Sunday 4 October 2026
Evening EditionRSS
ZOTPAPER

News without the noise

ZOTPAPER

Zotpaper: today's edition

Business

Amazon pledges $1 billion to data center communities, halts NDAs amid backlash

AWS CEO Matt Garman announces 'Built Together' initiative as local opposition to data center construction spreads across the US

Amazon has committed to investing more than $1 billion over the next five years in communities near its data centers, while also announcing it will stop using nondisclosure agreements (NDAs) with government agencies. The moves come as local opposition to data center construction grows, with New York State imposing a one-year moratorium on new permits and more than 100 similar moratoriums reportedly under consideration nationwide.

Zotpaper·4 Oct·3 min·5 sources

Why this leads: Amazon's billion-dollar pledge and pledge to halt NDAs marks a significant shift in how hyperscalers engage with host communities, a move that will shape data center expansion for years.

Read Full Account

Tech

AI

Business

Research Threads

All threads →

Latest Research

All research →
Weekly Digest · 60 papers · 2 Oct

This Week in Agentic AI Research: Runtime Contracts Beat Model Intuition

The strongest results move reliability from model reasoning into explicit state, memory, communication and audit mechanisms.

Threads:Audio-Video Generation Evaluation 1Token Pruning Efficiency 3Object Permanence in VLMs 1Theorem Proving Benchmarks 1Local Learning at Scale 1All threads →
Paper · 4 OctBig Tech

Semi-autoregressive continuous diffusion models match discrete baselines and enable KV caching.

Yair Schiff et al. · NVIDIA, Cornell University

The authors introduce Clock Diffusion, a semi-autoregressive framework for continuous diffusion language models that denoises token embeddings in blocks or sliding windows according to position-dependent noise schedules. ClockDLMs achieve state-of-the-art perplexity among continuous diffusion language models on OpenWebText, surpassing even block discrete diffusion baselines, and outperform continuous baselines on GSM8K. Their Cache Grab samplers accelerate inference by committing high-confidence tokens and applying self-speculative decoding.

Paper · 4 OctBig Tech

Post-training multi-token prediction heads match joint pretraining with far less data

Prachi Badarayani et al. · Microsoft

The authors show that post-training multi-token prediction (MTP) heads on approximately 2.5 billion tokens of target-generated chain-of-thought data matches or exceeds the expected speedup of jointly pretrained MTP heads on math, coding, and knowledge benchmarks, using 10^3 to 10^4 times fewer training tokens. They further propose a chain-aware relaxation of the verification rule that increases expected speedup by 12–16% without degrading task accuracy, and an adaptive controller that recovers 11–14% of the speedup lost by using a fixed draft length.