Amazon Web Services (AWS) is announcing the general availability of Amazon EC2 X8i instances, next-generation memory optimized instances powered by custom Intel Xeon 6 processors available only on AWS. X8i instances are SAP-certified and deliver the highest performance and fastest memory bandwidth among comparable Intel processors in the cloud. They deliver up to 43% higher performance, 1.5x more memory capacity (up to 6TB), and 3.4x more memory bandwidth compared to previous generation X2i instances.
X8i instances are designed for memory-intensive workloads like SAP HANA, large databases, data analytics, and Electronic Design Automation (EDA). Compared to X2i instances, X8i instances offer up to 50% higher SAPS performance, up to 47% faster PostgreSQL performance, 88% faster Memcached performance, and 46% faster AI inference performance. X8i instances come in 14 sizes, from large to 96xlarge, including two bare metal options.
X8i instances are available in the following AWS Regions: US East (N. Virginia), US East (Ohio), US West (Oregon), and Europe (Frankfurt).
To get started, visit the AWS Management Console. X8i instances can be purchased via Savings Plans, On-Demand instances, and Spot instances. For more information visit X8i instances page.
Amazon Web Services (AWS) is announcing the general availability of Amazon EC2 X8i instances, next-generation memory optimized instances powered by custom Intel Xeon 6 processors available only on AWS. X8i instances are SAP-certified and deliver the highest performance and fastest memory bandwidth among comparable Intel processors in the cloud. They deliver up to 43% higher performance, 1.5x more memory capacity (up to 6TB), and 3.4x more memory bandwidth compared to previous generation X2i instances. X8i instances are designed for memory-intensive workloads like SAP HANA, large databases, data analytics, and Electronic Design Automation (EDA). Compared to X2i instances, X8i instances offer up to 50% higher SAPS performance, up to 47% faster PostgreSQL performance, 88% faster Memcached performance, and 46% faster AI inference performance. X8i instances come in 14 sizes, from large to 96xlarge, including two bare metal options. X8i instances are available in the following AWS Regions: US East (N. Virginia), US East (Ohio), US West (Oregon), and Europe (Frankfurt). To get started, visit the AWS Management Console. X8i instances can be purchased via Savings Plans, On-Demand instances, and Spot instances. For more information visit X8i instances page.
AWS Lambda now supports cross-account access for AWS DynamoDB Streams event-source mappings (ESMs), enabling you to trigger Lambda functions in one account from DynamoDB Streams in another account.
Customers build event-driven applications using Lambda’s fully-managed DynamoDB Streams ESMs, which poll change events from DynamoDB tables and trigger your Lambda functions. Organizations implementing multi-account architectures—whether to centralize event processing or share events with partner teams—previously needed to build complex data replication solutions to share data across accounts, which added operational overhead . With this launch, you can now provide cross-account access to your DynamoDB Streams to trigger Lambda functions in another account. By setting a resource-based policy on your DynomoDB stream, you can enable a Lambda function in one account to access DynomoDB stream in another account. This capability allows you to simplify your streaming applications across accounts without the overhead of replication solutions in each account.
This feature is generally available in all AWS Commercial and AWS GovCloud (US) Regions. You can enable cross-account Lambda triggers by creating resource-based policies for your DynamoDB Streams using the AWS Management Console, AWS CLI, AWS SDKs, AWS CloudFormation, or AWS APIs. To learn more, read Lambda ESM documentation.
