Nvidia Chips Linked to China's Top AI Models
New research suggests Chinese AI models were trained on U.S.-banned Nvidia chips, raising concerns about export controls.
Browse all published articles.
New research suggests Chinese AI models were trained on U.S.-banned Nvidia chips, raising concerns about export controls.
Nvidia's co-founder and CEO stepping down has left the startup ecosystem without a key investor, potentially impacting funding, valuations, and competition.
Cloudflare has open-sourced Kitesurf, a stateless browser for AI agents, under the Apache 2.0 license, allowing developers to run their own browser engine on Cloudflare Workers.
Google introduces new observability and evaluation tools for AI agents in BigQuery and Looker, helping enterprises monitor performance, costs, and data usage.
AMD's RCCL and rocSHMEM now support GPU-initiated SDMA for intra-node data movement, improving bandwidth efficiency in GPU-accelerated workloads.
Amazon SageMaker HyperPod now includes Data Capture, enabling teams to record live inference data for model refinement and analysis.
AssemblyAI reduced inference compute costs by 70% using AWS EC2 Spot Instances while maintaining service level agreements across European regions.
AWS now offers OpenFold3 via SageMaker JumpStart, enabling users to predict 3D structures of proteins, DNA, RNA, and small-molecule ligands with minimal setup.
AWS now allows developers to route coding agent requests to the most suitable model on Amazon Bedrock using NVIDIA NeMo Switchyard, optimizing cost and efficiency.
GS Caltex automates model deployment using Amazon SageMaker Canvas and Pipelines, reducing deployment time from a week to hours.
The DeepSeek App's core features are now accessible through the DeepSeek API, enabling developers to integrate search, file processing, and safety checks into their applications.
AllenAI introduces XYBench, a benchmark of 8K+ questions designed to evaluate how AI assistants address user requests based on misconceptions. The benchmark won the Best Resource Paper Award at EMNLP 2026.