AWS unveils next-gen chips and serverless innovations at 're:Invent’; see details

AWS CEO Adam Selipsky revealed an array of fresh offerings on Tuesday, including a bespoke chip tailored for generative AI (Artificial Intelligence) and ML (Machine Learning) training to drive widespread adoption of GenAI among enterprises of all sizes. Selipsky highlighted that 80 percent of unicorn companies (those valued at $1 billion and beyond) globally rely on their cloud services, and the innovation spurred by GenAI is set to expand across businesses of varying scales. The cloud giant has been immersed in AI and ML advancements for over 25 years.

During the keynote at the annual 're:Invent' conference, Selipsky emphasized AWS as the driving force for innovators worldwide, catering to diverse businesses—from major financial service providers to automotive titans.

"Innovative strides with GenAI continue. Businesses, regardless of size, are at the initial stages of exploration with GenAI, pioneering early-use cases with large language models (LLMs). We're presenting an enterprise-ready generative AI stack," he shared with a captivated audience.

Selipsky introduced the latest generation of two AWS-developed chips—AWS Graviton4 and AWS Trainium2—tailored for a wide spectrum of customer workloads encompassing machine learning (ML) training and generative artificial intelligence (AI) applications.

Graviton4 boasts up to 30 percent improved compute performance, 50 percent more cores, and 75 percent enhanced memory bandwidth compared to the current Graviton3 processors, ensuring optimal price performance and energy efficiency for various workloads running on Amazon EC2.

Currently, AWS offers over 150 distinct Graviton-powered Amazon EC2 instance types globally at scale, has produced over 2 million Graviton processors, and serves more than 50,000 customers—among them, the top 100 EC2 customers—utilising Graviton-based instances for peak price performance in their applications.

During the event, the company unveiled three novel serverless advancements across its database and analytics suite, streamlining the scalability of customers' data infrastructure to support their most challenging use cases.

Dr Swami Sivasubramanian, AWS's Vice President of Data and Artificial Intelligence, emphasised that these serverless innovations "enhance the groundwork, simplifying customers' scalability to millions of transactions per second, rapid capacity augmentation, and dynamic workload pattern adaptation for performance and cost optimisation."

A new palm-scanning identity service was also announced, enabling companies to authenticate individuals upon physical premises entry. Termed 'Amazon One Enterprise,' this service eradicates operational complexities linked with traditional enterprise authentication methods, such as badges and PINs, as highlighted during the 'AWS re:Invent 2023' event.

The palm-recognition technology leverages advanced AI and ML to generate a palm signature linked to identification credentials like badges, employee IDs, or PINs.

"Amazon One Enterprise's palm recognition technology aims to deliver a highly precise identification service, heightening an organisation's overall security while streamlining authentication management with reduced operational overhead," mentioned Dilip Kumar, AWS Applications Vice President.

Amazon also introduced new devices priced at $195 that enable enterprise users to access virtual desktop environments, like Amazon WorkSpaces, via the internet, housed within Fire TV Cube hardware.

Also Read TCS partners with AWS on generative AI practices, will train 25,000 employees for certifications