Redis adds SSD-backed Redis Search to Redis Flex in preview for Redis Cloud Pro, letting customers place large indexes on SSD and set RAM usage from 10% to 50% as server DRAM prices roughly double.
Federal prosecutors arrest California businessman Greg Lui, owner of Earthmade Computer Inc., for allegedly sending roughly $300 million in Nvidia AI-chip servers to China from 2023 to 2024 without required U.S. export licenses.
Amazon launches thinner, lighter Kindle, Kindle Paperwhite and Kindle Colorsoft e-readers without AI, adding edge-to-edge displays and book-opening speeds up to 30% faster.
Nvidia launches the Open Agent Safety Platform to quarantine AI agents that breach their boundaries within milliseconds, using OpenShell on its Vera AI CPU and a separate Sentry chip to monitor and enforce access limits.
China is considering approving Nvidia RTX Pro 5500 chip imports for Alibaba and ByteDance, with ByteDance prepared to order 1 million units—potentially reopening millions of sales previously restricted by Beijing.
NVIDIA launches the Open Agent Safety Platform, pairing OpenShell with the Sentry reference design on BlueField-4 DPUs to independently monitor and quarantine AI agents within milliseconds, while Salesforce brings approval and audit controls into Slack.
NVIDIA launches the Open Agent Safety Platform, pairing open-source OpenShell sandboxes with NVIDIA Sentry and BlueField-4 DPUs to enforce verifiable AI-agent rules on files, networks, tools, processes and credentials before and during execution.
Qualcomm says its Snapdragon 8 Elite Extreme Gen 6 can run 30-billion-parameter Mixture-of-Experts AI models directly on phones by loading needed experts from flash storage instead of requiring roughly 32GB of RAM.
Qualcomm launches the Snapdragon 8 Elite Extreme Gen 6, a mobile chip designed to run 30-billion-parameter mixture-of-experts AI models directly on phones, targeting next year’s Android flagships alongside Snapdragon Sound Elite Gen 2 wearables hardware.
Inferact open-sources TPU v7 megakernels and claims its Kimi K3 implementation surpasses 700 tokens per second with speculative decoding on 16 TPU v7 chips, using one Pallas kernel to fuse all 92 MoE layers.