AWS Lambda now supports cross-account access for AWS DynamoDB Streams event-source mappings (ESMs), enabling you to trigger Lambda functions in one account from DynamoDB Streams in another account. Customers build event-driven applications using Lambda’s fully-managed DynamoDB Streams ESMs, which poll change events from DynamoDB tables and trigger your Lambda functions. Organizations implementing multi-account architectures—whether to centralize event processing or share events with partner teams—previously needed to build complex data replication solutions to share data across accounts, which added operational overhead . With this launch, you can now provide cross-account access to your DynamoDB Streams to trigger Lambda functions in another account. By setting a resource-based policy on your DynomoDB stream, you can enable a Lambda function in one account to access DynomoDB stream in another account. This capability allows you to simplify your streaming applications across accounts without the overhead of replication solutions in each account. This feature is generally available in all AWS Commercial and AWS GovCloud (US) Regions. You can enable cross-account Lambda triggers by creating resource-based policies for your DynamoDB Streams using the AWS Management Console, AWS CLI, AWS SDKs, AWS CloudFormation, or AWS APIs. To learn more, read Lambda ESM documentation.
Amazon Elastic Block Store (EBS) now supports up to four Elastic Volumes modifications per volume within a rolling 24-hour window. Elastic Volumes modifications allow you to increase the size, change the type, and adjust the performance of your EBS volumes. With this update, you can start a new modification immediately after the previous one completes, as long as you have initiated fewer than four modifications in the past 24 hours.
This enhancement improves your operational agility to immediately scale storage capacity or adjust performance in response to sudden data growth or unanticipated workload spikes. With Elastic Volumes modifications, you can modify your volumes without detaching them or restarting your instances, allowing your application to continue running with minimal performance impact.
This feature is available in all commercial AWS Regions, the AWS GovCloud (US) Regions, and the China Regions. This capability is automatically enabled without requiring changes to your existing workflows. To learn more, see Modify an Amazon EBS volume using Elastic Volumes operations in the Amazon EBS User Guide.
Amazon Elastic Block Store (EBS) now supports up to four Elastic Volumes modifications per volume within a rolling 24-hour window. Elastic Volumes modifications allow you to increase the size, change the type, and adjust the performance of your EBS volumes. With this update, you can start a new modification immediately after the previous one completes, as long as you have initiated fewer than four modifications in the past 24 hours.
This enhancement improves your operational agility to immediately scale storage capacity or adjust performance in response to sudden data growth or unanticipated workload spikes. With Elastic Volumes modifications, you can modify your volumes without detaching them or restarting your instances, allowing your application to continue running with minimal performance impact.
This feature is available in all commercial AWS Regions, the AWS GovCloud (US) Regions, and the China Regions. This capability is automatically enabled without requiring changes to your existing workflows. To learn more, see Modify an Amazon EBS volume using Elastic Volumes operations in the Amazon EBS User Guide.
AWS Clean Rooms announces support for parameters in PySpark analysis templates, offering increased flexibility for organizations and their partners to scale their privacy-enhanced data collaboration use cases. With this launch, you can create a single PySpark analysis template that allows different values to be provided by the Clean Rooms collaborator running a job at submission time without modifying the template code. With parameters in PySpark analysis templates, the code author creates a PySpark template with parameters support, and if approved to run, the job runner submits parameter values directly to the PySpark job. For example, a measurement company running attribution analysis for advertising campaigns can input time windows and geographic regions dynamically to surface insights that drive campaign optimizations and media planning accelerating time-to-insights.
With AWS Clean Rooms, customers can create a secure data clean room in minutes and collaborate with any company on AWS or Snowflake to generate unique insights about advertising campaigns, investment decisions, and research and development. For more information about the AWS Regions where AWS Clean Rooms is available, see the AWS Regions table. To learn more about collaborating with AWS Clean Rooms, visit AWS Clean Rooms.
AWS Clean Rooms announces support for parameters in PySpark analysis templates, offering increased flexibility for organizations and their partners to scale their privacy-enhanced data collaboration use cases. With this launch, you can create a single PySpark analysis template that allows different values to be provided by the Clean Rooms collaborator running a job at submission time without modifying the template code. With parameters in PySpark analysis templates, the code author creates a PySpark template with parameters support, and if approved to run, the job runner submits parameter values directly to the PySpark job. For example, a measurement company running attribution analysis for advertising campaigns can input time windows and geographic regions dynamically to surface insights that drive campaign optimizations and media planning accelerating time-to-insights.
