Amazon Announces New Services to Compete with OpenAI and GitHub Copilot
Amazon Announces New Services to Compete with OpenAI and GitHub Copilot
A marketplace of AI models and a coding companion
Amazon today announced a new generative AI service called Bedrock and made its CodeWhisperer text-to-code solution generally available and free for individual use. This move pits Amazon directly against Microsoft’s OpenAI services and its GitHub Copilot code generation solution.
With Bedrock, Amazon Web Services (AWS) is bringing to foundation AI models the same simplicity it brought to computing infrastructure. Key Amazon Bedrock foundation models (FM) include:
Text Generation
Chatbots
Search
Text Summarization
Image Generation
Personalization
Amazon’s first foundation model options include AI21 Labs' Jurassic-2, Anthropic’s Claude , Stability.ai’s Stable Diffusion, and Amazon’s Titan. Noticeably absent from the list is Hugging Face, which selected AWS to host its AI model training and inference, including the open-source BLOOM large language model (LLM).
CodeWhisperer Now Available
CodeWhisperer was first announced at Amazon’s re:MARS conference in June 2022, which was the same month that GitHub’s Copilot launched in general availability. Amazon’s product has been in a beta preview since that time.
Amazon said that “productivity challenges” during the preview found that CodeWhisperer users were “27% more likely to complete tasks and did so an average of 57% faster.” This is in line with GitHub Copilot’s claim of 54% developer productivity increases.
A key difference between the services is the price. While an individual can access Copilot for $10 per month, CodeWhisperer is launching with an “individual tier that’s free to use for all developers.
Taking on Microsoft
Microsoft has staked out strong positioning based on its collaboration, investment, and effective control of OpenAI. However, it is very much tied to OpenAI at this point for Azure, Copilot, and a number of productivity applications. That has helped the company because of OpenAI’s early lead in generative AI mindshare and market share. It also leaves an opening for a competition that offers alternative solutions.
Amazon has its own foundation models that compete directly with OpenAI offerings. The Bedrock approach is to offer an “everything else” strategy and not just push its own products. Eventually, Bedrock may include access to OpenAI models. Sam Altman, OpenAI’s CEO has said the company’s deal with Azure was not exclusive. Of course, that comment was prior to the $10 billion investment that reportedly left Microsoft with a 49% ownership stake in OpenAI.
It could be that OpenAI is exclusive to Azure. Regardless, Bedrock is, for now, promoting easy access to competing foundation models. This should give a meaningful boost to AI21 and Anthropic and put Amazon’s Titan in the mix. Stable Diffusion probably doesn’t nee the boost but making AI models easier to adopt may be the next key area of competition.
The introduction of CodeWhisperer for general use is not surprising. However, the offer of a free coding companion when Copilot requires a monthly fee is another direct challenge to Microsoft’s early market leadership in the segment.
Flanking Maneuver on Nvidia and Google
Amazon’s entry may prove to be more challenging for Nvidia and Google. Nvidia recently announced an AI model hosting service and is promoting its Picasso service for image and video generation and NeMo for LLMs. Google announced the PaLM LLM API last month, but it is still in a limited preview. For LLMs, everyone now is competing to become the alternative to OpenAI. Amazon Bedrock is going to make that harder for Nvidia and Google to accomplish.
In the text-to-image space, Stability AI and its Stable Diffusion already seems to have a lead over OpenAI’s DALL-E. AWS may help solidify Stable Diffusion as a market leader and will make it harder for Nvidia’s Picasso and Google’s yet-to-be-released Imagen to gain market momentum. Nvidia and Google may turn out to have strong offerings, but competition is increasing, and that introduces uncertainty around any one company’s product adoption prospects.
A few months ago, Amazon looked like it was asleep as the generative AI market began its Cambrian explosion. However, the strength of the current AWS user base and what looks like a logical strategy in Bedrock may help put Amazon in the generative AI leadership group far faster than expected.
