Source: Wall Street See

As the world's largest cloud computing service provider, $Amazon (AMZN.US)$ in a recent collaboration with $Microsoft (MSFT.US)$ and $Alphabet-C (GOOG.US)$ Falling behind in the competition. To regain its former glory, Amazon is accelerating the pace of building its advantage in the field of artificial intelligence computing.

On Wednesday local time, Amazon held the Re:Invent conference in Las Vegas and announced a series of latest heavyweight plans and products for AWS (Amazon Cloud) at the conference, including:

Partnering with rival OpenAI's Anthropic to build the world's most powerful AI supercomputer; preparing to collaborate with Apple to train Apple Intelligence (Apple AI); and launching the next-generation more advanced 3-nanometer training chip Trainium 3 by the end of next year. In addition, Amazon also introduced the new generation foundational model Amazon Nova.

Joining hands with Anthropic to build an AI supercomputer.

Matt Garman, CEO of Amazon Web Services, announced at the conference that the company is creating an AI training platform called UltraServers - including 64 Trainium 2 chips - to expand generative AI workloads.

This cloud computing platform will be used for an ambitious plan - an AI supercomputer. Amazon announced at the conference that it will cooperate with Anthropic to develop the AI supercomputer UltraCluster, codenamed Rainer.

Amazon expects this computer to be equipped with hundreds of thousands of Amazon's latest AI training chips Trainium 2. Once completed, it will be five times larger than the cluster of Anthropic's current most powerful model, becoming the world's largest AI supercomputer.

Facing formidable competition from Microsoft and OpenAI, Amazon has chosen to team up with Anthropic, investing 8 billion US dollars in Anthropic this year.

Will the more powerful new generation of chips challenge the dominant position of nvidia?

Garman stated at the Re:Invent conference that Tranium 2 will be widely available in the Trn2 UltraServer cluster dedicated to training cutting-edge ai.

Many companies have started using amazon's cloud to build and train custom ai models, typically in conjunction with $NVIDIA (NVDA.US)$ GPUs. However, Garman stated that the new AWS clusters are 30% to 40% cheaper than clusters using nvidia GPUs.

Garman stated that amazon's goal is to reduce the cost of ai.

Today, when it comes to GPUs, there is really only one choice, and that is nvidia. We believe customers will appreciate having multiple options.

I hope Trainium can carve out a good niche market. I think it will be an excellent choice for many workloads — not all workloads.

At the meeting, Amazon also released its next generation of more advanced 3-nanometer training chip Trainium 3, claiming that the chip's performance will be four times that of existing chips, and is scheduled to be launched to customers by the end of 2025.

Moore Insight & Strategy's CEO and Chief Analyst Patrick Moorhead stated that the "data on Trainium 3 is quite amazing", and its performance seems to have been significantly improved due to improvements in chip interconnects. Interconnects are crucial for developing super large AI models, as they enable fast data transfer between chips, with AWS seemingly optimizing this in their latest design.

Moorehead also mentioned that in the near future, Nvidia may still dominate the AI training field, but in the coming years, it will face increasingly intense competition, with Amazon's innovation showing that Nvidia is not the exclusive player in the training field.

Is a New Chapter Opening in Apple's Collaboration?

According to previous reports, Apple is considering using the Trainium 2 chip for Apple Intelligence training. Garman stated at the meeting that Apple is one of its customers, having adopted some AWS chips such as Amazon Graviton and Inferentia to support services including Siri.

Apple's Director of AI and Machine Learning Benoit Dupin later took the stage at a conference in Las Vegas. He stated that Apple has collaborated with AWS in "almost every stage" of the AI and machine learning lifecycle.

AWS can keep up with the pace, we have been a customer for over a decade.

Dupin mentioned that Apple is conducting preliminary testing on the Trainium 2 chip to assist in training Apple Intelligence.

The new generation of generative AI models.

At the conference, amazon also launched a new generation of foundational model called Amazon nova.

Amazon stated that AWS customers will be able to use generative AI applications supported by Amazon nova to 'understand videos, charts, and documents, or generate videos and other multimedia content,' with a range of models available at different costs.

Among them, Amazon nova micro, lite, and pro support over 200 languages. Amazon nova micro supports a context length of 128K input tokens, while Amazon nova lite and pro have a context length of up to 300K tokens, and can process videos up to 30 minutes in duration.

'Amazon nova micro, Amazon nova lite, and Amazon nova pro are at least 75% cheaper than the best performing models in their respective intelligent categories within Amazon Bedrock.'

In addition, amazon also introduced two new models - Amazon nova canvas for generating high-quality images and Amazon nova reel for generating high-quality videos.

Rohit Prasad, Vice President of General AI at amazon, revealed that internally, around 1000 generative AI applications are in progress at amazon.

AI agent administrators, reasoning inspectors...

Amazon has launched a series of AI tools on the AWS platform to help enterprises utilize and master generative AI technology.

For example, a new cloud tool called Bedrock Agents announced on Wednesday can be used to create and manage AI agents, which can automatically perform useful tasks such as customer support, order processing, and analysis. It includes a master agent responsible for managing subordinate AI teams, providing reports on how they operate, and assisting with adjustments.

Garman said:

Basically, you can create an agent that makes you the boss of all the other agents.

In addition, Amazon will also launch a system that uses larger models to enhance the capabilities of smaller models, as well as a tool to verify whether the outputs of chatbots are correct (used to combat AI illusions).

Editor / jayden