Publicado el Deja un comentario

Announcing new Amazon EC2 M9g instances powered by AWS Graviton5 processors (Preview)

Starting today, new general purpose Amazon Elastic Compute Cloud (Amazon EC2) M9g instances, powered by AWS Graviton5 processors, are available in preview. AWS Graviton5 is the latest in the Graviton family of processors that are custom designed by AWS to provide the best price performance for workloads in Amazon EC2. These instances offer up to 25% better compute performance, and higher networking and Amazon Elastic Block Store (Amazon EBS) bandwidth than AWS Graviton4-based M8g instances. They are up to 30% faster for databases, up to 35% faster web applications, and up to 35% faster for machine learning workloads compared to M8g.

M9g instances are built on the AWS Nitro System, a collection of hardware and software innovations designed by AWS. The AWS Nitro System enables the delivery of efficient, flexible, and secure cloud services with isolated multitenancy, private networking, and fast local storage. Amazon EC2 M9g instances are ideal for workloads such as application servers, microservices, gaming servers, midsize data stores, and caching fleets.

To learn more or request access to the M9g preview, see Amazon EC2 M9g instances. To begin your Graviton journey, visit the Level up your compute with AWS Graviton page.

 

​Starting today, new general purpose Amazon Elastic Compute Cloud (Amazon EC2) M9g instances, powered by AWS Graviton5 processors, are available in preview. AWS Graviton5 is the latest in the Graviton family of processors that are custom designed by AWS to provide the best price performance for workloads in Amazon EC2. These instances offer up to 25% better compute performance, and higher networking and Amazon Elastic Block Store (Amazon EBS) bandwidth than AWS Graviton4-based M8g instances. They are up to 30% faster for databases, up to 35% faster web applications, and up to 35% faster for machine learning workloads compared to M8g. M9g instances are built on the AWS Nitro System, a collection of hardware and software innovations designed by AWS. The AWS Nitro System enables the delivery of efficient, flexible, and secure cloud services with isolated multitenancy, private networking, and fast local storage. Amazon EC2 M9g instances are ideal for workloads such as application servers, microservices, gaming servers, midsize data stores, and caching fleets. To learn more or request access to the M9g preview, see Amazon EC2 M9g instances. To begin your Graviton journey, visit the Level up your compute with AWS Graviton page.  

Publicado el Deja un comentario

El imperativo del CISO: Construir resiliencia en una era de ciberamenazas aceleradas

El imperativo del CISO: Construir resiliencia en una era de ciberamenazas aceleradas

Mujer con camisa roja de manga larga y una cadena dorada, trabaja en una laptop

Por: Ann Johnson, vicepresidenta corporativa, subdirectora de CISO, Oficina de Gestión de Seguridad del Cliente

El más reciente Informe de Defensa Digital de Microsoft 2025 dibuja un vívido panorama de un panorama de ciberamenazas en constante evolución. El aumento de ciberataques motivados por temas financieros y el riesgo persistente de actores estatales exige atención urgente. Pero para quienes estamos en la Oficina del Director de Seguridad de la Información (CISO, por sus siglas en inglés), el verdadero desafío y la oportunidad residen en cómo responden, se adaptan y construyen resiliencia las organizaciones para lo que viene después.

Los hallazgos de este año revelan algo que todos hemos percibido: la amenaza del paisaje no solo está en evolución, sino que también se acelera. La IA ha cambiado de manera fundamental la ecuación y afecta la velocidad, escala y sofisticación de los ciberataques de formas que hacen obsoletas muchas suposiciones defensivas tradicionales. Sin embargo, la IA también representa nuestra herramienta más poderosa para la adaptación.

Descubran las últimas novedades del Informe de Defensa Digital de Microsoft 2025

Comprensión de la aceleración

Las métricas cuentan una historia contundente, pero las implicaciones operativas importan más. Hemos observado ciberataques que se ejecutan en el tiempo que tarda un usuario en hacer clic—técnicas ClickFix que eluden defensas en capas mediante ingeniería social a velocidad de máquina. En entornos de nube, la ventana entre el despliegue y el compromiso se ha reducido a 48 horas para contenedores, lo que desafía de manera fundamental nuestras suposiciones sobre el endurecimiento de los plazos.

La economía también ha cambiado. Las campañas de phishing impulsadas por IA ahora logran mejoras de rentabilidad 50 veces superiores al automatizar la personalización a gran escala. Seguimos operaciones norcoreanas que han incorporado a decenas de miles de trabajadores en todo el mundo, lo que convierte la fuerza laboral remota en un vector persistente de ciberamenazas. Esto no es oportunista. De hecho, es una infiltración a escala industrial.