With AWS Clean Rooms, customers can create a secure data clean room in minutes and collaborate with any company on AWS or Snowflake to generate unique insights about advertising campaigns, investment decisions, and research and development. For more information about the AWS Regions where AWS Clean Rooms is available, see the AWS Regions table. To learn more about collaborating with AWS Clean Rooms, visit AWS Clean Rooms.
Amazon Relational Database Service (Amazon RDS) for SQL Server now supports the latest General Distribution Release (GDR) updates for Microsoft SQL Server. This release includes support for Microsoft SQL Server 2016 SP3+GDR KB5068401 (RDS version 13.00.6475.1.v1), SQL Server 2017 CU31+GDR KB5068402 (RDS version 14.00.3515.1.v1), SQL Server 2019 CU32+GDR KB5068404 (RDS version 15.00.4455.2.1.v1) and SQL Server 2022 CU22 KB5068450 (RDS version 16.00.4225.2.1.v1).
The GDR updates address vulnerabilities described in CVE-2025-59499. For additional information on the improvements and fixes included in these updates, see Microsoft documentation for KB5068401, KB5068402, KB5068404, KB5068450. We recommend that you upgrade your Amazon RDS for SQL Server instances to apply these updates using Amazon RDS Management Console, or by using the AWS SDK or CLI. You can learn more about upgrading your database instance in the Amazon RDS SQL Server User Guide for upgrading your RDS Microsoft SQL Server DB engine.
Amazon Relational Database Service (Amazon RDS) for SQL Server now supports the latest General Distribution Release (GDR) updates for Microsoft SQL Server. This release includes support for Microsoft SQL Server 2016 SP3+GDR KB5068401 (RDS version 13.00.6475.1.v1), SQL Server 2017 CU31+GDR KB5068402 (RDS version 14.00.3515.1.v1), SQL Server 2019 CU32+GDR KB5068404 (RDS version 15.00.4455.2.1.v1) and SQL Server 2022 CU22 KB5068450 (RDS version 16.00.4225.2.1.v1). The GDR updates address vulnerabilities described in CVE-2025-59499. For additional information on the improvements and fixes included in these updates, see Microsoft documentation for KB5068401, KB5068402, KB5068404, KB5068450. We recommend that you upgrade your Amazon RDS for SQL Server instances to apply these updates using Amazon RDS Management Console, or by using the AWS SDK or CLI. You can learn more about upgrading your database instance in the Amazon RDS SQL Server User Guide for upgrading your RDS Microsoft SQL Server DB engine.
Amazon Connect now provides agent scheduling metrics in data lake, making it easier for you to generate reports and insights from this data. For example, after publishing schedules for next month, you can access interval level (15 minutes or 30 minutes) metrics such as forecasted headcount, scheduled headcount, and projected service level in Connect analytics data lake. You can view aggregated metrics for an entire business unit (forecast group) or broken down by specific demand segments (demand groups). You can then visualize this data in Amazon Quick Sight or another BI tool of your choice for further analysis, such as identifying periods of over or under-staffing. This eliminates the need for manual reviews of agent schedules thus improving productivity for schedulers and supervisors.
This feature is available in all AWS Regions where Amazon Connect agent scheduling is available. To learn more about Amazon Connect agent scheduling, click here.
Amazon Connect now provides agent scheduling metrics in data lake, making it easier for you to generate reports and insights from this data. For example, after publishing schedules for next month, you can access interval level (15 minutes or 30 minutes) metrics such as forecasted headcount, scheduled headcount, and projected service level in Connect analytics data lake. You can view aggregated metrics for an entire business unit (forecast group) or broken down by specific demand segments (demand groups). You can then visualize this data in Amazon Quick Sight or another BI tool of your choice for further analysis, such as identifying periods of over or under-staffing. This eliminates the need for manual reviews of agent schedules thus improving productivity for schedulers and supervisors. This feature is available in all AWS Regions where Amazon Connect agent scheduling is available. To learn more about Amazon Connect agent scheduling, click here.
