Amazon's "Build on Trainium" will allocate up to $110M in credits for a new research cluster
Amazon is investing $110 million in its Build on Trainium program to provide academic researchers access to powerful AI computing resources and support open-source AI innovation through university partnerships.
Amazon has announced the "Build on Trainium" grant program, which will provide researchers the resources necessary to build new AI architectures, machine learning libraries, and performance optimizations for large-scale distributed AWS Trainium UltraClusters. As part of the program, Amazon has built a 40,000-Trainium chip research cluster to enable large-scale experiments previously out of reach for many institutions.
Partnering universities will receive up to $11 million in Trainium credits, and individual grants of up to $500,000 will be available for the broader AI research community. These grants will be awarded to the selected research proposals from multiple Amazon Research Awards calls for proposals. The research cluster will be accessed through self-managed capacity block reservations using Amazon EC2 Capacity Blocks for ML.
Furthermore, the new Neuron Kernel Interface (NKI) will give researchers direct access to the chips' instructions set, allowing them to build optimized compute kernels for their innovations. Participants in the Build on Trainium program also have access to technical education programs and networking opportunities through Amazon's Neuron Data Science community. The only requirement for participants is that they open-source the advances created with the support of the Build on Trainium program, so the community can benefit from and continue to build on these innovations.
