{"id":419861,"date":"2025-12-02T16:28:10","date_gmt":"2025-12-02T16:28:10","guid":{"rendered":"https:\/\/www.europesays.com\/us\/419861\/"},"modified":"2025-12-02T16:28:10","modified_gmt":"2025-12-02T16:28:10","slug":"amazon-releases-an-impressive-new-ai-chip-and-teases-a-nvidia-friendly-roadmap","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/us\/419861\/","title":{"rendered":"Amazon releases an impressive new AI chip and teases a Nvidia-friendly roadmap\u00a0\u00a0"},"content":{"rendered":"

Amazon Web Services, which has been building its own AI training chips<\/a> for years now, just introduced a new version known as Trainium3 that comes with some impressive specs. <\/p>\n

The cloud provider, which made the announcement<\/a> Tuesday at its AWS re:Invent 2025, also teased the next product in on its AI training product roadmap: Trainium4, which is already in the works and will be able to work with Nvidia\u2019s chips. <\/p>\n

AWS used its annual tech conference to formally launch Trainium3 UltraServer, a system powered by the company\u2019s state-of-the art, 3 nanometer Trainium3 chip, as well as its homegrown networking tech.\u00a0As you might expect, the third-generation chip and system offer big bumps in performance for AI training and inference over the second-generation, according to AWS. <\/p>\n

AWS says the systems are more than four times faster, with four times more memory, not just for training, but for delivering AI apps at peak demand. Additionally, thousands of UltraServers can be linked together to provide an app with up to 1 million Trainium3 chips \u2014 10 times the previous generation. Each UltraServer can host 144 chips, according to the company.\u00a0<\/p>\n

Perhaps more importantly, AWS says the chips and systems are also 40% more energy efficient than the previous generation.\u00a0\u00a0While the world races to build bigger data centers powered by astronomical gigawatts of electricity,<\/a> data center giant AWS is trying to make systems that drink less, not more.<\/p>\n

It is, obviously, in AWS\u2019s direct interests to do so. But in its classic, Amazon cost-conscience way, it promises that these systems save its AI cloud customers money, too.\u00a0\u00a0<\/p>\n

AWS customers like Anthropic (of which Amazon is also an investor), Japan\u2019s LLM Karakuri, Splashmusic, and Decart have already been using the third-gen chip and system and significantly cut their inference costs, Amazon said.\u00a0<\/p>\n

Techcrunch event<\/p>\n

\n\t\t\t\t\t\t\t\t\tSan Francisco
\n\t\t\t\t\t\t\t\t\t\t\t\t\t|
\n\t\t\t\t\t\t\t\t\t\t\t\t\tOctober 13-15, 2026\n\t\t\t\t\t\t\t<\/p>\n

AWS also presented a bit of a roadmap for the next chip, Trainium4, which is already in development. AWS promised the chip will provide another big step-up in performance and support Nvidia\u2019s NVLink Fusion high-speed chip interconnect technology.\u00a0\u00a0<\/p>\n

This means the AWS Trainium4-powered systems will be able to interoperate and extend their performance with Nvidia GPUs while still using Amazon\u2019s homegrown, lower-cost server rack technology.\u00a0\u00a0<\/p>\n

It\u2019s worth noting, too, that Nvidia\u2019s CUDA (Compute Unified Device Architecture), has become the de facto standard that all the majorAI apps are built to support. The Trainium4-powered systems may make it easier to woo big AI apps built with Nvidia GPUs in mind to Amazon\u2019s cloud.<\/p>\n

Amazon did not announce a timeline for Trainium4. If Amazon follows previous rollout timelines, we\u2019ll likely hear more about Trainium4 at next year\u2019s conference.<\/p>\n","protected":false},"excerpt":{"rendered":"Amazon Web Services, which has been building its own AI training chips for years now, just introduced a…\n","protected":false},"author":3,"featured_media":419862,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[21],"tags":[691,21233,78199,738,5170,196402,158,67,132,68],"class_list":{"0":"post-419861","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-ai-chips","10":"tag-ai-system","11":"tag-artificial-intelligence","12":"tag-aws","13":"tag-aws-reinvent-2025","14":"tag-technology","15":"tag-united-states","16":"tag-unitedstates","17":"tag-us"},"share_on_mastodon":{"url":"https:\/\/pubeurope.com\/@us\/115650920760130193","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/posts\/419861","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/comments?post=419861"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/posts\/419861\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/media\/419862"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/media?parent=419861"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/categories?post=419861"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/us\/wp-json\/wp\/v2\/tags?post=419861"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}