La curva de sofisticación continúa su ascenso pronunciado. Nuestra telemetría muestra un aumento del 87% en campañas disruptivas dirigidas a entornos Microsoft Azure. Los intentos de robo de credenciales han subido un 23%, la exfiltración de datos un 58%. Ahora seguimos los primeros indicadores de malware autónomo capaz de moverse de manera lateral y de comportamiento adaptativo sin necesidad de dirección humana.

Lo que más me llama la atención es la coordinación operativa. A través de Microsoft Threat Intelligence, observamos campañas que abarcan más de 130 países donde estados-nación, sindicatos criminales y mercenarios comerciales comparten infraestructuras y tácticas. Los corredores de acceso han creado mercados que difuminan las líneas entre espionaje y crimen. Los modelos: escalables, resistentes y alarmantemente eficientes.

Desde la conciencia de amenazas hasta la acción estratégica

Aquí está la paradoja a la que se enfrenta todo CISO: las amenazas se aceleran, pero nuestras capacidades defensivas nunca han sido tan fuertes. La brecha no es la tecnología. La brecha está en cómo pensamos y operacionalizamos la seguridad. Enfoques heredados que separan la seguridad de la estrategia empresarial, que priorizan la prevención sobre la resiliencia, que tratan los incidentes de amenaza como fracasos en lugar de eventos inevitables—estas mentalidades ahora son una carga.

El camino a seguir requiere cambios fundamentales:

La seguridad como facilitador empresarial, no como punto de control. Simplemente integramos la seguridad en todos los procesos de negocio, desde el desarrollo de productos hasta la gestión de la cadena de suministro. Cuando la seguridad se convierte en parte integral del funcionamiento de las organizaciones, en lugar de una puerta por la que deben pasar, avanzamos más rápido mientras gestionamos el riesgo de forma más eficaz. Esto no va de bajar los estándares. Se trata de construir seguridad en los cimientos en lugar de añadirla como una fachada.

La resiliencia como objetivo principal. La cuestión no es si ocurrirá un incidente, sino qué tan rápido podremos detectarlo, contenerlo y recuperarnos de él. Cuando los ciberataques se ejecutan en segundos y los compromisos ocurren en 48 horas, nuestras capacidades de respuesta deben igualar esa velocidad. Esto significa manuales probados, equipos empoderados y mecanismos de respuesta automatizados que funcionan a velocidad de máquina.

Inteligencia y automatización como multiplicadores de fuerza. Las mismas tecnologías de IA que permiten a los ciberatacantes escalar operaciones pueden amplificar nuestras capacidades de defensa, si las desplegamos de manera estratégica. La automatización no consiste en reemplazar equipos de seguridad. Se trata de permitir que operen a la velocidad y escala que exigen las amenazas modernas.

El mandato evolucionado del CISO

El papel del CISO se ha ampliado de manera fundamental. Ya no somos tan solo tecnólogos. Somos gestores de riesgos, asesores estratégicos y agentes de cambio organizacional. El consejo necesita que traduzcamos las ciberamenazas técnicas en riesgos empresariales y las estrategias de resiliencia en ventajas competitivas.

Esta evolución exige nuevas capacidades:

Liderazgo transversal que trasciende la informática. Cuando un ataque de ingeniería social puede comprometer una organización en segundos, la respuesta requiere acciones coordinadas entre TI, legal, recursos humanos, comunicación y liderazgo ejecutivo. Debemos construir estas alianzas antes de la crisis, no durante ella.

Adaptación continua como disciplina operativa. La ventana de compromiso de contenedores de 48 horas y los vectores de infección instantáneos que observamos hacen que la monitorización continua, las pruebas regulares y la iteración rápida no sean buenas prácticas. Son requisitos de supervivencia. Nuestras defensas, políticas y capacidades de respuesta deben evolucionar tan rápido como las amenazas.

Gobernanza que anticipa la evolución regulatoria. A medida que los gobiernos aumentan los requisitos de transparencia e imponen consecuencias por actividades maliciosas, debemos asegurarnos de que nuestras organizaciones puedan cumplir tanto la letra como el espíritu de las normativas emergentes. Esto incluye comprender los riesgos de terceros, desde intermediarios de acceso hasta ciberamenazas integradas en nuestra plantilla y cadenas de suministro.

