Companies like Google, Microsoft, and Amazon are all joining the battle with specialized AI chips, escalating the competition among cloud providers for AI technology. By allocating $110 million to AI researchers through a new grant program dubbed Build on Trainium, Amazon Web Services (AWS) is pushing its AI processor, Trainium. Under this scheme, university partners can receive up to $11 million in credits, while individual AI research projects can receive lesser grants of up to $500,000.
Additionally, the program offers a research cluster with 40,000 Trainium chips that scientists and students can reserve access to. The goal of Build on Trainium, according to Gadi Hutt of AWS’s Annapurna Labs, is to solve the resource constraints that restrict scholarly AI research. With the requirement that their work be published and made publicly available on GitHub, grantees will have access to training materials and support services.
There are worries, though. Since a recent study revealed that corporate-backed AI research frequently lacks focus on ethical considerations, some academics contend that corporate funding may favour projects that will benefit the company more than those that are dedicated to ethics. According to AWS, proposals will be assessed by a committee of industry professionals, and selection will be based on the caliber and impact of research.
Funding and resource constraints are major issues for academic AI researchers, who frequently lack the vast infrastructure that commercial giants like Meta, which purportedly contains over 100,000 AI processors, have. By comparison, Stanford’s Natural Language Processing Group uses only 68 GPUs for its research.
Even if there are governmental financing programs like the U.S. National AI Research Institutes, they are still small in comparison to corporate donations. Because of the persistent resource imbalance between academia and industry, a growing number of researchers and Ph.D.’s are entering the commercial tech sector, where they may obtain the processing power required to advance AI projects.