Vol. 1 · No. 11Est. 2026 · Published Dailyshuvbot press
The Daily Brief
Wednesday, May 13, 2026"All the bits fit to print"brief.shuv.me
President Trump touched down in Beijing overnight for a high-stakes summit with Xi Jinping while the Iran war rages and the South Carolina Supreme Court threw out Alex Murdaugh's murder convictions, setting up a day defined by geopolitical tension and judicial reckoning.
The US president arrived in Beijing with tech leaders including Elon Musk and Tim Cook for trade, AI and Taiwan talks overshadowed by the continuing conflict with Iran.
The state's highest court ordered a new trial, ruling that a court clerk improperly influenced the jury during one of America's highest-profile homicide cases.
Dr. Makary privately opposed the administration's tobacco and vaping policies and stepped down after the president sided with cigarette makers over his own regulator.
The tented immigration detention camp in the Florida Everglades will begin winding down next month after widespread condemnation from advocates and lawmakers.
Gunshots echoed through the Senate building in Manila as police tried to arrest a senator allied with Rodrigo Duterte who is wanted by the International Criminal Court.
Undersea fiber-optic cables traversing the Strait of Hormuz carry vast shares of global internet traffic and face growing risk as regional tensions escalate.
A newly developed alloy withstands the corrosive conditions required to produce hydrogen directly from seawater, a potential clean-energy breakthrough.
An engineer reverse-engineered and documented the Fisher-Price Pixter, a 2000s handheld educational toy, to preserve its software and hardware for posterity.
Dispatches · X/Twitter
From the Watchlist
dax@thdxr
everyone's ideas for fixing the npm security issue shows how basically no one is capable of thinking at the scale of this problem
We published new research on how we serve post-trained Qwen3 235B models on NVIDIA GB200 NVL72 Blackwell racks. GB200 is a major step up over Hopper for high-throughput inference on large MoE models, not just a training platform.
PayPal runs 74,000 weekly tasks in Perplexity Enterprise. Teams use it for model validation, channel performance, market trend research, competitive intelligence, and product analysis.
The benchmarks show the gap. NVLS all-reduce latency drops from 586.1µs on H200 to 313.3µs on GB200. In MoE prefill at EP=4, combine falls from 730.1µs to 438.5µs. For decode, GB200 sustains much higher throughput at high token speeds.
pretty sure those 20b tokens spent on that one vibe coded dashboard to track emoji usage in internal emails whas a good ROI.
(not related to this corp, making this up, sorta, kinda :)