Estrategias probadas para operacionalizar la resiliencia en seguridad

De nuestro trabajo con los clientes, nuestra propia experiencia operativa y la implementación de la Iniciativa Futuro Seguro (SFI, por sus siglas en inglés), destacan tres prioridades:

Los controles de identidad modernos son innegociables. Con el 97% de los ataques de identidad dirigidos a contraseñas, la MFA resistente al phishing altera de manera fundamental la ecuación de riesgo. Esto no va de añadir capas, sino de eliminar vectores de ataque completos. Las organizaciones que implementan autenticación resistente al phishing ven reducciones drásticas en compromisos exitosos.

La preparación para la respuesta a incidentes determina el resultado. Cuando los ataques se mueven a velocidad de máquina, el tiempo de respuesta se convierte en la variable crítica. Esto significa simulaciones regulares, libros de jugadas probados y equipos capacitados para actuar con decisión. Debemos practicar para los escenarios que nos enfrentaremos, no para los que esperamos evitar. Las organizaciones que se recuperan más rápido son aquellas que han fallado en simulación y han aprendido antes del evento real.

La defensa colectiva ya no es opcional. Frente a campañas que abarcan más de 130 países y ecosistemas de ciberatacantes que comparten infraestructura, la defensa aislada es ineficaz. El intercambio de inteligencia, las mejores prácticas colaborativas y la coordinación sectorial son multiplicadores de fuerza que benefician a todos. Las ciberamenazas a las que nos enfrentamos son demasiado sofisticadas y demasiado coordinadas para que cualquier organización las defienda sola.

Hemos aplicado estos mismos principios a nivel interno a través de nuestra Iniciativa Futuro Seguro. En lugar de mantener nuestras lecciones de implementación internas, publicamos los patrones y prácticas reales que hemos utilizado: los enfoques específicos que funcionaron, los compromisos que encontramos y los pasos prácticos que otras organizaciones pueden adaptar. La biblioteca de patrones y prácticas SFI incluye directrices detalladas sobre desafíos como la seguridad de entornos multi-inquilino, la protección de cadenas de suministro de software e implementación de Zero Trust (Confianza Cero) para el acceso al código fuente.

Lo que aprecio de estos patrones es que están escritos por profesionales que en verdad los han implementado. Cada uno describe el problema, explica cómo lo resolvimos a nivel interno en Microsoft y ofrece recomendaciones que ustedes pueden evaluar para su propio entorno. No hay resúmenes brillantes, solo los detalles operativos de lo que funcionó y lo que no.

Descubran más con la Iniciativa Futuro Seguro

Pasos para fortalecer la resiliencia y la respuesta en toda su organización

La aceleración que presenciamos—velocidad de ciberataque, escala operativa y sofisticación técnica—exige una aceleración equivalente en nuestra respuesta. No se trata de trabajar más duro; se trata de trabajar de forma diferente. Significa tratar la IA y la automatización como imperativos operativos, no como proyectos futuros. Significa construir la seguridad de identidad como infraestructura fundamental, no como una casilla de cumplimiento. Significa desarrollar capacidades de respuesta a incidentes que igualen la velocidad de los ciberataques modernos.

Lo más fundamental, significa abrazar nuestro papel evolucionado como CISOs. Somos arquitectos de la resiliencia organizativa en una era en la que las ciberamenazas se mueven a velocidad de máquina y abarcan continentes. Esto requiere partes iguales de profundidad técnica, visión estratégica y liderazgo colaborativo.

El panorama de las ciberamenazas seguirá su evolución. Nuestro mandato es evolucionar más rápido, construir organizaciones que no solo sean seguras, sino también resilientes, adaptables y preparadas para lo que venga. Ese es el reto al que se enfrenta todo CISO hoy en día. También es la oportunidad de construir algo más fuerte que lo que vino antes.

Para un análisis detallado y exhaustivo, consulten el Informe completo de Defensa Digital de Microsoft 2025.

Descubran más con Microsoft Security

Para saber más sobre las soluciones de seguridad de Microsoft, visiten nuestra página web. Guarden el blog de Seguridad en sus Favoritos para seguir nuestra cobertura experta en temas de seguridad. Además, síganos en LinkedIn (Microsoft Security) y X (@MSFTSecurity) para las últimas noticias y actualizaciones sobre ciberseguridad.

The post El imperativo del CISO: Construir resiliencia en una era de ciberamenazas aceleradas appeared first on Source LATAM.

 

​The post El imperativo del CISO: Construir resiliencia en una era de ciberamenazas aceleradas appeared first on Source LATAM.  

