Amazon has announced its ambitious plan to build one of the world’s most powerful artificial intelligence supercomputers in partnership with Anthropic, an AI company rivaling OpenAI. This initiative, revealed at Amazon’s Re:Invent conference in Las Vegas, aims to construct a supercomputer five times larger than the setup currently behind Anthropic’s most advanced AI model. This new AI infrastructure will use hundreds of thousands of Amazon’s newest AI training chip, Trainium 2, and upon completion, it is expected to be the largest AI machine of its kind.
According to Matt Garman, CEO of Amazon Web Services, the project, known as Rainer, signifies Amazon’s growing influence in the field of generative AI. Alongside this, Garman introduced the availability of Trainium 2 through new AWS clusters that offer a cost-effective alternative to those built on Nvidia’s GPUs. He claimed these clusters would be 30-40% cheaper, making Amazon’s cloud services increasingly attractive for companies seeking AI solutions.
Historically seen as trailing behind major players like Microsoft and Google in AI, Amazon has now emerged as a formidable contender with significant investments, including an $8 billion injection into Anthropic. Through its AWS platform, Amazon provides a range of innovative tools and services designed to facilitate the development and deployment of AI applications for enterprises.
Among the key announcements at the conference was Amazon’s next-generation chip, Trainium 3, boasting four times the performance of its predecessor and slated for release in late 2025. Industry analysts like Patrick Moorhead from Moore Insight & Strategy praise the advancements in Amazon’s chip technology, highlighting the improvements in interconnects that enhance data transfer between chips—a crucial factor for developing extensive AI systems.
While Nvidia currently dominates AI training hardware, Amazon’s advancements suggest increasing competition in the coming years. Steven Dickens of HyperFRAME Research points out that Amazon’s extensive cloud service offerings provide a distinct advantage in selling generative AI capabilities, despite not having a direct ChatGPT-like product.
Amazon’s efforts to democratize AI include a new service called Model Distillation, which allows companies to create smaller, more efficient AI models from larger ones, reducing cost and resource demands without sacrificing functionality. Furthermore, Bedrock Agents was launched, a tool facilitating the creation and management of autonomous AI agents for tasks like customer support and data analysis.
In addressing concerns over AI reliability, Amazon introduced Automated Reasoning, a verification tool that ensures the accuracy of chatbot responses. This system translates natural language into a logical format, providing a rigorous validation process of AI outputs. Bryon Cook, head of Amazon’s Autonomous Reasoning Group, emphasizes that this approach has been effective in fields like chip design and cryptography and could extend to more complex applications.
Overall, Amazon’s substantial investments and innovations underline its commitment to being a key player in the AI landscape. As AWS customers transition from trial to commercial applications of AI, Amazon’s strategic use of custom silicon and cutting-edge AI tools positions it well to meet future demands in the industry.
You can read the original article here: [https://www.wired.com/story/amazon-reinvent-anthropic-supercomputer/](https://www.wired.com/story/amazon-reinvent-anthropic-supercomputer/)
