The cloud computing division of Amazon.com unveiled a suite of technologies Thursday aimed at helping other companies develop their own chatbot and AI-powered imaging services.
Microsoft Corp and Alphabet Inc are incorporating AI chatbots into consumer products like their search engines, but they’re also aiming for another big market: selling the underlying technology to other companies through their cloud operations.
Amazon Web Services (AWS), the largest provider of Cloud Computing in the world, launched into that race on Thursday with a suite of its own AI technologies, but is taking a different approach.
AWS will offer a service called Bedrock that allows companies to customize what are called core models—the basic AI technologies that do things like respond to queries with human-like text or generate images from a prompt—with their own data to create a unique model. OpenAI, the creator of ChatGPT, for example, offers a similar service that allows customers to fine-tune ChatGPT models to create a custom chatbot.
Bedrock will allow customers to work with Amazon’s own base models, called the Amazon Titan, but will also offer a menu of models offered by other companies. The first third-party choices will come from the startups AI21 Labs, Anthropic and Stability AI, along with Amazon’s own models.
The service allows customers of AWS testing those technologies without having to deal with the servers in the underlying data centers that power them.
“It’s unnecessary complexity from a user perspective,” Vasi Philomin, vice president of generative AI at AWS, told Reuters. “Behind the scenes, we can abstract that away.”
Those underlying servers will use a mix of custom AI chips from Amazon as well as processors from Nvidia Corp, the biggest provider of chips for AI work, but whose products have been in short supply this year.
