OpenAI Models Now Accessible on AWS for the First Time

Sam Altman’s competitive advantage is heating up, underscored by a new partnership with Amazon Web Services.

As OpenAI unveiled two open-weight reasoning models similar to its o-series, Amazon revealed that these models will soon be offered as a service on AWS. This marks the first instance where OpenAI’s models will be available to enhance Amazon AI services like Bedrock and SageMaker AI.

While the models can be accessed through Hugging Face, Amazon’s acquisition of them is more than a simple AWS service offering. This partnership with OpenAI was emphasized by Dmitry Pimenov, product lead for the model maker, in the announcement.

This is a pivotal strategic shift for both organizations. For AWS, this collaboration positions the cloud giant alongside the premier model creator, OpenAI.

Previously, AWS was primarily recognized as a key host and financial backer of Anthropic’s Claude, a significant contender to OpenAI. AWS provides Claude alongside other models from developers like Cohere, DeepSeek, Meta, and Mistral, as well as its proprietary options in its AI services. Specifically, Bedrock empowers AWS customers to build and host GenAI applications using their preferred models, while SageMaker enables users to train or create their own AI models primarily for analytics.

Although AWS’s main competitor, Microsoft, does not have exclusive access to OpenAI models as of January, Azure remains OpenAI’s most notable cloud partner to date. OpenAI has also disclosed that Microsoft will offer versions of these two new models optimized for Windows devices.

Amazon’s CEO, Andy Jassy, has been concerned about Microsoft capturing a larger share of cloud business tied to OpenAI. Recently, during Amazon’s quarterly earnings call, Jassy faced multiple queries from Wall Street analysts regarding the company’s perceived decline in AI competitiveness, especially against Microsoft.

For instance, JPMorgan analyst Doug Anmuth asked Jassy to clarify on “significantly faster cloud growth among the number two and number three players in the space,” which refers to Microsoft and Google. Later, Morgan Stanley analyst Brian Nowak informed Jassy that Wall Street views AWS as lagging in GenAI, raising concerns about market share depletion.

In his response, Jassy provided a comprehensive rebuttal, including a subtle jab at Microsoft: “I believe the second player is about 65% the size of AWS.”

Simultaneously, another AWS rival, Oracle, announced a $30 billion per year contract with OpenAI for data center services, indicating that OpenAI will pay Oracle more annually than all its other cloud service clients combined. Up until now, AWS had missed out on any accolades related to OpenAI.

As for how this collaboration benefits OpenAI: the AI provider’s relationship with Microsoft is reportedly tense, with ongoing negotiations about their long-term partnership. What better way for OpenAI to reinforce its position than by aligning with the largest cloud provider, even if it begins on a modest scale?

Moreover, this partnership allows numerous AWS enterprise clients to easily experiment with OpenAI models for their hosted AI applications.

Altman also gains a strategic advantage over Meta’s Mark Zuckerberg with this move. While OpenAI rolls out these two high-performance models under an Apache 2.0 open-source license, Meta has recently admitted it might not continue to open source all of its upcoming “superintelligence” models.

AWS did not promptly respond to our request for further comments.