Amazon Web Services Creates Striking Options for Cloud AI Applications
Demand for High-End AI Chips Leads to New Announcements
At a conference in Las Vegas, Amazon Web Services (AWS) unveiled a slew of innovations in cloud computing for AI applications. Announced at the Reinvent conference, these new changes include the availability of Nvidia’s latest H200 AI graphics processing units (GPU). It also introduced new chips of its own: the Trainium2 artificial intelligence chips and the general-purpose Graviton4 processor.
Demand for Nvidia’s latest GPUs has soared, especially since the unveiling of OpenAI’s ChatGPT chatbot last year. The release of this technology led to a rapid increase in the adoption of similar generative AI technologies, resulting in a shortage of Nvidia’s chips.
Amazon’s announcement of their new Trainium2 chips along with their Graviton4 processors comes as AWS tries to stand out as a cloud provider. AWS did not shy away from the fact that they not only offer these new chips but will also give customers access to Nvidia’s latest chips.
What’s New
AWS announced offering access to Nvidia’s latest H200 chips. It also announced new Trainium2 AI chips and Graviton4 processors. The H200 is expected to perform twice as fast as its predecessor. Amazon believes that the new Trainium2 chips will offer four times better performance than the original model. Amazon also unveiled its Graviton4 processors, which promise 30% better performance than its current Graviton3 chips.
According to Amazon, more than 50,000 of its customers are already using Graviton chips, which are based on Arm architecture and consume less energy than chips from Intel or AMD. Amazon also aims to gain more from its relationship with Nvidia with the operations of more than 16,000 Nvidia GH200 Grace Hopper Superchips. Allowing its Nvidia-based infrastructure to be utilized by Nvidia’s research and development group along with AWS customers.
Final Thoughts
AWS did not provide any specific release dates for virtual-machine instances with Nvidia H200 chips nor instances relying on the new Trainium2 silicon. However, customers can start testing Graviton4 virtual-machine instances now before they become commercially available in the next few months. AWS has launched more than 200 cloud products since 2006, and has continued to invest in the Graviton and Trainium programs suggesting it is responding to demand.
I have been featured in numerous publications, both online and offline, and am a regular speaker at industry events. I am also the founder of Crypto University, an online educational platform that helps people learn about cryptocurrencies and blockchain technology. In addition to my writing and teaching career, I am also an active investor in the cryptocurrency space. I have made investments in some of the leading projects in the space, and my portfolio has outperformed the market by a wide margin