Amazon spent $110 million to launch the "Build on Trainium" project, aiming to promote artificial intelligence research, reduce reliance on Nvidia, and promote the application of its self-developed Trainium chips. The project will provide university researchers with access to the Trainium chip, supporting them in developing new AI architectures, machine learning libraries, and optimizing large-scale distributed systems. This move is not only a major strategic investment by Amazon in the field of AI, but also marks its further efforts in independent research and development of chip technology, and indicates a possible change in the competitive landscape of the AI chip market in the future.
Recently, Amazon announced a major investment, planning to invest $110 million in artificial intelligence (AI) research, aiming to reduce reliance on Nvidia and promote the development of its own chips. The funding will be used to support the university's generative AI research under the project titled "Build on Trainium."
The project will provide researchers with the opportunity to use Trainium chips so they can develop new AI architectures, machine learning libraries, and performance improvements for large-scale distributed AWS Trainium UltraClusters.
Image source notes: The image is generated by AI, and the image authorized service provider Midjourney
AWS Trainium is a custom machine learning chip designed specifically for deep learning training and inference tasks. Amazon said the project covers a wide range of research directions from algorithm innovation to improvements in AI accelerator performance, and also includes research on large-scale distributed systems. As part of the Build on Trainium program, Amazon has built a research UltraCluster that contains up to 40,000 Trainium chips optimized for AI's unique workloads and computing structure.
Amazon further stated that any AI advances created in the project will be released in open source, allowing researchers and developers to continue to drive their innovations. In addition, Amazon announced a $4 billion investment in Claude developer and OpenAI competitor Anthropic in August.
The Build on Trainium project will also provide funding to support new research and student education, Amazon plans to conduct multiple rounds of research awards, selected proposals will earn AWS training points, and be able to access large Trainium UltraClusters for research. The Catalyst research team at Carnegie Mellon University has been involved in the project.
"AWS's Build on Trainium program gives our teachers and students access to modern accelerators, such as AWS Trainium, on a large scale, with open programming models," said Todd C. Mowry, a computer science professor at the school. This allows us to significantly expand our research on tensor program compilation, machine learning parallelization, and language model services and tuning."
Key points:
Amazon invests $110 million to promote AI research and reduce its reliance on Nvidia.
The “Build on Trainium” project supports university research and provides opportunities for use of Trainium chips.
Research results will be open source to help AI technology continue to develop and innovate.
Amazon's "Build on Trainium" project not only demonstrates its ambitions in the field of AI, but also indicates intensifying competition in the field of AI chips in the future. Open source strategy will further promote the development of AI technology and benefit developers and researchers around the world. This move will undoubtedly have a profound impact on the entire AI industry.