Pinterest, the visual inspiration and discovery platform, has expanded its long-standing partnership with AWS as its preferred cloud services provider. The new multi-year agreement represents a US$4 billion investment in cloud infrastructure — the single largest infrastructure commitment the company has ever made.

The deal is designed to accelerate Pinterest's AI and machine learning roadmap, deliver a faster and more responsive search and shopping experience for users, and further modernise the platform's underlying infrastructure. Pinterest has partnered with AWS since 2010, making this expansion a natural evolution of a relationship spanning over 16 years.

Pinterest has been rapidly upgrading from older legacy systems to generative AI, continually refining its own visual AI capabilities while also adopting best-in-class open-source models. At the heart of this transformation is the platform's Taste Graph — a proprietary AI engine that takes users from broad inspiration to highly personalised, actionable ideas.

The latest flagship product built on this foundation is the Pinterest Assistant — a conversational AI tool that transforms visual search into a seamless, multi-step dialogue experience. It is powered by scalable open-source vision-language models (VLMs) running on AWS infrastructure, enabling users to explore ideas through natural conversation.

Pinterest plans to expand its use of AWS Trainium chips to host and run large language models and vision-language models for personalised visual search and AI-assisted discovery. Meanwhile, AWS Graviton already powers roughly a third of Pinterest's entire compute workload, with further expansion planned.

The expanded AWS partnership signals a clear strategic direction for Pinterest: become the definitive AI-powered visual discovery platform. As Matt Madrigal, CTO of Pinterest, put it, the company is investing heavily in AI to make discovery more personal, visual, and actionable for the hundreds of millions of people who visit the platform each month.

The infrastructure investment is expected to yield tangible improvements in both consumer experience — via faster, smarter search and personalisation — and advertiser performance, as Pinterest's proprietary and open-source AI models continue to advance. With Trainium and Graviton providing scalable compute for training and inference at massive scale, Pinterest is well-equipped to move faster and deliver new AI-driven experiences to its global user base.

Key Takeaways

• →Pinterest has committed US$4 billion to AWS through 2031 — the largest infrastructure deal in the company's history — to accelerate its AI and machine learning roadmap.
• →The partnership builds on a 16-year relationship with AWS, with the expanded agreement deepening Pinterest's reliance on purpose-built AWS silicon: Trainium and Graviton.
• →AWS Graviton already powers roughly one-third of Pinterest's compute; Trainium will now be added to run large language models and vision-language models at scale.
• →Pinterest's new AI flagship, the Pinterest Assistant, uses open-source vision-language models on AWS to deliver conversational visual search to over 600 million monthly users.
• →The investment aims to improve both consumer experience — through faster, personalised discovery — and advertiser performance, as Pinterest advances its proprietary and open-source AI models.
• →The deal signals Pinterest's clear strategic intent: to become the world's leading AI-powered visual discovery platform, driven by its proprietary Taste Graph and next-generation generative AI capabilities.