AWS unveils new Trainium2 AI chips for EC2 instances

LAS VEGAS - Amazon Web Services (AWS), a key division of the $2.23 trillion market cap technology giant Amazon, announced the availability of new Amazon Elastic (NYSE:ESTC) Compute Cloud (Amazon EC2) instances powered by its latest Trainium2 AI chips, promising up to 40% better price performance compared to current GPU-based instances. The announcement was made at the AWS re:Invent conference, showcasing the company's commitment to advancing artificial intelligence (AI) and machine learning (ML) capabilities. With revenue reaching $620.13 billion in the last twelve months, Amazon continues to demonstrate its market leadership in cloud computing and AI innovation.According to InvestingPro data, 31 analysts have recently revised their earnings expectations upward for Amazon, reflecting strong confidence in the company's technological initiatives and market position.

The new EC2 Trn2 instances, which are now generally available in the US East (Ohio) region, feature 16 Trainium2 chips, providing 20.8 peak petaflops of compute. This enhancement is tailored for training and deploying large language models (LLMs) with billions of parameters, addressing the growing demand for high-performance computing in AI workloads.

In addition to the Trn2 instances, AWS introduced Trn2 UltraServers, a new EC2 offering that connects four Trn2 servers using the ultra-fast NeuronLink interconnect. This configuration scales up to 83.2 peak petaflops, quadrupling the compute, memory, and networking capabilities of a single instance, making it possible to train and deploy the world's largest models.

AWS is collaborating with Anthropic, an AI safety and research company, on Project Rainier, which involves building an EC2 UltraCluster of Trn2 UltraServers. This project aims to create the largest AI compute cluster reported to date, with hundreds of thousands of Trainium2 chips, to facilitate the training of future AI models.

Moreover, AWS unveiled the Trainium3 chip, its next-generation AI training chip, which is expected to offer four times the performance of Trn2 UltraServers. The Trainium3 chips are designed with a 3-nanometer process node and are slated to be available in late 2025.

The announcement also highlighted the Neuron software development kit (SDK), which helps developers optimize their AI models to run on Trainium chips. The Neuron SDK is integrated with popular frameworks like JAX and PyTorch, allowing for seamless adoption of Trainium2 technology.

Several companies, including Databricks and Hugging Face, plan to leverage the Trainium2's high performance and cost-efficiency to improve their AI offerings. Databricks' Mosaic AI and Hugging Face's open platform for AI builders are expected to benefit from the enhanced capabilities of Trainium2.

This development underscores AWS's leadership in providing a broad and deep selection of accelerated EC2 instances for AI/ML applications. The company's continued innovation in cloud computing infrastructure aims to meet the escalating compute and networking demands of sophisticated AI models.

The information in this article is based on a press release statement from Amazon Web Services. Amazon's financial health score of 3.15 (rated as "GREAT" by InvestingPro) underscores the company's robust position to continue investing in advanced technologies like Trainium2 and Trainium3.Unlock deeper insights into Amazon's financial performance and growth potential with an InvestingPro subscription, which includes access to detailed Pro Research Reports, over 15 additional ProTips, and comprehensive financial metrics that help you make informed investment decisions.

In other recent news, Amazon.com Inc (NASDAQ:AMZN). reported record-breaking holiday sales, with independent sellers accounting for over 60% of the sales. The company displayed significant revenue growth of 11.93%, reaching $620.13 billion. Amazon Web Services (AWS), an Amazon subsidiary, introduced new features to its Simple Storage Service (S3), enhancing data analytics and management capabilities. BMO Capital Markets reiterated its Outperform rating on Amazon, while MoffettNathanson maintained its Buy rating.

The company also committed over $500 million to advance nuclear energy technologies, aligning with its Climate Pledge to achieve net-zero carbon emissions by 2040. AWS also highlighted its collaboration with NVIDIA (NASDAQ:NVDA) on the Omniverse platform, aiming to advance Physical AI capabilities.

Amazon's recent developments also include the launch of AI-powered features such as Rufus and Amazon Lens to enhance the shopping experience. These recent developments are shaping Amazon's operations and strategic decisions.

This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.