Amazon Web Services (AWS) can now provide OpenAI models through its Bedrock service, but some AWS customers have responded tepidly.

The Information reported recently that some companies are satisfied with models already offered on Bedrock, such as Anthropic's Claude and Amazon's own Nova, while others already use OpenAI through other clouds.

Adam Sandman (애덤 샌드먼), CEO of IT testing and quality-assurance software company Inflectra, said the announcement that OpenAI can be used on AWS would have been important if it had come years ago. He said Claude is better for coding and many tasks. He added that Qwen and DeepSeek are models currently drawing attention. Phil Christianson (필 크리스찬슨), chief product officer at IT software company Xurrent, also said access to OpenAI would have been good 2 to 3 years ago. He said frontier models are all good enough now.

The Information said the situation shows OpenAI faces the task of catching up to Anthropic in the enterprise market.

Anthropic recently achieved an annualised revenue of $30 billion.

Amazon said in February it would provide OpenAI technology on Bedrock in a "stateful" runtime environment.

The stateful approach supports more continuous operation by AI agents by maintaining conversational context. Chris Daniluk (크리스 다닐룩), CEO of AWS consulting company LeadMic Technologies, said that in the current situation where AI has difficulty operating autonomously, a stateful runtime can smoothly coordinate agents, but being required to use OpenAI models is somewhat inconvenient. He added that the product could be attractive to customers that were unable to use OpenAI because they did not want to move to Microsoft Azure.