Publicado el Deja un comentario

Amazon SageMaker HyperPod now supports checkpointless training

Amazon SageMaker HyperPod now supports checkpointless training, a new foundational model training capability that mitigates the need for a checkpoint-based job-level restart for fault recovery. Checkpointless training maintains forward training momentum despite failures, reducing recovery time from hours to minutes. This represents a fundamental shift from traditional checkpoint-based recovery, where failures require pausing the entire training cluster, diagnosing issues manually, and restoring from saved checkpoints, a process that can leave expensive AI accelerators idle for hours, costing your organization wasted compute.

Checkpointless training transforms this paradigm by preserving the model training state across the distributed cluster, automatically swapping out faulty training nodes on the fly and using peer-to-peer state transfer from healthy accelerators for failure recovery. By mitigating checkpoint dependencies during recovery, checkpointless training can help your organization save on idle AI accelerator costs and accelerate time. Even at larger scales, checkpointless training on Amazon SageMaker HyperPod enables upwards of 95% training goodput on cluster sizes with thousands of AI accelerators.

Checkpointless training on SageMaker HyperPod is available in all AWS Regions where Amazon SageMaker HyperPod is currently available. You can enable checkpointless training with zero code changes using HyperPod recipes for popular publicly available models such as Llama and GPT OSS. For custom model architectures, you can integrate checkpointless training components with minimal modifications for PyTorch-based workflows, making it accessible to your teams regardless of their distributed training expertise.

To get started, visit the Amazon SageMaker HyperPod product page and see the checkpointless training GitHub page for implementation guidance.

 

​Amazon SageMaker HyperPod now supports checkpointless training, a new foundational model training capability that mitigates the need for a checkpoint-based job-level restart for fault recovery. Checkpointless training maintains forward training momentum despite failures, reducing recovery time from hours to minutes. This represents a fundamental shift from traditional checkpoint-based recovery, where failures require pausing the entire training cluster, diagnosing issues manually, and restoring from saved checkpoints, a process that can leave expensive AI accelerators idle for hours, costing your organization wasted compute.
Checkpointless training transforms this paradigm by preserving the model training state across the distributed cluster, automatically swapping out faulty training nodes on the fly and using peer-to-peer state transfer from healthy accelerators for failure recovery. By mitigating checkpoint dependencies during recovery, checkpointless training can help your organization save on idle AI accelerator costs and accelerate time. Even at larger scales, checkpointless training on Amazon SageMaker HyperPod enables upwards of 95% training goodput on cluster sizes with thousands of AI accelerators.
Checkpointless training on SageMaker HyperPod is available in all AWS Regions where Amazon SageMaker HyperPod is currently available. You can enable checkpointless training with zero code changes using HyperPod recipes for popular publicly available models such as Llama and GPT OSS. For custom model architectures, you can integrate checkpointless training components with minimal modifications for PyTorch-based workflows, making it accessible to your teams regardless of their distributed training expertise.
To get started, visit the Amazon SageMaker HyperPod product page and see the checkpointless training GitHub page for implementation guidance.  

Publicado el Deja un comentario

Announcing TypeScript support in Strands Agents (preview) and more

In May, we open sourced the Strands Agents SDK, an open source python framework that takes a model-driven approach to building and running AI agents in just a few lines of code. Today, we’re announcing that TypeScript support is available in preview. Now, developers can choose between Python and TypeScript for building Strands Agents.

TypeScript support in Strands has been designed to provide an idiomatic TypeScript experience with full type safety, async/await support, and modern JavaScript/TypeScript patterns. Strands can be easily run in client applications, in browsers, and server-side applications in runtimes like AWS Lambda and Bedrock AgentCore. Developers can also build their entire stack in Typescript using the AWS CDK.

We’re also announcing three additional updates for the Strands SDK. First, edge device support for Strands Agents is generally available, extending the SDK with bidirectional streaming and additional local model providers like llama.cpp that let you run agents on small-scale devices using local models. Second, Strands steering is now available as an experimental feature, giving developers a modular prompting mechanism that provides feedback to the agent at the right moment in its lifecycle, steering agents toward a desired outcome without rigid workflows. Finally, Strands evaluations is available in preview. Evaluations gives developers the ability to systematically validate agent behavior, measure improvements, and deploy with confidence during development cycles.

Head to the Strands Agents GitHub to get started building.

