Pinterest signs $4bn AWS deal to expand AI infrastructure for visual discovery
Pinterest said it will increase use of AWS custom silicon, including Trainium for training and running large-scale AI models, and Graviton processors, which already support part of its existing compute infrastructure
17 June, 2026
TT
16
Pinterest has agreed a planned $4bn infrastructure commitment with Amazon Web Services (AWS) through 2031, its largest cloud agreement to date, as it expands artificial intelligence systems powering visual search and recommendations for more than 600 million monthly users.
The agreement extends a long-standing partnership between the two companies that began in 2010 and covers cloud infrastructure, AI training and inference workloads.
Pinterest said it will increase use of AWS custom silicon, including Trainium for training and running large-scale AI models, and Graviton processors, which already support part of its existing compute infrastructure.
The platform uses its proprietary Taste Graph to map user interests and deliver personalised recommendations across areas such as shopping, recipes and home design. It has been shifting its AI stack from traditional retrieval systems toward transformer-based and multimodal generative models.
Pinterest Assistant – A new feature
Pinterest has also introduced Pinterest Assistant, a conversational discovery feature built using vision-language models to support more interactive search and content exploration.
“Pinterest is investing in AI to make discovery more personal, visual and actionable for our users,” said Matt Madrigal, chief technology officer at Pinterest.
The company said the expanded AWS agreement will support continued scaling of AI model training and inference, alongside broader infrastructure modernization efforts, including migration toward a Kubernetes-based architecture using Amazon Elastic Kubernetes Service (EKS).
AWS said the partnership reflects its continued work with large-scale digital platforms building AI-native products.
“AWS is focused on helping customers scale AI systems efficiently across compute and infrastructure,” said Dave Brown, senior vice president of AWS Compute and Machine Learning Services.
The deal underscores the growing reliance of consumer internet platforms on long-term cloud and AI infrastructure commitments as competition intensifies in generative and multimodal AI.
























