Search

Tag: #cloud-infra23 results

Meta Adopts AWS Graviton5 for Agent AI

Meta Adopts AWS Graviton5 for Agent AI

Meta partners with AWS to enhance agentic AI by deploying tens of millions of AWS Graviton cores, primarily the latest Graviton5. It optimizes CPU-intensive tasks like inference, code generation, and autonomous task adjustment. This diversifies AI infrastructure and boosts energy efficiency for next-gen AI development.

ITmedia AI+ (日本)MediaApr 24#arm-processor#agentic-ai#cloud-infra
Amazon's $25B Bet on Anthropic

Amazon's $25B Bet on Anthropic

Amazon announced up to $25 billion investment in Anthropic, with $5B upfront and $20B conditional on milestones, following prior $8B total. Anthropic commits to over $100B AWS spend over a decade and 5GW Trainium chip capacity. Claude models integrate directly into AWS portal for customers.

Oracle Cuts 30K Jobs for AI Data Centers

Oracle Cuts 30K Jobs for AI Data Centers

Oracle sent termination emails to employees in the US, India, Canada, and Mexico on March 31 without warning. TD Cowen estimates the layoffs will impact 18% of Oracle's 162,000 workforce, freeing up $8-10 billion for AI infrastructure investments. Oracle has not confirmed the exact number of cuts.

The Next Web (TNW)MediaMar 31#layoffs#cloud-infra#ai-investment
OpenClaw Tops GitHub, Mac Minis Sell Out

OpenClaw Tops GitHub, Mac Minis Sell Out

OpenClaw AI agent surges to GitHub's all-time top starred repo in 100 days, outpacing Linux and igniting 'raise lobster' trend. Demand for local runs causes nationwide Mac mini shortages. Clouds like Tencent and Alibaba launch hosted services amid privacy warnings for exposed instances.

AI Search: God-Level Infra Opportunity

AI Search: God-Level Infra Opportunity

Article positions AI search as a crucial intermediate layer for constraining LLM context in agents and apps, akin to object storage in cloud computing. It highlights low technical barriers, massive small-dev market, and expansion potential like bundled compute. Strategies include starting with 'fake crawling' via third-party APIs before building proprietary indexes.

Page 1 of 3