Pinterest announced on Thursday a major expansion of its long-term partnership with Amazon Web Services (AWS), committing up to $4 billion for cloud services through 2031.
The agreement, the largest infrastructure commitment in Pinterest’s history, will support the company’s AI initiatives to enhance visual search, personalization, and discovery for more than 600 million monthly active users worldwide, reported AmazonNews.
Pinterest and AWS have collaborated since 2010, jointly optimizing one of the largest-scale data lakes on AWS. The new deal deepens this relationship, focusing on AI model training, inference, and platform infrastructure modernization.
“Pinterest is heavily investing in AI to make discovery more personal, visual and actionable for the hundreds of millions of people who use our platform every month,” said Matt Madrigal, Chief Technology Officer, Pinterest, in a June 4 press release. “This expanded commitment with AWS gives us the compute flexibility, hardware optionality, and infrastructure efficiency to accelerate our AI vision for the next generation of visual discovery on Pinterest.”
Pinterest plans to expand its use of AWS custom silicon, including Trainium for large language and vision-language models, and Graviton processors, which already power roughly a third of the company’s compute infrastructure. These tools will support advanced recommendation systems, multimodal models, and features like Pinterest Assistant, a multi-turn conversational discovery tool.
The company will also transition more of its infrastructure to a Kubernetes-based architecture on Amazon Elastic Kubernetes Service (EKS) to improve developer velocity and operational reliability.
“AWS is the best place to do AI at this scale, and we’re committed to helping Pinterest’s teams move faster and think bigger—benefiting users all over the world,” said Dave Brown, Senior Vice President of AWS Compute & ML Services in the release. “AWS compute and purpose-built silicon like Trainium and Graviton give Pinterest the price-performance to train and run AI models at massive scale across both training and inference.”
The visual discovery platform, known for helping users find inspiration and shop products, has evolved its AI capabilities from traditional embedding-based retrieval to transformer-based generative models. The Taste Graph powers personalized, actionable results ranging from recipes to home renovations.
