AWS reshuffles enterprise AI... the core lies in "agents"

As advancements in artificial intelligence systems garner widespread attention, Amazon Web Services (AWS) is targeting the next phase of enterprise AI. At this year’s AWS re:Invent event, “Agent AI” emerged as the core architecture to replace traditional cloud strategies, shifting the discussion focus beyond mere model performance to how to integrate AI into real business processes. Although this momentum has been forming for years, industry evaluations suggest that this event marks the official start of a structural overhaul of AI architecture and the entire cloud infrastructure.

AWS’s new CEO Matt Garman stated, “80% to 90% of the value of enterprise AI in the future will come from ‘Agent’ architecture,” positioning the agent concept as the central axis of the new cloud era. He emphasized that through a comprehensive tech stack—including scalable custom semiconductors, new model series, learning-based architectures, and runtime operations—companies can build valuable intelligence systems. AWS’s design of AI factories to accommodate the needs of a very small number of high-performance clients is also an extension of this strategy.

This direction is not just a vision. AWS has set a goal to evolve beyond AI assistance toward “team-member AI” capable of collaboratively handling entire organizational operations. Swami Sivasubramanian, Vice President of the Agent AI division, explained, “Most AI systems are like interns, needing to learn from scratch every day; what we want is AI that can understand the context of work over the long term.” Along this vision, new storage services like S3 vector, capable of directly processing internal enterprise data, reducing latency, and improving accuracy, are also part of the plan.

Industry leaders are also responding enthusiastically to this change. Jerry Chen, Partner at Grelock Partners, pointed out, “Cloud architectures are beginning to evolve from traditional ‘cloud’ structures to ‘cloud + AI’ structures,” emphasizing the need to prepare for sovereignty cloud environments driven by AI. In line with this trend, AWS has launched the Nova Forge platform, supporting fine-tuning of cutting-edge models across various industries, demonstrating a move to accelerate model training based on enterprise-specific data.

The agent-based architecture is moving beyond simple automation toward redesigning the overall customer experience. AWS Vice President Colleen O’Brien used the shopping tracking technology “Just Walk Out” as an example: “A system capable of real-time visual reasoning decisions in a fully anonymized state could meet diverse privacy protection requirements in the future.”

This shift is also clearly visible in developer trends. Tools are evolving from simple code assistance to integrated agent-type tools capable of handling everything from planning to documentation and task execution. According to AWS Vice President Deepak Singh, the Kiro platform focuses on converting conversational instructions into explicit specifications to quickly deliver results-level products.

The wave of transformation is also affecting the entire cloud ecosystem. Salesforce Vice President Nick Johnston said, “Customers want an open platform that avoids vendor lock-in and can activate data anywhere,” announcing the integration plan “Agentforce 360” with AWS. Oracle is accelerating its multi-cloud strategy, while Snowflake is porting Cortex-based financial and security agents to AWS Bedrock Core, aiming to enhance intelligent analysis capabilities.

To ensure consistent AI performance across different environments, multi-cloud integration strategies are also accelerating. Companies are freely migrating mission-critical data between clouds to effectively harness the potential of evolving agent AI. User-centric customized insights are also being deployed more rapidly within such hybrid architectures.

Cloud infrastructure is being redefined. Systems are shifting toward AI-first architectures, with agent AI now seen not just as a tool but as a “digital colleague.” Led by AWS and its partners, this trend is likely to establish a new benchmark for enterprise technology strategies.