 

​In May, we open sourced the Strands Agents SDK, an open source python framework that takes a model-driven approach to building and running AI agents in just a few lines of code. Today, we’re announcing that TypeScript support is available in preview. Now, developers can choose between Python and TypeScript for building Strands Agents. TypeScript support in Strands has been designed to provide an idiomatic TypeScript experience with full type safety, async/await support, and modern JavaScript/TypeScript patterns. Strands can be easily run in client applications, in browsers, and server-side applications in runtimes like AWS Lambda and Bedrock AgentCore. Developers can also build their entire stack in Typescript using the AWS CDK. We’re also announcing three additional updates for the Strands SDK. First, edge device support for Strands Agents is generally available, extending the SDK with bidirectional streaming and additional local model providers like llama.cpp that let you run agents on small-scale devices using local models. Second, Strands steering is now available as an experimental feature, giving developers a modular prompting mechanism that provides feedback to the agent at the right moment in its lifecycle, steering agents toward a desired outcome without rigid workflows. Finally, Strands evaluations is available in preview. Evaluations gives developers the ability to systematically validate agent behavior, measure improvements, and deploy with confidence during development cycles. Head to the Strands Agents GitHub to get started building.  

Publicado el Deja un comentario

New serverless model customization capability in Amazon SageMaker AI

Amazon Web Services (AWS) announces a new serverless model customization capability that empowers AI developers to quickly customize popular models with supervised fine-tuning and the latest techniques like reinforcement learning. Amazon SageMaker AI is a fully managed service that brings together a broad set of tools to enable high-performance, low-cost AI model development for any use case. 

Many AI developers seek to customize models with proprietary data for improved accuracy, but this often requires lengthy iteration cycles. For example, AI developers must define a use case and prepare data, select a model and customization technique, train the model, then evaluate the model for deployment. Now AI developers can simplify the end-to-end model customization workflow, from data preparation to evaluation and deployment, and accelerate the process. With an easy-to-use interface, AI developers can quickly get started and customize popular models, including Amazon Nova, Llama, Qwen, DeepSeek, and GPT-OSS, with their own data. They can use supervised fine-tuning and the latest customization techniques such as reinforcement learning and direct preference optimization. In addition, AI developers can use the AI agent-guided workflow (in preview), and use natural language to generate synthetic data, analyze data quality, and handle model training and evaluation—all entirely serverless. 

You can use this easy-to-use interface in the following AWS Regions: Europe (Ireland), US East (N. Virginia), Asia Pacific (Tokyo), and US West (Oregon). To join the waitlist to access the AI agent-guided workflow, visit the sign-up page

To learn more, visit the SageMaker AI model customization page and blog.

 

​Amazon Web Services (AWS) announces a new serverless model customization capability that empowers AI developers to quickly customize popular models with supervised fine-tuning and the latest techniques like reinforcement learning. Amazon SageMaker AI is a fully managed service that brings together a broad set of tools to enable high-performance, low-cost AI model development for any use case. 
Many AI developers seek to customize models with proprietary data for improved accuracy, but this often requires lengthy iteration cycles. For example, AI developers must define a use case and prepare data, select a model and customization technique, train the model, then evaluate the model for deployment. Now AI developers can simplify the end-to-end model customization workflow, from data preparation to evaluation and deployment, and accelerate the process. With an easy-to-use interface, AI developers can quickly get started and customize popular models, including Amazon Nova, Llama, Qwen, DeepSeek, and GPT-OSS, with their own data. They can use supervised fine-tuning and the latest customization techniques such as reinforcement learning and direct preference optimization. In addition, AI developers can use the AI agent-guided workflow (in preview), and use natural language to generate synthetic data, analyze data quality, and handle model training and evaluation—all entirely serverless. 
You can use this easy-to-use interface in the following AWS Regions: Europe (Ireland), US East (N. Virginia), Asia Pacific (Tokyo), and US West (Oregon). To join the waitlist to access the AI agent-guided workflow, visit the sign-up page. 
To learn more, visit the SageMaker AI model customization page and blog.  

Publicado el Deja un comentario

Amazon Bedrock now supports reinforcement fine-tuning delivering 66% accuracy gains on average over base models

Amazon Bedrock now supports reinforcement fine-tuning, helping you improve model accuracy without needing deep machine learning expertise or large sums of labeled data. Amazon Bedrock automates the reinforcement fine-tuning workflow, making this advanced model customization technique accessible to everyday developers. Models learn to align with your specific requirements using a small set of prompts rather than the large sums of data needed for traditional fine-tuning methods, enabling teams to get started quickly. This capability teaches models through feedback on multiple possible responses to the same prompt, improving their judgement of what makes a good response. Reinforcement fine-tuning in Amazon Bedrock delivers 66% accuracy gains on average over base models so you can use smaller, faster, and more cost-effective model variants while maintaining high quality.

