Jason Says
NVIDIA spending $12B to acqui-hire Poolside proves AI talent is now worth more than most AI companies. Instead of chasing model capabilities, build a better harness — Nvidia's own research confirms this.
NVIDIA executes a $12B reverse acqui-hire of AI coding company Poolside: founders stay for $1B, employees move to NVIDIA for $6B, and the remaining funds scale a 7GW neocloud infrastructure. The deal marks an unprecedented escalation in the AI talent war.
Zhipu AI (Z.ai) CEO Jie Tang discusses GLM 5.3, arguing that parameter scaling is dead and post-training is the new scaling law. This aligns with the industry trend: pre-training gains are plateauing, and techniques like RLHF, distillation, and synthetic data are now the real differentiators.
TypeScript expert Matt Pocock open-sourced his daily Claude Code Agent Skills, including /wayfinder for greenfield project planning and engineering review. Unlike heavy frameworks like GSD/BMAD, these skills are small and adaptable for real engineering work.
Nvidia research shows AI agents perform well through fine-tuning and a well-designed harness, even when the underlying model isn't top-tier. For indie developers, this means investing in your agent workflow may matter more than chasing the latest model.
OpenAI launches AI Futures, a blog exploring how transformative AI could reshape power, governance, the economy, and individual freedom. As global AI regulation accelerates, OpenAI's move to proactively frame the discourse is worth watching.
Simile AI CEO Joon Sung Park, author of the viral Generative Agents paper, is building digital twins of all 8 billion humans, positioning simulation as a new scaling law. From academic fun to serious business, AI simulation is accelerating in marketing and policy domains.
MoneyPrinterTurbo tops GitHub Trending: input a topic or keyword, and it automatically generates scripts, matches footage, adds subtitles and background music, and produces HD short videos. A ready-to-use monetization tool for creators looking to scale content with AI.
Investors: Iconiq、Sequoia 等
AI 会计独角兽，48 小时内从董事会分享数据到融资 1 亿美元成为独角兽——增长数据好到投资人自己追着送钱
轨道数据中心初创，在发射资源紧缺的背景下融资 2.5 亿美元——把 AI 算力搬到太空的赛道正在加速
Subscribe for daily AI updates + free playbook
📘 Subscribe Free