Pinterest has expanded its partnership with Amazon Web Services (AWS) and is increasingly relying on its AI infrastructure. The social media platform has confirmed a $4 billion commitment. This means that AWS Trainium and Graviton will be used for training and operating AI models on a large scale by 2031.
The agreement, considered the largest infrastructure deal in the company’s history, calls for Pinterest to use Amazon’s special silicon chips. These are used to host and operate speech and image-to-speech models that enable personalized visual search and AI-powered discovery. Pinterest will also expand its use of Graviton, the AWS processors for cloud computing workloads, to power discovery systems.
This latest agreement expands the existing relationship between Pinterest and AWS, which dates back to 2010. Since the beginning of the partnership, the two companies have worked to improve Pinterest’s reliability, efficiency and performance. Together they have built what they say is one of the “largest data lakes” on AWS, a repository that stores raw data.
Pinterest has also expanded its use of AI to improve discovery capabilities. The platform has introduced features such as the ‘Taste Graph’ and the ‘Pinterest Assistant’. These offer users advanced recommendation systems and conversational discovery options to improve and personalize the online experience.
According to Pinterest Chief Technology Officer (CTO) Matt Madrigal, the renewed commitment supports the next phase of growth in implementing AI models. AWS will provide further flexibility, hardware options and more efficient infrastructure to accelerate Pinterest’s AI vision.
“This strategic partnership will help accelerate AI innovation at Pinterest, improving both the experience for our consumers and performance for advertisers by advancing our own models and the use of open source models,” Madrigal added.
This article was created using digital tools translated.
FashionUnited uses artificial intelligence to speed up the translation of articles and improve the end result. They help us to make FashionUnited’s international reporting quickly and comprehensively accessible to a German-speaking readership. Articles translated using AI-based tools are proofread and carefully edited by our editors before they are published. If you have any questions or comments, please email [email protected]