AWS joins cloud price war against Google and Microsoft on gen AI

Nova foundation models intended to attract converts from Gemini and OpenAI

AWS has released its Nova generative AI services, intended to provide a competitive alternative to Google and Microsoft’s models.

The company, two years behind its competitors in releasing its own AI systems, is offering a range of models, either available now or coming soon. They will roll out first in AWS’s North American regions.

The company claims its pricing is at least 75% lower on AWS Bedrock, Amazon’s full managed service for foundation models, than the respective offerings from Google and Microsoft.

The services available now are as follows:

Amazon Nova Micro is a text-only model that delivers low latency responses at low cost. AWS claims Nova Micro is highly performant at language understanding, translation, reasoning, code completion, brainstorming and mathematical problem-solving. It offers a generation speed of more than 200 tokens per second.

Amazon Nova Lite is a low-cost multi-modal model offering fast responses for image and video processing, as well as text inputs. AWS claims Nova Lite provides a high level of accuracy and high speed across a range of tasks. This, it adds, makes it suitable for a wide range of interactive and high-volume applications where cost is a key consideration.

Amazon Nova Pro is a multi-modal model offering a combination of accuracy, speed and low-cost for a range of tasks. AWS says that its capabilities and speed make it a suitable model for almost any task, including video summarisation, Q&A, mathematical reasoning, software development and AI agents that can execute multi-step workflows.

The following services are listed as “coming soon”:

Amazon Nova Canvas is an image generation model that creates professional grade images from text or images provided in prompts. The service also provides features to make it easy to edit images using text inputs, controls for adjusting colour schemes and layout and built-in controls the company claims will support safe and responsible use of AI.

Amazon Nova Reel is a video generation model that customers can use to create high quality video from text and images. Nova Reel supports use of natural language prompts to control visual style and pacing, including camera motion control, and built-in controls to support safe and responsible use of AI.

Amazon will also offer a service called Nova Premier, a model for complex reasoning tasks that is still in training but, it claims, will be available early next year.

Computing says:

Pricing its services lower than competitors is a must if Amazon wants to break into the gen AI market at this stage. While most companies are still in the trial/PoC stage, decisions are starting to firm up over which platform they will pursue, and to-date that’s largely been a choice between two key players.

More competition is never a bad thing for the end customer, but has Amazon struck too late? It was a frontrunner in the cloud market, where it beat Microsoft and Google by several years, but on AI it risks being a late entrant – doomed to the same low-adoption fate as Google Cloud.