Organizations struggle to adapt AI models to their unique business needs, forcing them to choose between generic models with average performance or expensive, complex customization that requires specialized talent, infrastructure, and risky data movement. Reinforcement fine-tuning in Amazon Bedrock removes this complexity by making advanced model customization fast, automated, and secure. You can train models by uploading training data directly from your computer or choose from datasets already stored in Amazon S3, eliminating the need for any labeled datasets. You can define reward functions using verifiable rule-based graders or AI-based judges along with built-in templates to optimize your models for both objective tasks such as code generation or math reasoning, and subjective tasks such as instruction following or chatbot interactions. Your proprietary data never leaves AWS’s secure, governed environment during the entire customization process, mitigating security and compliance concerns.

You can get started with reinforcement fine-tuning in Amazon Bedrock through the Amazon Bedrock console and via the Amazon Bedrock APIs. At launch, you can use reinforcement fine-tuning with Amazon Nova 2 Lite with support for additional models coming soon. To learn more about reinforcement fine-tuning in Amazon Bedrock, read the launch blog, pricing page, and documentation.

 

​Amazon Bedrock now supports reinforcement fine-tuning, helping you improve model accuracy without needing deep machine learning expertise or large sums of labeled data. Amazon Bedrock automates the reinforcement fine-tuning workflow, making this advanced model customization technique accessible to everyday developers. Models learn to align with your specific requirements using a small set of prompts rather than the large sums of data needed for traditional fine-tuning methods, enabling teams to get started quickly. This capability teaches models through feedback on multiple possible responses to the same prompt, improving their judgement of what makes a good response. Reinforcement fine-tuning in Amazon Bedrock delivers 66% accuracy gains on average over base models so you can use smaller, faster, and more cost-effective model variants while maintaining high quality.
Organizations struggle to adapt AI models to their unique business needs, forcing them to choose between generic models with average performance or expensive, complex customization that requires specialized talent, infrastructure, and risky data movement. Reinforcement fine-tuning in Amazon Bedrock removes this complexity by making advanced model customization fast, automated, and secure. You can train models by uploading training data directly from your computer or choose from datasets already stored in Amazon S3, eliminating the need for any labeled datasets. You can define reward functions using verifiable rule-based graders or AI-based judges along with built-in templates to optimize your models for both objective tasks such as code generation or math reasoning, and subjective tasks such as instruction following or chatbot interactions. Your proprietary data never leaves AWS’s secure, governed environment during the entire customization process, mitigating security and compliance concerns.
You can get started with reinforcement fine-tuning in Amazon Bedrock through the Amazon Bedrock console and via the Amazon Bedrock APIs. At launch, you can use reinforcement fine-tuning with Amazon Nova 2 Lite with support for additional models coming soon. To learn more about reinforcement fine-tuning in Amazon Bedrock, read the launch blog, pricing page, and documentation.  

Publicado el Deja un comentario

Introducing elastic training on Amazon SageMaker HyperPod

Amazon SageMaker HyperPod now supports elastic training, enabling organizations to accelerate foundation model training by automatically scaling training workloads based on resource availability and workload priorities. This represents a fundamental shift from training with a fixed set of resources, as it saves hours of engineering time spent reconfiguring training jobs based on compute availability.

Any change in compute availability previously required manually halting training, reconfiguring training parameters, and restarting jobs—a process that requires distributed training expertise and leaves expensive AI accelerators sitting idle during training job reconfiguration. Elastic training automatically expands training jobs to absorb idle AI accelerators and seamlessly contracting when higher-priority workloads need resources—all without halting training entirely.

By eliminating manual reconfiguration overhead and ensuring continuous utilization of available compute, elastic training can help save time previously spent on infrastructure management, reduce costs by maximizing cluster utilization, and accelerate time-to-market. Training can start immediately with minimal resources and grow opportunistically as capacity becomes available.

SageMaker HyperPod is available in all regions where Amazon SageMaker HyperPod is currently available. Organizations can enable elastic training with zero code changes using HyperPod recipes for publicly available models including Llama and GPT OSS. For custom model architectures, customers can integrate elastic training capabilities through lightweight configuration updates and minimal code modifications, making it accessible to teams without requiring distributed systems expertise.

To get started, visit the Amazon SageMaker HyperPod product page and see the elastic training documentation for implementation guidance.

