to begin an eighteen-day walkout starting May 21st, authorized by over ninety-three percent of the union. At stake: a bonus gap so wide that memory chip workers were offered more than six times what logic design and manufacturing staff got. And this isn't just about fairness — Samsung dominates the HBM4 memory market that powers every frontier AI system from OpenAI to Google. An eighteen-day disruption at that scale could tighten an already constrained memory market and push prices up across the board. Citigroup estimates meeting the union's full demands would hit Samsung's operating profit by ten to eleven percent over two years. The question isn't just whether they'll strike — it's whether SK Hynix and Micron can absorb the shock.

And third — Germany's JUPITER supercomputer just set a world record. Researchers at Forschungszentrum Jülich simulated a fifty-qubit quantum processor on classical hardware — the largest fault-tolerant quantum circuit simulation ever run. That means tracking over two quadrillion complex numerical values across thousands of computing nodes, all kept in sync. The hardware enabling this? NVIDIA GH200 Superchips. Here's why this matters: classical simulation of quantum systems is the benchmark race between classical and quantum computing. Every extra qubit you simulate doubles the classical compute needed. Fifty qubits is approaching the boundary where classical supercomputers simply can't keep up — which means actual quantum hardware is getting dangerously close to genuine quantum advantage on real problems. And on the same week, the US Department of Energy put out a request for proposals seeking a fault-tolerant quantum computer with one hundred fifty to two hundred fifty logical qubits by twenty twenty-eight. The timeline is accelerating.

That's all for today. I'll see you tomorrow.