AWS ships CloudWatch Logs intelligent tiering, DynamoDB Bulk Executor, and G7e in new regions
AWS shipped a batch of cost- and ops-focused updates. CloudWatch Logs now offers intelligent tiering, automatically classifying log data across Standard, Infrequent Access, and Archive Instant Access tiers based on access patterns, letting customers retain logs longer at lower cost without operational overhead. Separately, AWS open-sourced a Bulk Executor for DynamoDB that runs count, find, delete, and update operations across all items in a table at scale with no custom code.
On compute, EC2 G7e instances — powered by NVIDIA RTX PRO 6000 Blackwell GPUs — expanded to Frankfurt, Stockholm, and Mumbai, delivering up to 2.3x inference performance over G6e for deploying LLMs and agentic AI. AWS also added an AssetState dimension for Outposts capacity metrics, RDS/Aurora Graviton4 R8g and M8g instances in more regions (Melbourne, London, Paris, Zurich), and MSK Express Broker support for Apache Kafka 4.2 with a new consumer rebalance protocol.
These incremental infrastructure releases matter because they compound into cost and reliability advantages for AI workloads. The G7e regional expansion in particular targets inference latency for European and Indian customers deploying agentic systems, dovetailing with AWS's recent GPU management fee cuts (35% G-series, 60% P-series/Trainium) announced at the Builder Center's one-year mark.
What to watch: whether intelligent tiering meaningfully lowers observability bills (a persistent enterprise pain point) and how G7e's Blackwell-based price/performance compares to competing cloud GPU offerings. None of these are headline-grabbing on their own, but they reflect AWS's steady grind to be the default AI-deployment substrate.
