An Interview with OpenAI CEO Sam Altman and AWS CEO Matt Garman About Bedrock Managed Agents

2026-04-28 · Source: Stratechery by Ben Thompson · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Corporate Strategy & Leadership · Depth: Intermediate, extended

Summary

Microsoft and OpenAI have amended their partnership agreement, allowing OpenAI to offer its models on other cloud providers, including AWS. This change addresses the previous exclusivity with Azure, which hindered OpenAI's reach to enterprises preferring other cloud environments. The core of this new collaboration is "Bedrock Managed Agents, powered by OpenAI," a service that integrates OpenAI's frontier models within an AWS-native agent runtime, providing identity, permissions, logging, and governance. This offering aims to simplify the development and deployment of stateful AI agents for organizations with existing AWS infrastructure. The shift acknowledges that Azure's exclusivity, while a competitive advantage, was ultimately detrimental to Microsoft's investment in OpenAI, especially given Anthropic's growth. OpenAI is also releasing Microsoft from the AGI clause, ensuring the agreement runs through 2032.

Key takeaway

For CTOs and VP of Engineering evaluating AI strategy, the amended Microsoft-OpenAI deal and the launch of Bedrock Managed Agents on AWS signal a critical shift towards multi-cloud AI deployment and integrated agent solutions. You should prioritize cloud-native managed agent services that embed AI models within your existing infrastructure, as this approach simplifies security, governance, and scalability, accelerating time-to-value for complex AI workflows like virtual co-workers. This move also validates a partner-centric approach over full vertical integration for broader market reach.

Key insights

AI agent development requires deep integration of models with cloud-native runtime environments for enterprise adoption.

Principles

• Platform shifts enable startups with faster cycles and less capital.
• Demand for intelligence is highly elastic at low enough prices.
• Partnerships can grow the market more effectively than full vertical integration.

Method

Bedrock Managed Agents package OpenAI's models within an AWS-native agent runtime, handling identity, permissions, logging, and governance to simplify stateful agent deployment in enterprise VPCs.

In practice

• Utilize managed agent services for easier enterprise AI integration.
• Prioritize frontier models for current AI applications.
• Expect AI costs to decrease dramatically over time.

Topics

• Bedrock Managed Agents
• OpenAI AWS Partnership
• AI Agent Development
• Model Harness Integration
• Microsoft OpenAI Agreement

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Architect, AI Product Manager

Related on AIssential

• OpenAI ends its exclusive partnership with Microsoft — OpenAI's amended Microsoft deal enables multi-cloud deployment and delinks revenue share from AGI progress.
• Integrating AWS API MCP Server with Amazon Quick using Amazon Bedrock AgentCore Runtime — Conversational AI streamlines AWS operations by translating natural language into secure CLI commands via Bedrock AgentCore and MCP.
• OpenAI, Anthropic fight on the frontier — Leading AI labs are rapidly advancing models with self-improvement, multi-agent capabilities, and deep enterprise integration.
• OpenAI models now available on Amazon Web Services — OpenAI's GPT-5.5, GPT-5.4, and Codex are now available on AWS Bedrock, integrating with existing AWS infrastructure and security.
• OpenAI ends Microsoft legal peril over its $50B Amazon deal — Renegotiated terms between Microsoft and OpenAI resolve exclusivity conflicts, enabling broader multi-cloud deployment for OpenAI products.
• OpenAI and Amazon announce strategic partnership — OpenAI and Amazon partner to integrate advanced AI models and infrastructure, expanding enterprise AI capabilities and compute capacity.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Stratechery by Ben Thompson.