You are currently viewing How Amazon is Using Customized AWS Processors to Catch up to Microsoft and Google in Generative AI

How Amazon is Using Customized AWS Processors to Catch up to Microsoft and Google in Generative AI

In an undisclosed office space in Austin, Texas, Amazon employees are crafting custom microchips, Inferentia and Trainium, designed to accelerate generative AI and provide an alternative to training large language models on Nvidia GPUs. These chips offer an option for AWS customers facing challenges in procuring Nvidia GPUs due to their scarcity and cost. Amazon Web Services (AWS) CEO, Adam Selipsky, expressed confidence in Amazon’s capability to meet the growing demand for AI chips.

However, Amazon’s competitors have already made significant strides in capturing the generative AI market. Microsoft garnered attention by investing $13 billion in OpenAI and quickly integrating generative AI models into its products. Google introduced its own language model, Bard, and invested $300 million in OpenAI’s rival, Anthropic. Amazon, comparatively, announced its Titan family of language models later, indicating a shift from its usual market-creating approach to playing catch-up in this emerging domain.

Despite this, Amazon’s strengths lie in its custom silicon capabilities, such as the Trainium and Inferentia chips, which are focused on training and inference, respectively. These chips are unique technical assets that could potentially give Amazon an edge in generative AI. Amazon’s dominance in cloud computing, represented by AWS’ 40% market share in 2022, can also aid its foray into generative AI. AWS’ extensive customer base and established cloud infrastructure provide a strong foundation for deploying generative AI applications.

Amazon’s strategy revolves around its AWS ecosystem, empowering existing customers to expand into value-creating generative AI applications. This approach leverages AWS’ cloud dominance and growing portfolio of developer tools for generative AI, which includes AWS HealthScribe and CodeWhisperer. Despite its relatively delayed entry, Amazon is actively working to bolster its presence in the generative AI landscape, and its established cloud infrastructure and customer base could potentially propel its success in this arena.