OffNet Newsroom

Daily topic roundup

AWS

Saturday, September 12, 2026 · 8 stories, curated & summarized — click any story for the source.

OpenAI News llmaiagents

OpenAI scales Habitat to 1B users, 22M req/s

OpenAI transformed its internal Python library, Habitat, into a globally distributed storage platform capable of handling over one billion ChatGPT users. The system now manages a peak load of 22 million requests per second. This evolution highlights the infrastructure changes required to support massive consumer-scale AI services.

  • Habitat evolved from a simple Python library into a global distributed storage system.
  • The platform currently serves over 1 billion ChatGPT users.
  • System handles peak traffic of 22 million requests per second.
  • Demonstrates the storage architecture behind massive AI scaling.

The Terraform AWS Provider has released version 6.62.0, continuing its rapid evolution to support emerging AWS infrastructure features. This update enhances Terraform's ability to accurately understand and manage existing cloud environments as they grow in complexity. The release reflects ongoing efforts to keep the IaC tool aligned with the expanding AWS service landscape.

  • Terraform AWS Provider v6.62.0 is now available for immediate adoption.
  • New AWS capabilities are added to support complex infrastructure states.
  • Improved state management helps Terraform better track existing resources.
  • The provider continues rapid iteration to match AWS service expansion.
Hacker News (100+ points) general

EPA to scrap public review rules for data center pollution permits

The EPA is planning to eliminate public review requirements for pollution permits related to data centers. This regulatory shift removes a layer of community oversight regarding environmental impacts from these facilities. The change aims to streamline the permitting process for energy-intensive infrastructure.

  • Public scrutiny of data center emissions is being removed from the permit process.
  • Streamlining permits may accelerate data center construction and deployment.
  • Environmental impact assessments for these facilities will face less transparency.
  • Operators may see faster approval times for new or expanded power usage.
AWS What's New awsdatabase

EC2 X2idn instances now live in Asia Pacific Hong Kong

AWS has enabled memory-optimized X2idn instances in the Asia Pacific (Hong Kong) region. Powered by third-generation Intel Xeon Scalable processors and the Nitro System, these instances offer performance gains over previous X1 generations. They are specifically optimized for memory-intensive workloads and are certified for SAP HANA-based applications.

  • New X2idn availability in Hong Kong supports low-latency SAP deployments in APAC.
  • Instances leverage 3rd gen Intel Xeon and Nitro System for enhanced memory performance.
  • SAP-certified for HANA, S/4HANA, BW/4HANA, and NetWeaver workloads.
  • Ideal replacement for legacy X1 instances requiring high memory density.
BY THE NUMBERSSAP Workloads Supported4Certified SAP application typesHANA, S/4HANA, BW/4HANA, NetWeaver

Form3 engineers detail their migration from a single-cloud baseline to a triple-active multi-cloud setup spanning the UK, Europe, and US. The architecture leverages CockroachDB for distributed data consistency and NATS for messaging, supported by custom Kubernetes operators to manage cross-cloud networking. The presentation also covers how distinct regional disaster recovery expectations in financial markets influenced these technical decisions.

  • CockroachDB enables strong consistency across geographically dispersed cloud regions.
  • NATS handles low-latency messaging requirements in a multi-active topology.
  • Custom K8s operators are essential for managing complex cross-cloud dependencies.
  • Regional regulatory and DR expectations dictate architectural boundaries.
  • Triple-active setups require rigorous cross-cloud networking strategies.
Hacker News (100+ points) general

Google to buy 50% of output from Finnish nuclear plant

Google has agreed to purchase half of the electricity generated by a nuclear power plant in Finland. This move secures a large, stable baseload power supply for its expanding data center operations. The deal highlights the growing reliance on nuclear energy to meet the escalating energy demands of AI and cloud infrastructure.

  • Secures stable baseload power for data centers, reducing reliance on intermittent renewables.
  • Signals increasing energy intensity of AI workloads driving large-scale power deals.
  • Demonstrates trend of hyperscalers partnering directly with nuclear operators for supply certainty.
AWS What's New awsdatabase ↺ since 09-11

Amazon Redshift RG instances now available in Europe (Zurich) Region

AWS has expanded the availability of Amazon Redshift RG instances, built on AWS Graviton processors, to the Europe (Zurich) Region. These new instances offer up to 2.4x better performance and 30% lower price per vCPU compared to the previous RA3 generation. They utilize a custom vectorized query engine to process Apache Iceberg and Parquet data directly, unifying data warehouse and data lake analytics.

  • RG instances in Zurich enable local low-latency access for European workloads.
  • Performance improves up to 2.4x over RA3 with significant cost savings per vCPU.
  • Built-in vectorized engine processes Iceberg and Parquet without external storage.
  • Existing RA3 clusters can upgrade via Snapshot & Restore or Elastic Resize.
TRADE-OFFRG vs RA3 InstancesNew RG InstancesUp to 2.4x better performance30% lower price per vCPUNative Iceberg and Parquet supportPrevious RA3 GenerationStandard query engineHigher cost per vCPURequires external storage for lakesvs
AWS What's New awsdatabase ↺ since 09-10

AWS Lambda extends LMI timeout to 90 minutes for heavy workloads

AWS Lambda Managed Instances now support a 90-minute function timeout, a sixfold increase from the previous 15-minute limit. This change applies to asynchronous and event source mapping invocations, enabling longer continuous execution without architectural workarounds. The update specifically targets data-intensive tasks such as media transcoding, financial modeling, and AI inference.

  • LMI timeout increased from 15 to 90 minutes for async and ESM invocations.
  • Eliminates need for complex architectural workarounds for long-running jobs.
  • Ideal for media transcoding, Monte Carlo simulations, and AI inference.
  • Applies to Lambda Managed Instances, not standard container instances.
THE SHIFTLambda LMI Timeout Jumps15PREVIOUS LIMIT90NEW LIMITSixfold increase for async and ESM invocations
↑