CLOUD

https://image.cnbcfm.com/api/v1/image/108360951-1789001878121-gettyimages-2293873710-RNC_MIDTERM_CONVENTION.jpeg?v=1789001904&w=1920&h=1080

TECH NEWS

President Trump says Scott Bessent is not “going to be Super Intelligence (SI) Czar”, because “he doesn't want to” and Trump wants to keep him at the Treasury

Sponsor Posts Subquadratic: the LLM built for 12M-token reasoning — SubQ can reason across entire codebases and document sets in one pass with no RAG workarounds. Read how SubQ 1.1 Small holds near-perfect retrieval out to 12M tokens. Cloudflare Connect 2026: The Agentic Conference — Stop preparing for agents. Start running

https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/09/24/ML-21346-featured-image.png

TECH NEWS

Scaling MoE reinforcement learning on Amazon EKS with EFA and DeepEP with 40% more throughput | Amazon Web Services

When you post-train a Mixture-of-Experts (MoE) model with Reinforcement Learning from Human Feedback (RLHF) or Group Relative Policy Optimization (GRPO) at scale, three simultaneous challenges emerge. The first requires coordinating heterogeneous compute for rollout generation and policy training. Second, sustaining high-throughput communication across hundreds of accelerators. And third, dynamically orchestrating