nextbig.dev reads 300+ curated sources, scores every story with Claude's frontier model, and writes the daily briefing on AI and the compute underneath it — models, GPUs, datacenters, and the economics of inference. Every edition closes with The Wager: one falsifiable call, with a date on it. We keep score.
DiffusionGemma ships open weights with 4x faster decode and day-0 SGLang support; Google books Intel for 3M+ TPUs; DeepSeek goes GW-scale.
Every story below is real, sourced, and scored — hover to pause, click to read. Open the full wire →
One pipeline, three editions — built so a working engineer can hold the entire AI news cycle in fifteen minutes a day.
300+ curated sources read continuously. Claude scores each story 1–10 for builder relevance — what you see survived the cut; the noise never makes the page.
Open the wire →The mechanism behind every headline, the arithmetic the source skipped, and a position you can act on — closed by The Wager, a falsifiable call with a date on it.
Read today's →One thesis, developed from the week's strongest wager — an original claim you won't find in any source story, settled in public when the date comes due.
Read the essays →One email per week. What changed in AI and compute, what it costs, and what to do about it. No spam, no filler.
First briefing lands in your inbox next week. In the meantime, open the wire or read today's edition.