¿Puede la IA aprender el lenguaje de la biología para reinventar la medicina?
Por: Samantha Kubota, escritora de Microsoft.
Todos tenemos unos 20.000 genes en nuestro genoma. Aunque esta diversidad es lo que hace tan rica la experiencia humana, nuestras diferencias genéticas pueden complicar las cosas en lo que respecta a la medicina y al tratamiento de enfermedades.
Hoy en día, la mayoría de los tratamientos son un enfoque único para todos. Solo una pequeña fracción de los pacientes con cáncer, por ejemplo, recibe terapias dirigidas. Pero si la IA pudiera aprender a leer y escribir el lenguaje de la biología, podría ayudar a personalizar los tratamientos según la composición única de cada paciente.
Ava Amini, investigadora principal en Microsoft Research, trabaja para que eso suceda. De manera reciente habló sobre el potencial de la IA para la biología en una cervecería concurrida en Cambridge, Massachusetts, como parte de «Lectures on Tap«, una serie de eventos que combina conferencias de expertos con diversión interactiva en pubs informales de todo Estados Unidos.
Aquí tienen cinco de los conceptos que abordó, desde cómo funciona la medicina de precisión hasta la gran visión de desarrollar IA capaz de predecir cómo se comportan las células.
Cómo la IA puede ayudar a entender la biología
La biología es bastante compleja: la composición genética y el comportamiento celular de cada persona son únicos. Hoy en día, la medicina suele tratar a los pacientes basándose en promedios, no en diferencias individuales. Amini afirma que la IA ofrece una forma de descifrar patrones en enormes conjuntos de datos biológicos que los humanos no pueden procesar por sí solos.
«La computación nos da un conjunto de herramientas muy potente para entender lo que creo que es el sistema más complejo e intrincado que tenemos, que es el sistema y el lenguaje de la biología», afirma. «Tenemos esta oportunidad de construir sistemas computacionales, modelos de IA, que puedan aprovechar la escala de datos que generamos, aprender este lenguaje biológico y, en última instancia, poder utilizarlo para hacer nuevos descubrimientos, diseñar nuevos medicamentos y, con suerte, acercarnos a esa visión de empoderar a las personas para vivir un futuro más saludable.»
Amini afirma que una sola biopsia de cáncer, por ejemplo, puede generar casi 50 millones de datos individuales. La IA podría ayudar a filtrar estos datos enormes, encontrar patrones y permitir un tratamiento personalizado y preciso en lugar de una atención generalizada.
Cómo la medicina de precisión puede ayudar a las personas
La medicina de precisión tiene como objetivo adaptar los tratamientos a la composición genética, molecular y celular única de cada paciente. Pero la mayoría de los tratamientos son genéricos, y solo una pequeña fracción de los pacientes con cáncer recibe terapias dirigidas. Aún menos experimentan un éxito duradero, dice Amini.
«La verdad es que, basándose en las terapias dirigidas actuales, menos del 5% de esta población va a responder de forma efectiva», dice Amini sobre el tratamiento del cáncer. «Eso se debe a que hay cosas como la resistencia o que el cáncer evoluciona, se extiende y crece, y estos pacientes no verán resultados duraderos, duraderos y curativos.»
La medicina de precisión busca superar estas limitaciones al aprovechar la diversidad y heterogeneidad de enfermedades como el cáncer, para ir más allá de las medias poblacionales hacia una atención individualizada.
Utilizar el lenguaje de la biología para diseñar nuevas proteínas
En 1965, la biofísica estadounidense Margaret Dayhoff dio a la biología un alfabeto — un código de una letra para los 20 aminoácidos naturales, los bloques básicos de las proteínas. Su creación de este código para los aminoácidos permitió la representación de proteínas como lenguaje.
