Summary Bullets:
• Various new capabilities in cloud migration, AI, and agentic AI that are aligned with business needs in APAC.
• This shows strong momentum, but there are a few considerations for AWS to strengthen its position in the region.
In a recent briefing with analysts in APAC, AWS shared its key launches in H1 2025. In line with the market direction, most new services and features are around AI. Cloud adoption is growing while AI is evolving rapidly in the region. The focus has shifted from LLMs and use case creations to efficient deployments and advanced automation. For example, using the right model (third-party, custom model, model distillation, fine-tuning, and SLM) and agentic AI (multi-agent applications and agent development, including support for third-party agents and open-source agent SDK). The new capabilities are crucial for AWS to address the growing customer needs. Businesses have higher awareness of AI and are beginning to feel the push to adopt the technology to keep up with user demands and gain a competitive edge. The new capabilities are also crucial for AWS to retain its market position and to respond to competitors.
Cloud migration: AWS launched Amazon Elastic VMware Service (EVS) to simplify migration to the AWS’ environment (including AWS Outposts). While AWS’ support for VMware workloads is not new, Amazon EVS enables enterprises to retain VCF architecture (e.g., SDDC manager, vSphere, vSAN, and NSX) while providing deployment flexibility (e.g., self-manage or partners’ managed services and pay-per-use or bring-your-own-subscription models). Besides, AWS also launched AWS Transform, an agentic AI service (in both web-based and IDE), to accelerate VMware migration to EC2. The agent is designed to analyze workloads, dependencies, and readiness; convert VMware networking configurations to AWS; generate plans; and user validation (human in the loop). This can address the growing cloud migration in the region, but also minimize challenges such as enterprises’ integration, vendor lock-in, security, scalability, licensing and costs, and skill gap. Besides, with cloud-native environments, it can also future-proof enterprise workloads through options to refactor, replatform, and even repatriate the applications, which enables businesses to move away from VMware. AWS Transform is also available for mainframe and .NET application modernization.
Agentic AI: Apart from AWS Transform, there are several other new features highlighted by the vendor. AWS introduced Amazon Bedrock Agents by choosing the right models and data to execute specific tasks. The vendor has also added multi-agent collaboration as part of its Amazon Bedrock capabilities to enable management of multiple agents to address complex workflows. AWS is increasingly promoting open-source by adding support for (1) Strands Agent, an open-source agent SDK, and (2) Model Context Protocol (MCP), an open standard for integration across agents as well as data sources and tools. This provides wider flexibility for enterprises to deploy agentic AI, from specialized agents (Amazon Q), to fully managed agents (Amazon Bedrock) and DIY (open-source). This is crucial for enterprises to achieve greater efficiency and scalability, especially when they have implemented multiple agents from various providers for different business processes. Besides, Amazon Q can index data from various third-party sources including Salesforce, Zoom, Google, and Microsoft Exchange.
Other AI capabilities: There are also many other new features and capabilities of Amazon Bedrock including latency-optimized inference, model distillation, and intelligent prompt routing for model optimization, as well as support for new third-party models such as Deepseek, TwelveLabs, and Poolside. Another interesting new capability of Amazon Bedrock is cross-region inference which distributes its GPU capacity within a geographical region. This can provide cost-efficient solutions for enterprises who are developing AI applications that are not latency-sensitive nor bound by data sovereignty requirements. For Amazon Nova (its in-house models), the vendor highlighted Amazon Nova Sonic, a speech-to-speech model that provides higher performance (faster and more accurate) compared to the traditional approach (speech-text-model-text-speech again). It also introduced Amazon Nova Act, a model that allows a human interface (e.g., selecting an option on a web interface).
Conclusion: The new capabilities show AWS’ strong momentum in the rapidly evolving cloud and AI markets. AWS has also demonstrated various customer references with the new capabilities, across multiple industries. However, competitors are also moving at a similar pace. There are still some areas for consideration for AWS to further drive its position in the market. This includes showcasing wider references in APAC, supporting broader AI service availability in new regions in Asia, and AI edge (e.g., Outposts deployment).