 

​Amazon SageMaker HyperPod now supports elastic training, enabling organizations to accelerate foundation model training by automatically scaling training workloads based on resource availability and workload priorities. This represents a fundamental shift from training with a fixed set of resources, as it saves hours of engineering time spent reconfiguring training jobs based on compute availability.
Any change in compute availability previously required manually halting training, reconfiguring training parameters, and restarting jobs—a process that requires distributed training expertise and leaves expensive AI accelerators sitting idle during training job reconfiguration. Elastic training automatically expands training jobs to absorb idle AI accelerators and seamlessly contracting when higher-priority workloads need resources—all without halting training entirely.
By eliminating manual reconfiguration overhead and ensuring continuous utilization of available compute, elastic training can help save time previously spent on infrastructure management, reduce costs by maximizing cluster utilization, and accelerate time-to-market. Training can start immediately with minimal resources and grow opportunistically as capacity becomes available.
SageMaker HyperPod is available in all regions where Amazon SageMaker HyperPod is currently available. Organizations can enable elastic training with zero code changes using HyperPod recipes for publicly available models including Llama and GPT OSS. For custom model architectures, customers can integrate elastic training capabilities through lightweight configuration updates and minimal code modifications, making it accessible to teams without requiring distributed systems expertise.
To get started, visit the Amazon SageMaker HyperPod product page and see the elastic training documentation for implementation guidance.  

Publicado el Deja un comentario

Announcing Amazon EC2 General purpose M8azn instances (Preview)

Starting today, new general purpose high-frequency high-network Amazon Elastic Compute Cloud (Amazon EC2) M8azn instances are available for preview. These instances are powered by fifth generation AMD EPYC (formerly code named Turin) processors, offering the highest maximum CPU frequency, 5GHz in the cloud. The M8azn instances offer up to 2x compute performance versus previous generation M5zn instances. These instances also deliver 24% higher performance than M8a instances.

M8azn instances are built on the AWS Nitro System, a collection of hardware and software innovations designed by AWS. The AWS Nitro System enables the delivery of efficient, flexible, and secure cloud services with isolated multitenancy, private networking, and fast local storage. These instances are ideal for applications such as gaming, high-performance computing, high-frequency trading (HFT), CI/CD, and simulation modeling for the automotive, aerospace, energy, and telecommunication industries.

To learn more or request access to the M8azn instances preview, visit the Amazon EC2 M8a page.

 

​Starting today, new general purpose high-frequency high-network Amazon Elastic Compute Cloud (Amazon EC2) M8azn instances are available for preview. These instances are powered by fifth generation AMD EPYC (formerly code named Turin) processors, offering the highest maximum CPU frequency, 5GHz in the cloud. The M8azn instances offer up to 2x compute performance versus previous generation M5zn instances. These instances also deliver 24% higher performance than M8a instances. M8azn instances are built on the AWS Nitro System, a collection of hardware and software innovations designed by AWS. The AWS Nitro System enables the delivery of efficient, flexible, and secure cloud services with isolated multitenancy, private networking, and fast local storage. These instances are ideal for applications such as gaming, high-performance computing, high-frequency trading (HFT), CI/CD, and simulation modeling for the automotive, aerospace, energy, and telecommunication industries. To learn more or request access to the M8azn instances preview, visit the Amazon EC2 M8a page.  

Publicado el Deja un comentario

Announcing the Apache Spark upgrade agent for Amazon EMR

AWS announces the Apache Spark upgrade agent, a new capability that accelerates Apache Spark version upgrades for Amazon EMR on EC2 and EMR Serverless. The agent converts complex upgrade processes that typically take months into projects spanning weeks through automated code analysis and transformation. Organizations invest substantial engineering resources analyzing API changes, resolving conflicts, and validating applications during Spark upgrades. The agent introduces conversational interfaces where engineers express upgrade requirements in natural language, while maintaining full control over code modifications.

The Apache Spark upgrade agent automatically identifies API changes and behavioral modifications across PySpark and Scala applications. Engineers can initiate upgrades directly from SageMaker Unified Studio, Kiro CLI or IDE of their choice with the help of MCP (Model Context Protocol) compatibility. During the upgrade process, the agent analyzes existing code and suggests specific changes, and engineers can review and approve before implementation. The agent validates functional correctness through data quality validations. The agent currently supports upgrades from Spark 2.4 to 3.5 and maintains data processing accuracy throughout the upgrade process.

The Apache Spark upgrade agent is now available in all AWS Regions where SageMaker Unified Studio is available. To start using the agent, visit SageMaker Unified Studio and select IDE Spaces or install the Kiro CLI. For detailed implementation guidance, reference documentation, and migration examples, visit the documentation.