Microsoft construye sobre esta base con EvoDiff y El Atlas Dayhoff, modelos de IA generativa, para diseñar nuevas proteínas. Amini dice que el concepto es como Copilot para biología: introducir un prompt y sacar una proteína novedosa guiada por ese prompt.
Estos modelos pueden ser motivados en el lenguaje biológico para diseñar proteínas con funciones específicas.
Ava Amini, investigadora principal de Microsoft Research, habla sobre el potencial de la IA en biología en un evento Lectures on Tap.
Las proteínas diseñadas por IA muestran progreso y promesa
Proteínas diseñadas por IA podrían ayudar a dirigirse a las células cancerosas o unirse a receptores para la administración de fármacos, según Amini.
Ella afirma que los modelos EvoDiff y Dayhoff de Microsoft han generado proteínas probadas en laboratorio con resultados funcionales exitosos. Al aprender a mayor escala y diversidad de datos, los modelos de Dayhoff mejoraron la tasa de éxito en la producción de nuevas proteínas del 16% con métodos anteriores al 50%. Estos avances demuestran que la IA generativa para la biología no es solo teoría; esto ya ha comenzado a ocurrir.
«De hecho, hemos medido y probado en el laboratorio del mundo real para demostrar que estas proteínas tienen las funciones que queríamos y que queríamos tener», dice Amini.
Sin embargo, la calidad y diversidad de los datos todavía son críticas para el rendimiento del modelo, y aún existen limitaciones significativas, en especial en el modelado de células completas.
Trabajo para modelar células humanas
Un modelo de IA diseñado para simular la complejidad de una célula humana mediante el aprendizaje de patrones en datos biológicos podría predecir cómo responden las células a los fármacos, para desbloquear la medicina de precisión. Muchos lo consideran un «santo grial» en la ciencia, dice Amini, y han seguido la idea de construir modelos de IA para predecir cómo se comportan las células. Amini dice que sus experimentos en Microsoft han demostrado que los modelos existentes de IA de células a menudo predicen solo valores medios, en lugar de diferencias biológicas reales. Aumentar el volumen de datos no mejora el rendimiento: los modelos se saturan con rapidez y no escalan como se espera. Estudios críticos recientes, incluidos los de Amini y su equipo, han puesto de manifiesto estas limitaciones.
Amini aún tiene esperanza. Aunque la promesa de la IA en biología es inmensa, afirma, darse cuenta de que la medicina personalizada y precisa requerirá una integración y colaboración continuas entre disciplinas. Co-dirige Project Ex Vivo, una colaboración de investigación entre Microsoft y el Broad Institute con el apoyo del Dana-Farber Cancer Institute, que construye un nuevo marco para la oncología de precisión, donde se integran experimentación y computación desde cero hacia el objetivo final de mejorar los resultados para los pacientes.
«Como tecnóloga, usamos estos hallazgos como combustible y queremos aprovechar todo lo posible para en verdad avanzar», dice. «Y toda esta información, todas estas evaluaciones, nos ayudan a mejorar y acercarnos a esa promesa.»
Imagen principal de Andriy Onufriyenko / Moment / Getty Images.
The GDR updates address vulnerabilities described in CVE-2025-59499. For additional information on the improvements and fixes included in these updates, see Microsoft documentation for KB5068404 and KB5068406. We recommend that you upgrade your Amazon RDS Custom for SQL Server instances to apply these updates using Amazon RDS Management Console, or by using the AWS SDK or CLI. You can learn more about upgrading your database instance in the Amazon RDS Custom User Guide.
