Introduction#
Marvell Technology's shares surged following reports of a partnership with Google to create two new chips aimed at improving the efficiency of artificial intelligence (AI) models. This collaboration highlights the growing demand for advanced AI computing solutions.
Details of the Partnership#
According to reports, Google is in discussions with Marvell to develop a memory processing unit that will complement Google’s existing tensor processing unit (TPU). Additionally, they plan to create a new TPU specifically designed for AI inference, which is the process of using a trained AI model to make predictions or decisions.
Timeline and Goals#
The companies are working to finalize the design of the memory chip by next year, with plans to move into test production shortly thereafter. This initiative is part of Google’s strategy to position its TPUs as a competitive alternative to Nvidia’s widely used graphics processing units (GPUs), which dominate the AI computing market.
Google’s Shift in Strategy#
Historically, Google primarily used TPUs for its internal operations. However, in 2022, the company shifted its approach, allowing its cloud division to sell TPUs externally. This change has significantly increased the availability of TPUs to customers, responding to the rising demand for AI computing resources. Recently, Google also launched TorchTPU, a project aimed at making its chips compatible with PyTorch, a popular AI software framework. This move is designed to simplify the transition for developers already using PyTorch, thereby enhancing the adoption of Google’s AI solutions and increasing competition against Nvidia.