 

​AWS announces the Apache Spark upgrade agent, a new capability that accelerates Apache Spark version upgrades for Amazon EMR on EC2 and EMR Serverless. The agent converts complex upgrade processes that typically take months into projects spanning weeks through automated code analysis and transformation. Organizations invest substantial engineering resources analyzing API changes, resolving conflicts, and validating applications during Spark upgrades. The agent introduces conversational interfaces where engineers express upgrade requirements in natural language, while maintaining full control over code modifications. The Apache Spark upgrade agent automatically identifies API changes and behavioral modifications across PySpark and Scala applications. Engineers can initiate upgrades directly from SageMaker Unified Studio, Kiro CLI or IDE of their choice with the help of MCP (Model Context Protocol) compatibility. During the upgrade process, the agent analyzes existing code and suggests specific changes, and engineers can review and approve before implementation. The agent validates functional correctness through data quality validations. The agent currently supports upgrades from Spark 2.4 to 3.5 and maintains data processing accuracy throughout the upgrade process. The Apache Spark upgrade agent is now available in all AWS Regions where SageMaker Unified Studio is available. To start using the agent, visit SageMaker Unified Studio and select IDE Spaces or install the Kiro CLI. For detailed implementation guidance, reference documentation, and migration examples, visit the documentation.  

Publicado el Deja un comentario

Announcing Amazon Nova 2 Sonic for real-time conversational AI

Today, Amazon announces the availability of Amazon Nova 2 Sonic, our speech-to-speech model for natural, real-time conversational AI that delivers industry leading quality and price for voice-based conversational AI. It offers best-in-class streaming speech understanding with robustness to background noise and users’ speaking styles, efficient dialog handling, and speech generation with expressive voices that can speak natively in multiple languages (Polyglot voices). It has superior reasoning, instruction following, and tool invocation accuracy over the previous model.

Nova 2 Sonic builds on the capabilities introduced in the original Nova Sonic model with new features including expanded language support (Portuguese and Hindi), polyglot voices that enable the model to speak different languages with native expressivity using the same voice, and turn-taking controllability to allow developers to set low, medium, or high pause sensitivity. The model also adds cross-modal interaction, allowing users to seamlessly switch between voice and text in the same session, asynchronous tool calling to support multi-step tasks without interrupting conversation flow, and a one-million token context window for sustained interactions.

Developers can integrate Nova Sonic 2 directly into real-time voice systems using Amazon Bedrock’s bidirectional streaming API. Nova Sonic 2 now also seamlessly integrates with Amazon Connect and other leading telephony providers, including Vonage, Twilio, and AudioCodes, as well as open source frameworks such as LiveKit and Pipecat.

Amazon Nova 2 Sonic is available in Amazon Bedrock in the following AWS Regions: US East (N. Virginia), US West (Oregon), Asia Pacific (Tokyo), and Europe (Stockholm). To learn more, read the AWS News Blog and the Amazon Nova Sonic User Guide. To get started with Nova Sonic 2 in Amazon Bedrock, visit the Amazon Bedrock console.

 

​Today, Amazon announces the availability of Amazon Nova 2 Sonic, our speech-to-speech model for natural, real-time conversational AI that delivers industry leading quality and price for voice-based conversational AI. It offers best-in-class streaming speech understanding with robustness to background noise and users’ speaking styles, efficient dialog handling, and speech generation with expressive voices that can speak natively in multiple languages (Polyglot voices). It has superior reasoning, instruction following, and tool invocation accuracy over the previous model.
Nova 2 Sonic builds on the capabilities introduced in the original Nova Sonic model with new features including expanded language support (Portuguese and Hindi), polyglot voices that enable the model to speak different languages with native expressivity using the same voice, and turn-taking controllability to allow developers to set low, medium, or high pause sensitivity. The model also adds cross-modal interaction, allowing users to seamlessly switch between voice and text in the same session, asynchronous tool calling to support multi-step tasks without interrupting conversation flow, and a one-million token context window for sustained interactions.
Developers can integrate Nova Sonic 2 directly into real-time voice systems using Amazon Bedrock’s bidirectional streaming API. Nova Sonic 2 now also seamlessly integrates with Amazon Connect and other leading telephony providers, including Vonage, Twilio, and AudioCodes, as well as open source frameworks such as LiveKit and Pipecat.
Amazon Nova 2 Sonic is available in Amazon Bedrock in the following AWS Regions: US East (N. Virginia), US West (Oregon), Asia Pacific (Tokyo), and Europe (Stockholm). To learn more, read the AWS News Blog and the Amazon Nova Sonic User Guide. To get started with Nova Sonic 2 in Amazon Bedrock, visit the Amazon Bedrock console.