Amazon Relational Database Service (Amazon RDS) Custom for SQL Server now supports the latest General Distribution Release (GDR) updates for Microsoft SQL Server. This release includes support for SQL Server 2019 CU32+GDR KB5068404 (RDS version 15.00.4455.2.1.v1) and SQL Server 2022 CU21+GDR KB5068406 (RDS version 16.00.4222.2.1.v1). The GDR updates address vulnerabilities described in CVE-2025-59499. For additional information on the improvements and fixes included in these updates, see Microsoft documentation for KB5068404 and KB5068406. We recommend that you upgrade your Amazon RDS Custom for SQL Server instances to apply these updates using Amazon RDS Management Console, or by using the AWS SDK or CLI. You can learn more about upgrading your database instance in the Amazon RDS Custom User Guide.
Amazon Redshift Serverless introduces queue-based query resource management. You can create dedicated query queues with customized monitoring rules for different workloads. This feature provides granular control over resource usage. Queues let you set metrics-based predicates and automated responses. For example, you can configure rules to automatically abort queries that exceed time limits or consume too many resources.
Previously, Query Monitoring Rules (QMR) were applied only at the Redshift Serverless workgroup level, affecting all queries run in this workgroup uniformly. The new queue-based approach lets you create queues with distinct monitoring rules. You can assign these queues to specific user roles and query groups. Each queue operates independently, with rules affecting only the queries within that queue. The available monitoring metrics can be found in Query monitoring metrics for Amazon Redshift Serverless.
This feature is available in all AWS regions that support Amazon Redshift Serverless. You can manage QMR with queues through the AWS Console and Redshift APIs.
For implementation details, see the documentation in the Amazon Redshift management guide.
Amazon Redshift Serverless introduces queue-based query resource management. You can create dedicated query queues with customized monitoring rules for different workloads. This feature provides granular control over resource usage. Queues let you set metrics-based predicates and automated responses. For example, you can configure rules to automatically abort queries that exceed time limits or consume too many resources. Previously, Query Monitoring Rules (QMR) were applied only at the Redshift Serverless workgroup level, affecting all queries run in this workgroup uniformly. The new queue-based approach lets you create queues with distinct monitoring rules. You can assign these queues to specific user roles and query groups. Each queue operates independently, with rules affecting only the queries within that queue. The available monitoring metrics can be found in Query monitoring metrics for Amazon Redshift Serverless. This feature is available in all AWS regions that support Amazon Redshift Serverless. You can manage QMR with queues through the AWS Console and Redshift APIs. For implementation details, see the documentation in the Amazon Redshift management guide.
AWS Transform custom now supports AWS PrivateLink and is available in a new AWS Region, Europe (Frankfurt), in addition to the US East (N. Virginia) Region. AWS Transform custom helps organizations reduce technical debt by automating repetitive transformation tasks such as language version upgrades, API migrations, and framework updates. The agent is designed for enterprise development teams and consulting partners who need to execute consistent, repeatable code transformations across large codebases.
With AWS Transform custom, teams can create custom transformation definitions using natural language, documentation, and code samples, or use AWS-managed transformations for common scenarios including Java, Python, and Node.js version upgrades. Through continual learning, the service improves transformation quality with every execution over time. With AWS PrivateLink support, customers can now access AWS Transform custom from their Amazon VPC without routing traffic over the public internet, helping meet security and compliance requirements.
AWS Transform custom now supports AWS PrivateLink and is available in a new AWS Region, Europe (Frankfurt), in addition to the US East (N. Virginia) Region. AWS Transform custom helps organizations reduce technical debt by automating repetitive transformation tasks such as language version upgrades, API migrations, and framework updates. The agent is designed for enterprise development teams and consulting partners who need to execute consistent, repeatable code transformations across large codebases. With AWS Transform custom, teams can create custom transformation definitions using natural language, documentation, and code samples, or use AWS-managed transformations for common scenarios including Java, Python, and Node.js version upgrades. Through continual learning, the service improves transformation quality with every execution over time. With AWS PrivateLink support, customers can now access AWS Transform custom from their Amazon VPC without routing traffic over the public internet, helping meet security and compliance requirements. To learn more about AWS Transform custom, visit the product page and user guide.
