THE TECHNOLOGY BRIEFING
The technology archive. Every edition, in one place.
Earlier editions are retrospective collections of dated announcements, added to this archive later. Each briefing shows its source date and actual archive publication date. Historical dates do not imply that this website published at the time.
NASA launched the Nancy Grace Roman Space Telescope in 2026 to explore dark energy, dark matter, exoplanets, and astrophysics from a strategic orbit near the Sun-Earth Lagrange point L2.
In 2026, the ISS continued to play a vital role in human space exploration by testing technologies and studying human health for forthcoming Moon and Mars missions.
Amazon Quick's agentic AI capabilities were launched in AWS GovCloud (US-West) to support secure, compliant AI-driven workflows for government and regulated industries.
Amazon introduced EC2 R8a instances featuring 5th Gen AMD EPYC processors in Canada (Central) to boost compute and memory bandwidth for intensive workloads.
Amazon Bedrock extended cost allocation by AWS Identity and Access Management (IAM) principals to its bedrock-mantle endpoint, improving expense attribution for generative AI use.
Amazon SageMaker JumpStart incorporated NVIDIA’s LocateAnything-3B and Qwen’s large-scale AI models to expand enterprise AI capabilities in visual grounding and agent simulation.
NVIDIA's Nemotron 3.5 Lightning, optimized for persistent agents and enterprise automation, became available on SageMaker JumpStart, offering fast processing and reduced task completion time.
AWS Glue enhanced usability by enabling direct access to SageMaker Unified Studio for streamlined data analysis and AI workflow development.
AWS Secrets Manager expanded managed external secrets support to include Jenkins API and SonarQube tokens, enabling seamless credential rotation without custom code.
Amazon Connect Customer launched a dashboard to help managers monitor case volumes, resolution times, and SLA performance by team and case attributes.
Amazon SageMaker JumpStart broadened its AI model offerings by adding FLUX.2-small-decoder for image generation decoding and Google's gemma-4-12B-it for multimodal understanding, enhancing enterprise AI deployments.
AWS launched models including langcache for semantic caching, Mellum2 for agentic coding, and LightOnOCR for efficient document OCR, enabling diverse AI workloads via SageMaker JumpStart.
AWS integrated models targeting complex engineering workflows, high-throughput reasoning, and detailed document parsing to SageMaker JumpStart, empowering scalable AI applications.
Amazon EC2’s High Memory U7in-24TB instances became available in São Paulo, offering large DDR5 memory and high network bandwidth for demanding enterprise applications.
GameLift Streams introduced automated shader cache capture and distribution, reducing in-session visual stuttering and load times without requiring application modifications.
AWS Marketplace enhanced buyer communications by allowing selection of notification categories with routing to different recipients and delivery channels including email and team collaboration tools.
AWS Identity and Access Management introduced account access manager, enabling centralized assignment of IAM roles to workforce users via IAM Identity Center for simplified permissions management.
EC2 added application status checks that monitor application health indicators such as web server responsiveness or Docker daemon status, enhancing operational monitoring.
Amazon Managed Streaming for Kafka added support to migrate existing Kafka clusters to the KRaft consensus protocol without new infrastructure or client reconfiguration.
The next-generation Amazon OpenSearch Serverless platform scaled up support for multi-tenant environments by increasing collections per group from 1,500 to 10,000.
AWS introduced R8i and R8i-Flex instances in Europe (Milan) powered by custom Intel Xeon 6 processors, offering improved memory bandwidth and performance over prior Intel-based instances.
AWS enhanced Amazon Timestream for InfluxDB with features enabling users to create, schedule, and restore backups to improve data protection and recovery options.
AWS added Amazon Cognito as an essential skill within the Agent Toolkit for AWS, facilitating AI agent configurations for secure user and service sign-in flows.
AWS updated IAM Identity Center to allow customers to enable multi-Region support for new organization instances via a single-click during setup, reducing manual configuration steps.
AWS enhanced Amazon VPC IP Address Manager with tools to centrally monitor BGP route protection and automate RPKI management for Bring Your Own IP prefixes across accounts and regions.
The AgentCore platform for building and optimizing AI agents gains advanced context memory, policy controls, and simplified orchestration specifically for regulated environments in AWS GovCloud (US-West).
CloudWatch Alarms introduced wall clock evaluation windows to align metric alarm evaluations with fixed calendar periods like hours, days, or weeks, improving monitoring accuracy for scheduled workloads.
Amazon GameLift Servers extended its managed fleets to support additional instance types across new compute-optimized and general-purpose families, providing more architecture and processor options for game server workloads.
AWS prolonged security and operating system patch coverage for older Elasticsearch and OpenSearch versions, offering customers an extended upgrade runway with defined support timelines and fees.
Amazon Simple Email Service enhanced event notifications with an isBotEvent field to distinguish likely automated email opens and clicks from human interactions.
Amazon Connect enhanced supervisor tools with historical agent adherence data to improve scheduling oversight.
Amazon ECS introduced fractional GPU allocation for right-sizing container workloads using G6f instances.
AWS Lambda console now supports seamless transitions to Kiro and Cursor IDEs for local serverless development.
AWS WAF integrates Salt Security’s managed rules to detect and mitigate API attacks and AI-related threats without custom rule creation.
The Europe (Spain) region gained Amazon EC2 G7 instances featuring NVIDIA RTX PRO 4500 GPUs designed for accelerated AI and graphics workloads.
AWS Bedrock AgentCore offers runtime instances on EC2 for sustained, resource-intensive AI agent operations with managed infrastructure.
New Graviton4-powered ElastiCache nodes offer higher throughput, lower latency, and better price-performance over Graviton3 predecessors.
AWS Glue Schema Registry now available in more regions to centralize schema management for streaming data applications.
Deadline Cloud users can now monitor and manage rendering farms across all supported regions through one interface.
OpenSearch UI gains network-level access restrictions leveraging existing AWS IAM and VPC condition keys.
NASA's Perseverance rover recorded the first Earth disappearance behind another planetary body's moon from the Martian surface.
Amazon RDS for SQL Server broadened BYOM support to ten new AWS regions for license portability.
AWS Lambda functions outside VPCs gained scalable network throughput tied to memory configuration, improving data transfer speeds.
AWS expanded Amazon Keyspaces availability to Canada West, enabling low-latency Cassandra-compatible access within the region.
AWS upgraded Glue Data Quality’s anomaly detection with a new observation mode and eliminated charges for ETL anomaly monitoring.
AWS Marketplace added AI-powered descriptions that clarify product pricing structures prior to purchase.
DynamoDB introduced native vector search supporting billions to trillions of vectors with millisecond latency.
AWS enhanced Glue Data Catalog preview to export metadata as queryable S3 tables and support encrypted catalogs.
New IAM Identity Center instances can be configured to manage only AWS application access without provisioning AWS account access roles.
Aurora Serverless now achieves up to 12 ACUs within one second to handle agentic AI and bursty traffic efficiently.
New editions: Monday–Friday at 09:00 New York, subject to source availability and the generation allowance. AI-assisted briefings link to original sources. Twenty additional desk names are editorial pseudonyms, not profiles of real journalists. These briefings are separate from peer-reviewed research.