OpenAI has integrated its Codex model into Amazon Bedrock, a generative AI service from Amazon's cloud computing subsidiary, Amazon Web Services (AWS), the two companies announced on Tuesday.
The move comes a few days after OpenAI ended its exclusive cloud arrangement with Microsoft.
Amazon Bedrock allows developers and companies to build, test and scale AI applications using pre-built foundation models. Launched in 2023 and now available across multiple global regions including India, Bedrock gives users access to a range of AI models through a single platform, enabling them to develop applications such as chatbots, virtual assistants and content generation tools.
Amazon has also introduced Amazon Bedrock Managed Agents, powered by OpenAI, with the partnership enabling agentic capabilities to run on AWS’s global infrastructure.
"This is what our customers have been asking for a really long time. Their production applications run in AWS. Their data is AWS," said Matt Garman, CEO of AWS, at an event in San Francisco. Customers will no longer have to leave AWS to find the AI models they want, he said.
Driving the news
The rollout, currently in limited preview, includes OpenAI’s coding agent Codex, which is used by more than 4 million people each week to automate coding tasks, write and refactor software, generate tests, and accelerate development.
By making these tools available within Bedrock, the companies effectively bring AI-driven software development into the AWS environment, allowing enterprises to access, fine-tune and deploy OpenAI models using the same APIs and orchestration systems they already rely on.
This also marks the first time AWS customers can apply usage of OpenAI models, including Codex, toward their existing cloud spending commitments.
OpenAI’s latest models are being made available alongside offerings from Anthropic, Meta, Mistral, Cohere and Amazon, with what AWS describes as “unified security, governance and cost controls”.
What makes the OpenAI–AWS partnership significant?
AWS said the integration requires no additional infrastructure or new security configurations, allowing enterprises to adopt OpenAI models using existing operational systems.
Earlier this month, Amazon CEO Andy Jassy disclosed for the first time that AI services at AWS generated annualised revenue of more than $15 billion in the first quarter. AWS revenue growth has once again gone past the 20% mark in the past two quarters, boosting investor confidence.
Both companies confirmed that usage of OpenAI models on Bedrock will count towards existing AWS cloud commitments, enabling organisations to consolidate AI-related spending within their broader cloud budgets.
On the security front, OpenAI models inherit AWS’s enterprise controls by default. This includes access management via AWS Identity and Access Management (IAM) and secure connectivity through AWS PrivateLink.
Data protection measures, including encryption at rest and in transit, are applied alongside monitoring and audit capabilities through AWS CloudTrail, with integration into existing compliance frameworks.
Amazon OpenAI ties
Amazon and OpenAI have been deepening their ties for several months now. While Amazon invested $50 billion in OpenAI, the AI startup committed to spend $100 billion on AWS over the next eight years.
The companies have also reached an agreement for OpenAI to use two gigawatts of compute powered by Amazon's in-house Trainium AI chips.
The move comes a few days after OpenAI ended its exclusive cloud arrangement with Microsoft.
Amazon Bedrock allows developers and companies to build, test and scale AI applications using pre-built foundation models. Launched in 2023 and now available across multiple global regions including India, Bedrock gives users access to a range of AI models through a single platform, enabling them to develop applications such as chatbots, virtual assistants and content generation tools.
Amazon has also introduced Amazon Bedrock Managed Agents, powered by OpenAI, with the partnership enabling agentic capabilities to run on AWS’s global infrastructure.
"This is what our customers have been asking for a really long time. Their production applications run in AWS. Their data is AWS," said Matt Garman, CEO of AWS, at an event in San Francisco. Customers will no longer have to leave AWS to find the AI models they want, he said.
Driving the news
The rollout, currently in limited preview, includes OpenAI’s coding agent Codex, which is used by more than 4 million people each week to automate coding tasks, write and refactor software, generate tests, and accelerate development.
By making these tools available within Bedrock, the companies effectively bring AI-driven software development into the AWS environment, allowing enterprises to access, fine-tune and deploy OpenAI models using the same APIs and orchestration systems they already rely on.
This also marks the first time AWS customers can apply usage of OpenAI models, including Codex, toward their existing cloud spending commitments.
OpenAI’s latest models are being made available alongside offerings from Anthropic, Meta, Mistral, Cohere and Amazon, with what AWS describes as “unified security, governance and cost controls”.
What makes the OpenAI–AWS partnership significant?
AWS said the integration requires no additional infrastructure or new security configurations, allowing enterprises to adopt OpenAI models using existing operational systems.
Earlier this month, Amazon CEO Andy Jassy disclosed for the first time that AI services at AWS generated annualised revenue of more than $15 billion in the first quarter. AWS revenue growth has once again gone past the 20% mark in the past two quarters, boosting investor confidence.
Both companies confirmed that usage of OpenAI models on Bedrock will count towards existing AWS cloud commitments, enabling organisations to consolidate AI-related spending within their broader cloud budgets.
On the security front, OpenAI models inherit AWS’s enterprise controls by default. This includes access management via AWS Identity and Access Management (IAM) and secure connectivity through AWS PrivateLink.
Data protection measures, including encryption at rest and in transit, are applied alongside monitoring and audit capabilities through AWS CloudTrail, with integration into existing compliance frameworks.
Amazon OpenAI ties
Amazon and OpenAI have been deepening their ties for several months now. While Amazon invested $50 billion in OpenAI, the AI startup committed to spend $100 billion on AWS over the next eight years.
The companies have also reached an agreement for OpenAI to use two gigawatts of compute powered by Amazon's in-house Trainium AI chips.
